Power distribution network frequency adjusting method and system based on frequency characteristics of thermal power generating unit

Abstract

The invention discloses a power distribution network frequency regulation method and system based on thermal power generating unit frequency characteristics, and belongs to the technical field of power distribution network frequency regulation, and the method comprises the steps: S1, obtaining the historical output frequency of a thermal power generating unit, and judging whether the thermal power generating unit participates in the power distribution network frequency regulation or not according to the historical output frequency; s2, when the thermal power generating unit participates in power distribution network frequency adjustment, calculating an initial state quantity set based on the historical output frequency; s3, inputting the initial state quantity set into a target-strategy network for training to obtain an optimal target-strategy network; s4, acquiring the real-time output frequency of the thermal power generating unit, and inputting the real-time output frequency of the thermal power generating unit into the optimal target-strategy network to predict the optimal control parameter combination of the thermal power generating unit; s5, adjusting parameters of a controller corresponding to the thermal power generating unit in the power distribution network according to the optimal control parameter combination of the thermal power generating unit; the problem of poor stability of the power distribution network caused by fluctuation due to the fact that the new energy power generation device and the elastic resource change of the power distribution network are not considered in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network frequency regulation, and in particular to a distribution network frequency regulation method and system based on the frequency characteristics of thermal power units. Background Art

[0002] With the development of new energy technologies, more and more new energy power generation devices and their corresponding power electronic equipment are connected to the distribution network. Since the frequency regulation ability of new energy power generation devices is relatively weak, when they fluctuate, it will cause large fluctuations in the frequency of the distribution network. In addition, the number of traditional power generation devices has been reduced due to the replacement of new energy power generation devices, which also makes it difficult for the frequency regulation performance of traditional power generation devices to complement the frequency regulation performance of new energy power generation devices, making it difficult for the distribution network frequency to recover after fluctuations and remain in a low-frequency state for a long time. The existing technology generally uses conventional controllers to control the elastic resources in the distribution network to adjust the frequency of the distribution network, so that the frequency of the distribution network can recover quickly after fluctuations. However, the elastic resources of the distribution network are highly volatile. For example, the connection and disconnection of electric vehicles will cause rapid changes in the operating state of the distribution network. At this time, the frequent actions of the conventional controller will cause jitter problems, resulting in poor stability in the operation of the distribution network.

[0003] Chinese patent, publication number: CN117394407B, publication date: June 28, 2024, discloses an auto-disturbance rejection control method applied to a photovoltaic hybrid energy storage system, including: step 1, establishing a mathematical model of the energy storage system; step 1.1, establishing a mathematical model of the PCS; step 1.2, establishing a mathematical model of a bidirectional DC / DC converter; step 2, designing an auto-disturbance rejection control method; step 2.1, improving the design of an extended state observer; step 2.2, designing a state error feedback law based on a backstepping complementary sliding mode; step 3, designing an energy storage system control method: step 3.1, improving the auto-disturbance rejection design of the PCS voltage outer loop; step 3.2, improving the auto-disturbance rejection design of the bidirectional DC / DC converter current inner loop; however, the invention does not take into account the fluctuations in the distribution network caused by changes in new energy power generation devices and changes in the elastic resources of the distribution network, resulting in poor stability of the distribution network. Summary of the invention

[0004] The purpose of the present invention is to address the problem that the existing technology does not take into account the fluctuations in the distribution network caused by changes in the elastic resources of the distribution network, resulting in poor stability of the distribution network; a distribution network frequency regulation method and system based on the frequency characteristics of thermal power units are proposed, the thermal power units participating in the frequency regulation of the distribution network are determined by the historical output frequencies of the thermal power units, and the input quantity of the target-strategy network is calculated according to the historical output frequencies of the thermal power units participating in the frequency regulation of the distribution network to train the target-strategy network, and then the optimal control parameter combination of the corresponding thermal power units is predicted according to the real-time output frequencies of the thermal power units participating in the frequency regulation of the distribution network and the target-strategy network, finally, the parameters of the corresponding controllers of the thermal power units in the distribution network are adjusted according to the optimal control parameter combination, and the real-time frequency of the distribution network is adjusted by using the frequency characteristics of the thermal power units, thereby significantly improving the stability of the distribution network.

[0005] In a first aspect, a technical solution provided in an embodiment of the present invention is a method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit, comprising the following steps: S1. Obtain the historical output frequency of the thermal power unit, and determine whether the corresponding thermal power unit participates in the frequency regulation of the distribution network according to the historical output frequency; S2. When the thermal power unit participates in the frequency regulation of the distribution network, the initial state quantity set of the thermal power unit is calculated based on the historical output frequency of the thermal power unit; S3, inputting the initial state quantity set of the thermal power unit into the target-strategy network for training to obtain the optimal target-strategy network; S4, obtaining the real-time output frequency of the thermal power units participating in the frequency regulation of the distribution network, and inputting the real-time output frequency of the thermal power units into the optimal target-strategy network for prediction to obtain the optimal control parameter combination of the thermal power units; S5. Adjust the parameters of the controllers corresponding to the thermal power units in the distribution network according to the optimal control parameter combination of the thermal power units.

[0006] In this scheme, the thermal power units participating in the frequency regulation of the distribution network are determined by the historical output frequencies of the thermal power units, and the initial state quantities of the thermal power units are calculated according to the historical output frequencies of the thermal power units participating in the frequency regulation of the distribution network as the input quantities of the target-strategy network to train the target-strategy network, and the optimal target-strategy network that matches the operating state of the thermal power units is obtained. Subsequently, the real-time output frequency of the thermal power units participating in the frequency regulation of the distribution network is input into the optimal target-strategy network for prediction, and the optimal control parameter combination of the thermal power units under the current operating state is obtained. Finally, the parameters of the corresponding controllers of the thermal power units in the distribution network are adjusted according to the optimal control parameter combination to realize the control of the output frequency of the corresponding thermal power units, and then the output frequency of the thermal power units is used to adjust the frequency of the corresponding distribution network, thereby significantly improving the stability of the distribution network.

[0007] Preferably, in S1, the specific process of judging whether the corresponding thermal power unit participates in the frequency regulation of the distribution network according to the historical output frequency is as follows: The output frequency deviation is calculated based on the historical output frequency and the rated frequency of the corresponding thermal power unit, and the output frequency deviation is compared with the preset frequency regulation threshold. If the historical output frequency deviation is greater than or equal to the absolute value of the frequency regulation threshold, it is determined that the thermal power unit does not participate in the frequency regulation of the distribution network. If the historical output frequency deviation is less than the absolute value of the frequency regulation threshold, it is determined that the thermal power unit participates in the frequency regulation of the distribution network.

[0008] In this scheme, the thermal power units participating in the frequency regulation of the distribution network are determined based on the output frequency deviation of the thermal power units and the preset frequency regulation threshold, and the thermal power units whose output frequency is insufficient compared with the real-time frequency of the distribution network can be screened out.

[0009] Preferably, in S2, the specific process of calculating the initial state quantity set of the thermal power unit based on the historical output frequency of the thermal power unit is: S21, subtracting the historical output frequency of the thermal power unit from the rated frequency of the thermal power unit to obtain an output frequency deviation; synchronously, calculating the integral value of the historical output frequency according to an indefinite integral formula and the historical output frequency of the thermal power unit; S22, integrating the integral value of the historical output frequency and the output frequency deviation to obtain an initial state quantity set of the thermal power unit.

[0010] In this scheme, the initial state quantity set of the thermal power group is calculated according to the output frequency of the thermal power group, which can not only obtain the frequency periodic characteristics of the thermal power group according to the corresponding time scale of the thermal power group, but also convert the output frequency of the thermal power group into a form suitable for the subsequent target-strategy network, so as to support the subsequent training target-strategy network.

[0011] Preferably, in S3, the specific process of inputting the initial state quantity set of the thermal power unit into the target-strategy network for training to obtain the optimal target-strategy network is: S31, inputting the initial state quantity set of the thermal power unit into the target-strategy network for prediction and calculation to obtain the current action quantity, the current reward quantity and the next state quantity; S32, sorting out the initial state quantity, the current action quantity, the current reward quantity and the next state quantity to obtain a state set, and judging whether to end the state update of the thermal power unit according to the current time step of the target-strategy network and the preset time step threshold; S33. After the state update of the thermal power unit is completed, determine whether the target-strategy network training is completed based on the current training times of the target-strategy network and the preset training cycle. If so, mark the target-strategy network corresponding to the current training times as the optimal target-strategy network. If not, change the target-strategy network based on the random initialization formula, update the initial state quantity of the thermal power unit based on the state set and execute S31.

[0012] In this scheme, the current action amount, the current reward amount and the next state amount are calculated according to the initial state amount of the thermal power unit, and the state of the thermal power unit corresponding to the next time step can be predicted. Secondly, the predicted state amount of the thermal power unit can be continuously and automatically updated within the preset time step threshold to clarify the changing trend of the state of the thermal power unit in the current time period. Then, the target-strategy network is adjusted according to the training cycle to avoid underfitting and overfitting of the target-strategy network.

[0013] Preferably, in said S31, there is a preparatory step before inputting the initial state quantity set of the thermal power unit into the target-strategy network, and the preparatory step is specifically: The initial state quantity set of the thermal power unit is divided into at least one small batch state set according to the preset training times and training amount, and the small batch state set is input into the target-strategy network.

[0014] In this solution, the initial state quantity set of the thermal power unit is converted into a small batch state set, which can accelerate the convergence speed of the corresponding target-strategy network, make the training speed of the target-strategy network faster, and ensure that the performance of the target-strategy network after convergence is more stable.

[0015] Preferably, in S33, the specific process of changing the target-strategy network based on the random initialization formula and updating the initial state quantity of the thermal power unit based on the state set is: S331, calculating a selection index according to the current training times of the target-strategy network, a preset update frequency and an update selection formula, and selecting an update target based on the selection index; S332: when the update target is a target network in a target-strategy network, changing parameters of the target network based on a random initialization formula; When the update target is a target network and a policy network in a target-policy network, changing the parameters of the target network based on a random initialization formula, and changing the parameters of the policy network based on a random initialization formula; Synchronously, at least one state quantity is sampled from the state set as the initial state quantity of the thermal power unit.

[0016] In this scheme, different sub-networks in the target-strategy network are selected for updating according to the update selection formula. The parameters of the target-strategy network can be adjusted accordingly according to the real-time status of the thermal power unit, effectively improving the accuracy of the control parameter combination of the thermal power unit predicted by the target-strategy network, and stabilizing the training process of the target-strategy network to ensure the effectiveness of the training process.

[0017] Preferably, in S331, the update selection formula is specifically: k%×C=1; Where k is the current training number of the target-policy network, and C is the update frequency.

[0018] On the other hand, a technical solution also provided in an embodiment of the present invention is a distribution network frequency regulation system, comprising: a controller, a speed regulator, and a thermal power unit; The control end of the controller is electrically connected to the controlled end of the speed regulator, the data collection end of the controller is electrically connected to the data collection end of the thermal power unit, the control end of the speed regulator is electrically connected to the controlled end of the thermal power unit, and the output end of the thermal power unit is electrically connected to the external distribution network.

[0019] In this solution, the controller can collect output data of the thermal power unit to realize real-time monitoring of the thermal power unit, and the controller can determine whether the corresponding thermal power unit has the ability to participate in the frequency regulation of the distribution network based on the frequency data in the output data of the thermal power unit.

[0020] Preferably, the controller includes: a state observer, a proportional-differential controller; The input end of the state observer is electrically connected to the data acquisition end of the thermal power unit as the data collection end of the controller, the output end of the state observer is electrically connected to the input end of the proportional-differential controller, and the control end of the proportional-differential controller is electrically connected to the controlled end of the speed regulator as the control end of the controller.

[0021] In this solution, the proportional-differential controller can also accept a manually or automatically set control parameter combination corresponding to the thermal power unit to adjust the output of the controller to control the speed governor, thereby achieving control of the output of the thermal power unit.

[0022] Preferably, the thermal power unit comprises: a reheat boiler, a steam turbine and a generator set; The output end of the reheat boiler is connected to the input end of the steam turbine, the monitoring end of the reheat boiler is electrically connected to the output end of the steam turbine, the output end of the steam turbine is connected to the output end of the generator set, and the output end of the generator set is electrically connected to the external distribution network.

[0023] The present invention has at least the following substantial effects: The present application determines the thermal power units participating in the frequency regulation of the distribution network through the historical output frequencies of the thermal power units, and calculates the initial state quantity of the thermal power units according to the historical output frequencies of the thermal power units participating in the frequency regulation of the distribution network as the input quantity of the target-strategy network to train the target-strategy network, and obtains the optimal target-strategy network that matches the operating state of the thermal power units. Subsequently, the real-time output frequency of the thermal power units participating in the frequency regulation of the distribution network is input into the optimal target-strategy network for prediction, and the optimal control parameter combination of the thermal power units under the current operating state is obtained. Finally, the parameters of the corresponding controllers of the thermal power units in the distribution network are adjusted according to the optimal control parameter combination to realize the control of the output frequency of the corresponding thermal power units, and then the output frequency of the thermal power units is used to adjust the frequency of the corresponding distribution network, thereby significantly improving the stability of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts.

[0025] Figure 1 This is a flow chart of a method for frequency regulation of a distribution network based on the frequency characteristics of a thermal power unit according to the present invention; Figure 2 The figure is a schematic diagram of the structure of a distribution network frequency regulation system. DETAILED DESCRIPTION

[0026] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0027] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the drawings; the process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0028] Embodiment 1: like Figure 1As shown, this embodiment provides a distribution network frequency regulation method based on the frequency characteristics of thermal power units, including the following steps: S1, obtaining the historical output frequency of the thermal power units, and judging whether the corresponding thermal power units participate in the distribution network frequency regulation according to the historical output frequency.

[0029] In one embodiment, the specific process of judging whether the corresponding thermal power unit participates in the frequency regulation of the distribution network according to the historical output frequency is as follows: The output frequency deviation is calculated based on the historical output frequency and the rated frequency of the corresponding thermal power unit, and the output frequency deviation is compared with the preset frequency regulation threshold. If the historical output frequency deviation is greater than or equal to the absolute value of the frequency regulation threshold, it is determined that the thermal power unit does not participate in the frequency regulation of the distribution network. If the historical output frequency deviation is less than the absolute value of the frequency regulation threshold, it is determined that the thermal power unit participates in the frequency regulation of the distribution network.

[0030] In this embodiment, the frequency modulation threshold is specifically set to 0.1, so the comparison between the output frequency deviation Δf obtained by calculating the difference between the historical output frequency and the rated frequency of the corresponding thermal power unit and the frequency modulation threshold can be represented by a mathematical model. The mathematical model for comparing the output frequency deviation Δf with the frequency modulation threshold is as follows: Δf≥0.1Δf≤0.1; When the thermal power unit is determined not to participate in the frequency regulation of the distribution network, the corresponding thermal power unit can be periodically detected according to the preset thermal power unit output frequency acquisition cycle to achieve real-time monitoring of the thermal power unit that does not meet the distribution network frequency regulation.

[0031] S2. When the thermal power unit participates in the frequency regulation of the distribution network, the initial state quantity set of the thermal power unit is calculated based on the historical output frequency of the thermal power unit; S21, subtracting the historical output frequency of the thermal power unit from the rated frequency of the thermal power unit to obtain an output frequency deviation; synchronously, calculating the integral value of the historical output frequency according to an indefinite integral formula and the historical output frequency of the thermal power unit; S22, integrating the integral value of the historical output frequency and the output frequency deviation to obtain an initial state quantity set of the thermal power unit.

[0032] In this embodiment, the integral value of the historical output frequency can be expressed as ∫fdt to describe the change trend of the output frequency of the thermal power unit within the detection time period.

[0033] S3, inputting the initial state quantity set of the thermal power unit into the target-strategy network for training to obtain the optimal target-strategy network; S31, input the initial state quantity set of the thermal power unit into the target-strategy network for prediction and calculation to obtain the current action quantity, the current reward quantity and the next state quantity.

[0034] In one embodiment, there are further preparatory steps before inputting the initial state quantity set of the thermal power unit into the target-strategy network. The preparatory steps are specifically as follows: The initial state quantity set of the thermal power unit is divided into at least one small batch state set according to the preset training times and training amount, and the small batch state set is input into the target-strategy network.

[0035] In this embodiment, in order to speed up the training while ensuring the training effect, the maximum number of training times is set to 800 and the training volume is set to 400.

[0036] S32, sorting out the initial state quantity, the current action quantity, the current reward quantity and the next state quantity to obtain a state set, and judging whether to end the state update of the thermal power unit according to the current time step of the target-strategy network and the preset time step threshold.

[0037] In this embodiment, if the current time step is greater than the preset time step threshold, it is determined that the status update of the thermal power unit is completed, specifically referring to that the working process of the thermal power unit in the time period represented by the time step threshold has ended; if the current time step is less than or equal to the preset time step threshold, it is determined that the status update of the thermal power unit is not completed, specifically referring to that the working process of the thermal power unit in the time period represented by the time step threshold has not ended.

[0038] S33, when the state update of the thermal power unit is completed, judging whether the target-strategy network training is completed according to the current training times of the target-strategy network and the preset training cycle, if so, marking the target-strategy network corresponding to the current training times as the optimal target-strategy network, if not, changing the target-strategy network based on the random initialization formula, updating the initial state quantity of the thermal power unit based on the state set and executing S31; S331, calculating a selection index according to the current training times of the target-strategy network, a preset update frequency and an update selection formula, and selecting an update target based on the selection index; The update selection formula is specifically: k%×C=1; Where k is the current training number of the target-policy network, and C is the update frequency.

[0039] S332: when the update target is a target network in a target-strategy network, changing parameters of the target network based on a random initialization formula; When the update target is a target network and a policy network in a target-policy network, changing the parameters of the target network based on a random initialization formula, and changing the parameters of the policy network based on a random initialization formula; Synchronously, at least one state quantity is sampled from the state set as the initial state quantity of the thermal power unit.

[0040] In this embodiment, if the current number of training times is greater than the training cycle, it is determined that the target-strategy network training is completed; if the current number of training times is less than or equal to the training cycle, it is determined that the target-strategy network training is not completed, and on the basis of the incomplete training, the target network and the strategy network in the target-strategy network are selected for parameter update according to the current number of training times, the preset update frequency and the update selection formula, the maximum value of the training cycle is the above-mentioned maximum number of training times, the random initialization formula includes a state initialization formula and a time step initialization formula; the state initialization formula is as follows: s t =s0; In the formula, s t represents the state quantity sampled from the state set, s0 represents the initial state quantity of the thermal power unit after initialization; the time step initialization formula is specifically: t = 1; Where t represents the last time step when the state update of the thermal power unit ends.

[0041] S4. Obtain the real-time output frequency of the thermal power units participating in the frequency regulation of the distribution network, and input the real-time output frequency of the thermal power units into the optimal target-strategy network for prediction to obtain the optimal control parameter combination of the thermal power units.

[0042] S5. Adjust the parameters of the controllers corresponding to the thermal power units in the distribution network according to the optimal control parameter combination of the thermal power units.

[0043] On the other hand, a technical solution also provided in an embodiment of the present invention is a distribution network frequency regulation system, comprising: a controller, a speed regulator, and a thermal power unit; The control end of the controller is electrically connected to the controlled end of the speed regulator, the data collection end of the controller is electrically connected to the data collection end of the thermal power unit, the control end of the speed regulator is electrically connected to the controlled end of the thermal power unit, and the output end of the thermal power unit is electrically connected to the external distribution network.

[0044] Furthermore, the controller includes: a state observer, a proportional-differential controller; The input end of the state observer is electrically connected to the data acquisition end of the thermal power unit as the data collection end of the controller, the output end of the state observer is electrically connected to the input end of the proportional-differential controller, and the control end of the proportional-differential controller is electrically connected to the controlled end of the speed regulator as the control end of the controller.

[0045] Furthermore, the thermal power unit comprises: a reheat boiler, a steam turbine and a generator set; The output end of the reheat boiler is connected to the input end of the steam turbine, the monitoring end of the reheat boiler is electrically connected to the output end of the steam turbine, the output end of the steam turbine is connected to the output end of the generator set, and the output end of the generator set is electrically connected to the external distribution network.

[0046] like Figure 2 As shown, in order to better explain the distribution network frequency regulation method and system based on the frequency characteristics of the thermal power unit described in this embodiment, this embodiment builds a corresponding simulation model in the simulation platform Simulink for experimental verification, including a steam turbine, a reheat boiler, a speed governor, a generator set, and a controller LADRC; The corresponding mathematical model of the steam turbine is: In the formula, G t (s) is the turbine transfer function, K t is the turbine gain, T t is the turbine time constant, s is a complex variable, in this embodiment, the turbine gain is set to 1, and the turbine time constant is set to 10; The mathematical model corresponding to the reheat boiler is: In the formula, G r (s) is the reheat boiler transfer function, K r is the reheat coefficient, T r is the reheat time constant, s is a complex variable. In this embodiment, the reheat coefficient is set to 2.5 and the reheat time constant is set to 5; The corresponding mathematical model of the speed regulator is: In the formula, G g (s) is the speed regulator transfer function, K g is the speed regulator gain, T g is the speed regulator time constant, s is a complex variable, in this embodiment, the speed regulator gain is set to 1, and the speed regulator time constant is set to 0.2; The corresponding mathematical model of the generator set is: In the formula, G p (s) is the transfer function of the generator set, K p is the generator set gain, T p is the generator set time constant, s is a complex variable, in this embodiment, the generator set gain is set to 1, and the generator set time constant is set to 10; in addition Figure 2R2 in the figure is the feedback coefficient of the corresponding feedback device k. In this embodiment, the feedback coefficient is set to 2.4. w0 is the first weight parameter in the control parameter combination of the thermal power unit. c It is the second weight parameter in the control parameter combination of the thermal power unit, and b0 is the bias parameter in the control parameter combination of the thermal power unit.

[0047] After the distribution network frequency regulation system is built in the simulation platform Simulink, the distribution network frequency regulation method based on the frequency characteristics of the thermal power unit is arranged in the RLagent module of the simulation platform Simulink, and the critic network and the actor network in the deep learning network are used to jointly construct a target-strategy network, wherein the state path neural network layer of the critic network is 2, the action path neural network layer is 2, the number of neurons is 100, the learning rate is set to 0.0005, and the gradient threshold is 1; the actor network only sets one path, the neural network layer is 2, the number of neurons is 100, the learning rate is set to 0.0005, and the gradient threshold is 1; After the target-strategy network is arranged, the parameters of the distribution network frequency regulation method based on the frequency characteristics of the thermal power unit are set, wherein the target smoothing factor is 1e-3, the empirical buffer length is 1e-6, the discount factor is 0.8, the random disturbance factor is 0.2, the maximum number of training times is 800, and the training amount is 400.

[0048] Next, the simulation results of the constructed simulation model are compared with the operating data of the corresponding regulation system of the conventional controller. The comparison results are shown in Table 1: Table 1. Comparison results of simulation model simulation results and corresponding regulation system data of conventional controller Comparison indicators Conventional controller corresponding regulation system Simulation Model Performance Improvements <![CDATA[Maximum frequency deviation (×10 -3 )]]> 1.408 1.247 11.43％ Adjustment time(s) 36.585 31.529 13.82％ According to the data in Table 1, it can be inferred that the distribution network frequency regulation method and system based on the frequency characteristics of thermal power units provided in this embodiment can restore the frequency of the corresponding distribution network to stability in the shortest time, minimizing the maximum frequency deviation of the thermal power units.

[0049] This embodiment has at least the following substantial effects: In this embodiment, the thermal power units participating in the frequency regulation of the distribution network are determined by the historical output frequencies of the thermal power units, and the initial state quantities of the thermal power units are calculated according to the historical output frequencies of the thermal power units participating in the frequency regulation of the distribution network as the input quantities of the target-strategy network to train the target-strategy network, so as to obtain the optimal target-strategy network that matches the operating state of the thermal power units. Subsequently, the real-time output frequency of the thermal power units participating in the frequency regulation of the distribution network is input into the optimal target-strategy network for prediction, so as to obtain the optimal control parameter combination of the thermal power units under the current operating state. Finally, the parameters of the corresponding controllers of the thermal power units in the distribution network are adjusted according to the optimal control parameter combination to realize the control of the output frequency of the corresponding thermal power units, and then the output frequency of the thermal power units is used to adjust the frequency of the corresponding distribution network, thereby significantly improving the stability of the distribution network.

[0050] The above specific embodiments are preferred embodiments of the present invention, and are not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to the present specific embodiments. All equivalent changes made in accordance with the shape, structure, and method of the present invention are within the protection scope of the present invention.

Claims

1. A distribution network frequency regulation method based on the frequency characteristics of thermal power units, characterized in that: The method comprises the following steps: S1. Obtain the historical output frequency of the thermal power unit, and determine whether the corresponding thermal power unit participates in the frequency regulation of the distribution network according to the historical output frequency; S2. When the thermal power unit participates in the frequency regulation of the distribution network, the initial state quantity set of the thermal power unit is calculated based on the historical output frequency of the thermal power unit; S3, inputting the initial state quantity set of the thermal power unit into the target-strategy network for training to obtain the optimal target-strategy network; S4, obtaining the real-time output frequency of the thermal power units participating in the frequency regulation of the distribution network, and inputting the real-time output frequency of the thermal power units into the optimal target-strategy network for prediction to obtain the optimal control parameter combination of the thermal power units; S5. Adjust the parameters of the controllers corresponding to the thermal power units in the distribution network according to the optimal control parameter combination of the thermal power units.

2. The method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit according to claim 1, characterized in that: In S1, the specific process of judging whether the corresponding thermal power unit participates in the frequency regulation of the distribution network according to the historical output frequency is as follows: The output frequency deviation is calculated based on the historical output frequency and the rated frequency of the corresponding thermal power unit, and the output frequency deviation is compared with the preset frequency regulation threshold. If the historical output frequency deviation is greater than or equal to the absolute value of the frequency regulation threshold, it is determined that the thermal power unit does not participate in the frequency regulation of the distribution network. If the historical output frequency deviation is less than the absolute value of the frequency regulation threshold, it is determined that the thermal power unit participates in the frequency regulation of the distribution network.

3. The method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit according to claim 1, characterized in that: In S2, the specific process of calculating the initial state quantity set of the thermal power unit based on the historical output frequency of the thermal power unit is: S21, subtracting the historical output frequency of the thermal power unit from the rated frequency of the thermal power unit to obtain an output frequency deviation; Synchronously, calculating the integral value of the historical output frequency according to the indefinite integral formula and the historical output frequency of the thermal power unit; S22, integrating the integral value of the historical output frequency and the output frequency deviation to obtain an initial state quantity set of the thermal power unit.

4. The method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit according to claim 1, characterized in that: In S3, the specific process of inputting the initial state quantity set of the thermal power unit into the target-strategy network for training to obtain the optimal target-strategy network is: S31, inputting the initial state quantity set of the thermal power unit into the target-strategy network for prediction and calculation to obtain the current action quantity, the current reward quantity and the next state quantity; S32, sorting out the initial state quantity, the current action quantity, the current reward quantity and the next state quantity to obtain a state set, and judging whether to end the state update of the thermal power unit according to the current time step of the target-strategy network and the preset time step threshold; S33. After the state update of the thermal power unit is completed, determine whether the target-strategy network training is completed based on the current training times of the target-strategy network and the preset training cycle. If so, mark the target-strategy network corresponding to the current training times as the optimal target-strategy network. If not, change the target-strategy network based on the random initialization formula, update the initial state quantity of the thermal power unit based on the state set and execute S31.

5. The method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit according to claim 4, characterized in that: In S31, there is a preparatory step before inputting the initial state quantity set of the thermal power unit into the target-strategy network. The preparatory step is specifically: dividing the initial state quantity set of the thermal power unit into at least one small batch state set according to the preset number of training times and training amount, and inputting the small batch state set into the target-strategy network.

6. The method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit according to claim 4, characterized in that: In S33, the specific process of changing the target-strategy network based on the random initialization formula and updating the initial state quantity of the thermal power unit based on the state set is: S331, calculating a selection index according to the current training times of the target-strategy network, a preset update frequency and an update selection formula, and selecting an update target based on the selection index; S332: when the update target is a target network in a target-strategy network, changing parameters of the target network based on a random initialization formula; When the update target is a target network and a policy network in a target-policy network, changing the parameters of the target network based on a random initialization formula, and changing the parameters of the policy network based on a random initialization formula; Synchronously, at least one state quantity is sampled from the state set as the initial state quantity of the thermal power unit.

7. The method for adjusting the frequency of a distribution network based on the frequency characteristics of a thermal power unit according to claim 6, characterized in that: In S331, the update selection formula is specifically: k%×C=1; Where k is the current training number of the target-policy network, and C is the update frequency.

8. A distribution network frequency regulation system, applicable to the distribution network frequency regulation method based on the frequency characteristics of thermal power units as claimed in any one of claims 1 to 7, characterized in that: Including: controller, speed governor, thermal power unit; The control end of the controller is electrically connected to the controlled end of the speed regulator, the data collection end of the controller is electrically connected to the data collection end of the thermal power unit, the control end of the speed regulator is electrically connected to the controlled end of the thermal power unit, and the output end of the thermal power unit is electrically connected to the external distribution network.

9. A distribution network frequency regulation system according to claim 8, characterized in that: The controller includes: a state observer, a proportional-differential controller; The input end of the state observer is electrically connected to the data acquisition end of the thermal power unit as the data collection end of the controller, the output end of the state observer is electrically connected to the input end of the proportional-differential controller, and the control end of the proportional-differential controller is electrically connected to the controlled end of the speed regulator as the control end of the controller.

10. A distribution network frequency regulation system according to claim 8, characterized in that: The thermal power unit comprises: Reheat boilers, steam turbines and generator sets; The output end of the reheat boiler is connected to the input end of the steam turbine, the monitoring end of the reheat boiler is electrically connected to the output end of the steam turbine, the output end of the steam turbine is connected to the output end of the generator set, and the output end of the generator set is electrically connected to the external distribution network.

Images