CN118157224A - Automatic parameter adjusting method for wind farm and related components - Google Patents
Automatic parameter adjusting method for wind farm and related components Download PDFInfo
- Publication number
- CN118157224A CN118157224A CN202410284635.2A CN202410284635A CN118157224A CN 118157224 A CN118157224 A CN 118157224A CN 202410284635 A CN202410284635 A CN 202410284635A CN 118157224 A CN118157224 A CN 118157224A
- Authority
- CN
- China
- Prior art keywords
- power
- simulation
- target power
- field level
- wind turbine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000004088 simulation Methods 0.000 claims abstract description 148
- 238000012549 training Methods 0.000 claims abstract description 42
- 230000004044 response Effects 0.000 claims description 46
- 230000006870 function Effects 0.000 claims description 30
- 230000009471 action Effects 0.000 claims description 16
- 230000008859 change Effects 0.000 claims description 12
- 230000002787 reinforcement Effects 0.000 claims description 11
- 238000013531 bayesian neural network Methods 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000010248 power generation Methods 0.000 abstract description 7
- 230000008569 process Effects 0.000 description 7
- 230000007246 mechanism Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000013178 mathematical model Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000013459 approach Methods 0.000 description 2
- 238000011217 control strategy Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000000630 rising effect Effects 0.000 description 2
- 230000003828 downregulation Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000011990 functional testing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000003827 upregulation Effects 0.000 description 1
Landscapes
- Feedback Control In General (AREA)
Abstract
The invention discloses an automatic parameter adjusting method of a wind power plant and related components, and relates to the field of wind power generation, wherein a pre-learning model is obtained based on simulation plant level target power and simulation plant level output power training; based on the actual target power of the field level and the actual output power of the field level, the obtained learning model outputs final PID control parameters so that the wind power plant can adjust the actual output power of the field level according to the updated PID control parameters. And simulating a plurality of wind turbines through a simulation platform, and taking the obtained field-level simulation target power and the field-level simulation output power of the simulation platform as training data of a pre-learning model. In practical application, retraining the pre-learning model according to the actual target power of the field level and the actual output power of the field level as training data, and adjusting the PID control parameters of the wind power plant, so that the actual output power of the field level can track the actual target power rapidly and accurately.
Description
Technical Field
The invention relates to the field of wind power generation, in particular to an automatic parameter adjusting method for a wind farm and related components.
Background
The new energy stations connected in the grid are all provided with a set of energy management system, power instructions scheduled by the power grid are received upwards, and target power is issued to the on-site units downwards. The energy management system must employ a closed loop control strategy with higher accuracy to ensure timely successful grid connection of the wind farm sites. In order to ensure response accuracy and response speed, a common PID (Proportion-Integral-Differential) is used to realize closed-loop control, namely, a plurality of PID control parameters are set to carry out coordinated control on the real-time output of the units in the wind farm according to a field-level power instruction issued by real-time power grid dispatching and the real-time power feedback of all units in the wind farm, so that the deviation between target power and real-time power is reduced. Because wind power generation is influenced by surrounding environment, own components and single machine algorithm parameters, an accurate mathematical model cannot be established, so that control parameter setting is more difficult. The related technology adopts a manual parameter adjusting method, so that field personnel are required to have firm professional background knowledge and rich practical operation experience, the parameter adjusting difficulty is caused, and the time cost for parameter adjusting is high.
Disclosure of Invention
The invention aims to provide an automatic parameter adjusting method for a wind power plant and related components, which are used for adjusting PID control parameters of a wind power unit according to a learning model in practical application, so that the practical output power of the wind power unit approaches to the practical target power, and the automatic adjustment is more accurate and rapid.
In order to solve the technical problems, the invention provides an automatic parameter adjusting method for a wind farm, which comprises the following steps:
Outputting the simulation field level target power to a simulation platform to obtain simulation field level output power of the wind turbine output by the simulation platform, wherein the simulation platform is used for simulating the wind turbine;
Training a pre-learning model based on the simulation field level target power and the simulation field level output power, wherein the pre-learning model is used for generating PID control parameters, and the PID control parameters are used for controlling the simulation field level output power to change to the simulation field level target power;
When determining the actual field level target power of the wind turbine, transmitting the actual field level target power to the wind turbine;
Inputting the actual field level output power of the wind turbine corresponding to the actual field level target power into the pre-learning model so that the pre-learning model continues training to obtain a learning model after continuing training;
And acquiring updated PID control parameters output by the learning model so that the wind turbine generator can adjust the actual output power of the wind turbine generator according to the updated PID control parameters.
In another aspect, before outputting the simulation field level target power to the simulation platform, the method further comprises:
Dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level;
before sending the actual field level target power to the wind turbine, the method further comprises:
and dividing the actual unit target power for each wind turbine in the wind power plant according to the actual field level target power.
In another aspect, before outputting the simulation field level target power to the simulation platform, the method further comprises:
receiving a response state, a load level and a power variation of the wind farm, wherein the response state comprises a power limiting instruction and an unlimited power instruction;
Dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level, wherein the method comprises the following steps:
And dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level, the response state of the wind power field, the load level and the power variation.
In another aspect, determining an actual farm level target power for the wind turbine includes:
determining the actual field level target power of the wind turbine generator according to a preset acquisition period, wherein the relation of the preset acquisition period is that
Wherein T is the preset acquisition period, T n is the duration of the received continuous instruction, N is the number of times the variation of the power is changed, and G is different load levels.
In another aspect, the process of training the pre-learning model includes:
Adjusting an action input at of a Critic network in the learning model, wherein the expression of at is at=μ (Δ SetW, para) +b (Δ SetW, para);
Wherein μ (Δ SetW, para) is a policy value obtained by an Actor neural network of the deep reinforcement learning DDPG algorithm, B (Δ SetW, para) is motion search noise after bayesian neural network optimization, Δ SetW is an adjustment amount of full-field power, and Para is a corresponding field-level PID parameter.
In another aspect, training a pre-learning model based on the simulated field level target power and the simulated field level output power includes:
Constructing a reward function according to a response index of the wind turbine from the current field level output power to the field level target power, wherein the value of the reward function is inversely related to the difference between the current field level output power and the field level target power, and the response index is related to the time spent by the wind turbine and the current output power to be adjusted to the target power;
And outputting the PID control parameter corresponding to the current output power and the target power when the value of the reward function is maximum.
In another aspect, the bonus function is expressed as
When f (delta SetW, para) is the value of the reward function and the target power changes once, the reward function value of the PID parameter corresponding to the target power is taken to be-1 when the response index does not meet the national standard in the current period, 1 is taken when all the response indexes meet the national standard in the current period, delta SetW is the adjustment quantity of the full-field power, para is the corresponding field-level PID parameter, t0 is the lag time required by the national standard, the response time required by tp, the adjustment time required by ts national standard, the overshoot required by ct national standard, the number of times of change of the control target in the calculation period is taken, G is different load levels, and the load levels are the ratio of the theoretical power to the installed capacity of the full-field unit.
In order to solve the technical problems, the invention also provides an automatic parameter adjusting system of the wind farm, which comprises the following components:
the simulation unit is used for outputting the simulation field-level target power to a simulation platform to obtain the simulation field-level output power of the wind turbine output by the simulation platform, and the simulation platform is used for simulating the wind turbine;
The first training unit is used for training a pre-learning model based on the simulation field level target power and the simulation field level output power, wherein the pre-learning model is used for generating PID control parameters, and the PID control parameters are used for controlling the simulation field level output power to change to the simulation field level target power;
The actual target power determining unit is used for transmitting the actual field level target power to the wind turbine generator when determining the actual field level target power of the wind turbine generator;
the second training unit is used for inputting the actual field level output power of the wind turbine generator corresponding to the actual field level target power into the pre-learning model so that the pre-learning model continues training to obtain a learning model after continuing training;
And the control unit is used for acquiring the updated PID control parameters output by the learning model so that the wind turbine generator can adjust the actual output power of the wind turbine generator according to the updated PID control parameters.
In order to solve the technical problems, the invention also provides an automatic parameter adjusting device for a wind farm, which comprises the following components:
A memory for storing a computer program;
And the processor is used for realizing the steps of the automatic parameter adjusting method of the wind farm when executing the computer program.
In order to solve the technical problems, the invention also provides a wind farm, which comprises the wind farm automatic parameter adjusting device and a plurality of wind turbines connected with the wind farm automatic parameter adjusting device, wherein the wind farm automatic parameter adjusting device is used for outputting PID parameters of the wind turbines.
The application provides an automatic parameter adjusting method of a wind power plant and related components, and relates to the field of wind power generation, wherein a pre-learning model is obtained based on simulation plant level target power and simulation plant level output power training; based on the actual target power of the field level and the actual output power of the field level, the obtained learning model outputs final PID control parameters so that the wind power plant can adjust the actual output power of the field level according to the updated PID control parameters. The simulation platform simulates a plurality of wind turbines, the obtained field-level simulation target power and the field-level simulation output power of the simulation platform are used as training data of a pre-learning model, and the trained pre-learning model can be used as an initial model in practical application. In practical application, retraining the pre-learning model according to the actual target power of the field level and the actual output power of the field level as training data, and adjusting the PID control parameters of the wind power plant, so that the actual output power of the field level can track the actual target power rapidly and accurately.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required in the prior art and the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an automatic parameter adjustment method for a wind farm provided by the invention;
FIG. 2 is a flowchart of another wind farm automatic parameter adjustment method provided by the invention;
FIG. 3 is a schematic structural diagram of an automated parameter adjustment system for a wind farm provided by the invention;
fig. 4 is a schematic structural diagram of an automatic parameter adjusting device for a wind farm.
Detailed Description
The invention provides an automatic parameter adjusting method and related components for a wind power plant, wherein PID control parameters of the wind power plant are adjusted according to a learning model in practical application, so that the practical output power of the wind power plant approaches to practical target power, and the automatic adjustment is more accurate and rapid.
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The new energy stations connected with the grid at present are all provided with a set of energy management system, the power instruction of power grid dispatching is received upwards, and target power is issued downwards to the on-site units. As the national standard, industry standard and enterprise standard are more and more strict to the control performance index of the wind farm, the energy management system must adopt a closed-loop control strategy with higher precision to ensure that the wind farm stations are successfully connected in time. In order to ensure response accuracy and response speed, a common expert PID realizes closed-loop control, namely, a plurality of control parameters (trigger threshold values and proportional, integral and differential parameters of each rule of the expert PID) are set according to a field level power instruction issued by real-time power grid dispatching and real-time power feedback of all units in a wind power plant to carry out coordinated control on the real-time output of the units in the wind power plant, so that the deviation between target power and real-time power is reduced. Because the field-level wind power generation is influenced by surrounding environment, own components and single machine algorithm parameters, an accurate mathematical model cannot be established, so that the control parameter setting is more difficult. The manual parameter adjusting method requires on-site personnel to have firm professional background knowledge and rich practical operation experience, and the time cost for parameter adjustment is high.
FIG. 1 is a flowchart of an automatic parameter adjustment method for a wind farm, provided by the invention, comprising:
S11: outputting the simulation field level target power to a simulation platform to obtain simulation field level output power of the wind turbine output by the simulation platform, wherein the simulation platform is used for simulating the wind turbine;
S12: training a pre-learning model based on the simulation field level target power and the simulation field level output power, wherein the pre-learning model is used for generating PID control parameters, and the PID control parameters are used for controlling the simulation field level output power to change to the simulation field level target power;
Before the wind turbine generator is put into use, a learning model needs to be trained in advance, the learning model replaces manpower in the related technology, and PID control parameters of the wind turbine generator can be automatically generated by the learning model so as to control the wind turbine generator to adjust the current output power.
Since the data of the wind turbine cannot be used before the wind turbine is put into use, the wind turbine is simulated by adopting the simulation platform. The simulation platform is a semi-physical simulation platform and is provided with a true controller, but the fan mechanical part is a mathematical model. The simulation platform can model the actual work of the wind turbine generator, namely, after receiving the simulation target power, the simulation platform can work based on the simulation target power. The obtained simulation output power and simulation target power output by the simulation platform can be used as training data of a learning model. The learning model needs to generate PID control parameters, and the PID control parameters can control the simulation output power to change to the simulation target power. And when the simulation output power is equal to the simulation target power, determining that the wind turbine generator set corresponding to the simulation platform runs accurately.
Each large host manufacturer is provided with a semi-physical simulation platform, and simulation test and optimization of an algorithm are realized based on the platform. Compared with a pure digital model, the semi-physical simulation platform is closer to the field real environment. And replacing a real fan with the semi-physical simulation model, and responding to the power instruction of the energy management platform in real time. And pre-training an automatic parameter adjustment model of the on-line Bayes deep reinforcement learning by taking a real-time response result of the semi-physical simulation platform as input.
S13: when determining the actual field level target power of the wind turbine, transmitting the actual field level target power to the wind turbine;
s14: inputting the actual field level output power of the wind turbine corresponding to the actual field level target power into a pre-learning model so that the pre-learning model continues training to obtain a learning model after continuing training;
S15: and acquiring updated PID control parameters output by the learning model so that the wind turbine generator adjusts the actual output power of the wind turbine generator according to the updated PID control parameters.
In the actual application process, the actual target power of the wind turbine generator is determined, the actual output power and the actual target power of the wind turbine generator are input into a learning model, PID control parameters output by the learning model are obtained, at the moment, the wind turbine generator adjusts the output power according to the updated PID control parameters, and finally the actual output power of the wind turbine generator can reach the actual target power.
Specifically, the actual target power of the wind turbine generator may be the actual target power of a single wind turbine generator, or the actual target power of the wind farm may be received, and the controller determines how much actual target power is specifically allocated to each wind turbine generator.
The application provides an automatic parameter adjusting method for a wind power plant, which relates to the field of wind power generation and comprises the steps of obtaining a pre-learning model based on simulation plant level target power and simulation plant level output power training; based on the actual target power of the field level and the actual output power of the field level, the obtained learning model outputs final PID control parameters so that the wind power plant can adjust the actual output power of the field level according to the updated PID control parameters. And simulating a plurality of wind turbines through a simulation platform, and taking the obtained field-level simulation target power and the field-level simulation output power of the simulation platform as training data of a pre-learning model. In practical application, retraining the pre-learning model according to the actual target power of the field level and the actual output power of the field level as training data, and adjusting the PID control parameters of the wind power plant, so that the actual output power of the field level can track the actual target power rapidly and accurately.
Based on the above embodiments:
FIG. 2 is a flowchart of another wind farm automatic parameter adjustment method provided by the invention;
in some embodiments, before outputting the simulation field level target power to the simulation platform, further comprising:
Dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level;
before sending the actual field level target power to the wind turbine, the method further comprises:
and dividing the actual unit target power for each wind turbine in the wind power plant according to the actual field level target power.
The energy management platform in the controller is used as a control mechanism facing the fan, and needs to receive the instruction issued by the scheduling AGC. The AGC (Automatic Generation Control) automatic power generation control system is a set of control system between a power grid and a grid-connected wind power plant. For a grid-connected wind farm, a farm-level control instruction issued by an AGC system needs to be received in real time. When the wind farm is not connected with the grid or the wind farm needs to do relevant functional test, the AGC system is not connected or needs to be shielded. The virtual AGC system may replace the function of the AGC system at this point, issue control instructions and some other control information such as voltage, frequency, etc. to the wind farm station, and the instructions at this point may be issued in real time according to the test requirements. In the test stage, since the AGC system is not accessed, the function can be realized through the virtual AGC system, and the instruction is issued to the energy management system in a certain period. For providing effective data to an automated parameter tuning system. The virtual AGC system is required to circularly issue corresponding instructions according to a certain rule. To ensure the completeness of the test, the virtual AGC system issues a cycle command that includes all conditions of different response states, different load levels, and different amounts of variation.
In some embodiments, before outputting the simulation field level target power to the simulation platform, further comprising:
Receiving a response state, a load level and a power variation of a wind power plant, wherein the response state comprises a power limiting instruction and an unlimited power instruction;
Dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level, wherein the method comprises the following steps:
and dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level, the response state of the wind power field, the load level and the power variation.
The controller receives the command of the AGC system in real time, distributes power to the fan in real time based on the current control parameters, and performs data messaging and forwarding. Typically, the response state includes a full-field power-limited command and a full-field power-unlimited command; the load level should include a low load and a high load, and it is generally considered that the theoretical power is lower than 50% of the installed capacity, otherwise the load is higher. The variation is divided into 5%, 10%, 20% of the power up-regulation and down-regulation full-field power.
Specifically, when the wind force is large, a full-field power limiting command exists, and when the wind force is small, a full-field power limiting command exists.
In some embodiments, determining an actual farm level target power for a wind turbine includes:
Determining the actual field level target power of the wind turbine generator according to a preset acquisition period, wherein the relation of the preset acquisition period is that
Wherein, T is a preset acquisition period, T n is the duration of a received continuous instruction, N is the number of times of the variable quantity quality of the power, and G is different load levels.
In order to ensure more accurate evaluation of each set of parameters, real feedback under different working conditions of each set of parameters is required, corresponding preset acquisition periods are also different under different load levels, and the acquisition process can be more accurate by continuously adjusting the corresponding preset acquisition periods.
In some embodiments, the process of training the pre-learning model includes:
adjusting an action input at of a Critic network in the learning model, wherein the expression of at is at=μ (Δ SetW, para) +b (Δ SetW, para);
Wherein μ (Δ SetW, para) is a policy value obtained by the Actor neural network of the deep reinforcement learning DDPG algorithm, and B (Δ SetW, para) is action search noise after the bayesian neural network is optimized.
The action exploration mechanism of the deep reinforcement learning algorithm DDPG relies on the Actor network to derive an action value that maximizes rewards and adds random numbers that obey gaussian distribution to explore the action space around that value. However, a method of searching for an action space with a predetermined gaussian distribution requires more searching time.
The Bayesian optimization can be realized by continuously updating the priori according to the previous parameter information, so that the iteration times are reduced, and the local or global optimal value meeting the condition is obtained in a shorter time.
The parameters of the bayesian neural network model are not fixed values and the uncertainty of the data and model can be evaluated. Therefore, the action exploration strategy can be optimized by means of the Bayesian neural network, the iteration times are reduced, convergence is realized rapidly, and the local or global optimal value meeting the conditions is obtained in a shorter time. The Bayesian neural network takes historical data as a priori, takes current power adjustment quantity, current control parameters, response indexes of the current parameters and response indexes meeting national standard requirements as inputs, and takes the difference value between the control parameters meeting the national standard requirements and the current parameters as output to train the neural network.
In some embodiments, training the pre-learning model based on the simulated field level target power and the simulated field level output power includes:
constructing a reward function according to a response index of the wind turbine from the current field level output power to the field level target power, wherein the value of the reward function is inversely related to the difference between the current field level output power and the field level target power, and the response index is related to the time spent by the wind turbine from the current output power to the target power;
And outputting PID control parameters corresponding to the current field level output power and the field level target power when the value of the reward function is maximum.
The automatic parameter adjustment model of the online Bayes deep reinforcement learning is built, and a clear rewarding mechanism and an action exploration mechanism are required to be built. Based on the data of the energy management platform, the response index results in the time period comprise lag time, rising time, adjusting time and overshoot (the response index comprises lag time, rising time, adjusting time and overshoot), the response index is the response index of the whole field unit responding to the command when the command issued by the AGC changes, and the reward function is constructed by taking national standard requirements as the maximum allowable value. And optimizing a deep reinforcement learning action exploration mode by taking real-time data of the energy management platform and corresponding rewarding values as data bases, and exploring the optimal action based on the Bayesian neural network so as to reduce iteration times.
The application evaluates the current PID control parameters by constructing the reward function, determines whether the PID control parameters reach the best or not from the angle of adjusting time, and is more accurate.
In some embodiments, the expression of the bonus function is
Wherein f (delta SetW, para) is a reward function value of a PID parameter corresponding to the target power when the target power is changed once, taking-1 when the response index does not meet the national standard in the current period, taking 1 when all the response indexes meet the national standard in the current period, delta SetW is an adjustment quantity of full-field power, para is a corresponding field-level PID parameter, t0 is a lag time required by the national standard, a response time required by the tp national standard, an adjustment time required by the ts national standard, overshoot required by the ct national standard, N is the number of times of change of a control target in the calculation period, G is different load levels, and the load level is the ratio of the theoretical power to the installed capacity of the full-field unit.
And establishing a reward mechanism by taking the result as a guide. The national standard is used for setting the rewarding function, and the accurate mathematical model is replaced by the rewarding function for deep strong chemistry. And evaluating the strategy obtained by the deep reinforcement learning by using a reward function. It is assumed that when the parameter value obtained by the deep reinforcement learning satisfies the response index and the value of the prize function is larger at this time due to the national standard, the deep reinforcement learning will find the parameter value that maximizes the prize when deciding.
Fig. 3 is a schematic structural diagram of an automatic parameter adjusting system for a wind farm, provided by the invention, the automatic parameter adjusting system for the wind farm comprises:
the simulation unit 31 is configured to output a simulation field level target power to a simulation platform, obtain a simulation field level output power of the wind turbine output by the simulation platform, where the simulation platform is used for simulating the wind turbine;
A first training unit 32 for training a pre-learning model based on the simulation field level target power and the simulation field level output power, the pre-learning model being used for generating PID control parameters for controlling the simulation field level output power to change to the simulation field level target power;
An actual target power determining unit 33, configured to send the actual field level target power to the wind turbine when determining the actual field level target power of the wind turbine;
the second training unit 34 is configured to input an actual field level output power of the wind turbine generator corresponding to the actual field level target power to the pre-learning model, so that the pre-learning model continues training, and a learning model after continuing training is obtained;
And the control unit 35 is used for acquiring the updated PID control parameters output by the learning model so that the wind turbine generator can adjust the actual output power of the wind turbine generator according to the updated PID control parameters.
On the basis of the above embodiment, the method further comprises:
The first dividing unit is used for dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level;
the second dividing unit is used for dividing the actual unit target power for each wind turbine unit in the wind power plant according to the actual field level target power.
The control instruction receiving unit is used for receiving the response state, the load level and the power variation of the wind power plant, wherein the response state comprises a power limiting instruction and an unlimited power instruction;
the first dividing unit is specifically configured to divide the target power of the simulation unit for each simulated wind turbine unit in the simulation platform according to the target power of the simulation field level, the response state of the wind power field, the load level and the variation of the power.
The actual total target power receiving unit is specifically configured to determine an actual field level target power of the wind turbine according to a preset acquisition period, where a relation of the preset acquisition period is as follows
Wherein, T is a preset acquisition period, T n is the duration of the received continuous instruction, N is the number of times of variation change, and G is different load levels.
An adjustment unit for adjusting an action input at of the Critic network in the learning model, wherein the expression of at is at=μ (Δ SetW, para) +b (Δ SetW, para);
Wherein μ (Δ SetW, para) is a policy value obtained by an Actor neural network of the deep reinforcement learning DDPG algorithm, B (Δ SetW, para) is motion search noise after bayesian neural network optimization, Δ SetW is an adjustment amount of full-field power, and Para is a corresponding field-level PID parameter.
The first training unit 32 is specifically configured to construct a reward function according to a response index of the wind turbine from the current field level output power to the field level target power, where a value of the reward function is inversely related to a difference between the current field level output power and the field level target power, and the response index is related to a time taken by the wind turbine and the current output power to be adjusted to the target power;
and the output unit is used for outputting the PID control parameter corresponding to the current output power and the target power when the value of the reward function is maximum.
The expression of the reward function is
Wherein f (delta SetW, para) is a reward function value of a PID parameter corresponding to the target power when the target power is changed once, taking-1 when the response index does not meet the national standard in the current period, taking 1 when all the response indexes meet the national standard in the current period, delta SetW is an adjustment quantity of full-field power, para is a corresponding field-level PID parameter, t0 is a lag time required by the national standard, a response time required by the tp national standard, an adjustment time required by the ts national standard, overshoot required by the ct national standard, N is the number of times of change of a control target in the calculation period, G is different load levels, and the load level is the ratio of the theoretical power to the installed capacity of the full-field unit.
The description of the wind farm automatic parameter adjustment system provided by the application refers to the above embodiment, and is not repeated here.
Fig. 4 is a schematic structural diagram of an automatic parameter adjusting device for a wind farm, provided by the invention, the automatic parameter adjusting device for the wind farm comprises:
a memory 41 for storing a computer program;
the processor 42 is configured to execute the computer program to implement the steps of the wind farm automatic parameter adjustment method described above.
The description of the automatic parameter adjusting device for the wind farm provided by the application refers to the above embodiment, and is not repeated here.
The application also provides a wind farm, which comprises the wind farm automatic parameter adjusting device and a plurality of wind turbines connected with the wind farm automatic parameter adjusting device, wherein the wind farm automatic parameter adjusting device is used for outputting PID parameters of the wind turbines.
The wind farm provided by the application is described with reference to the above embodiments, and will not be described herein.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
It should also be noted that in this specification, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. An automatic parameter adjusting method for a wind farm is characterized by comprising the following steps:
Outputting the simulation field level target power to a simulation platform to obtain simulation field level output power of the wind turbine output by the simulation platform, wherein the simulation platform is used for simulating the wind turbine;
Training a pre-learning model based on the simulation field level target power and the simulation field level output power, wherein the pre-learning model is used for generating PID control parameters, and the PID control parameters are used for controlling the simulation field level output power to change to the simulation field level target power;
When determining the actual field level target power of the wind turbine, transmitting the actual field level target power to the wind turbine;
Inputting the actual field level output power of the wind turbine corresponding to the actual field level target power into the pre-learning model so that the pre-learning model continues training to obtain a learning model after continuing training;
And acquiring updated PID control parameters output by the learning model so that the wind turbine generator can adjust the actual output power of the wind turbine generator according to the updated PID control parameters.
2. The automated wind farm parameter tuning method of claim 1, further comprising, prior to outputting the simulated farm-level target power to the simulation platform:
Dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level;
before sending the actual field level target power to the wind turbine, the method further comprises:
and dividing the actual unit target power for each wind turbine in the wind power plant according to the actual field level target power.
3. The automated wind farm parameter tuning method of claim 2, further comprising, prior to outputting the simulated farm-level target power to the simulation platform:
receiving a response state, a load level and a power variation of the wind farm, wherein the response state comprises a power limiting instruction and an unlimited power instruction;
Dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level, wherein the method comprises the following steps:
And dividing the target power of the simulation unit for each simulation wind turbine in the simulation platform according to the target power of the simulation field level, the response state of the wind power field, the load level and the power variation.
4. A method for automated wind farm scaling as defined in claim 3, wherein determining the actual farm level target power for the wind turbine includes:
determining the actual field level target power of the wind turbine generator according to a preset acquisition period, wherein the relation of the preset acquisition period is that
Wherein T is the preset acquisition period, T n is the duration of the received continuous instruction, N is the number of times the variation of the power is changed, and G is different load levels.
5. The automated wind farm parameter tuning method of claim 1, wherein training the pre-learning model comprises:
Adjusting an action input at of a Critic network in the learning model, wherein the expression of at is at=μ (Δ SetW, para) +b (Δ SetW, para);
Wherein μ (Δ SetW, para) is a policy value obtained by an Actor neural network of the deep reinforcement learning DDPG algorithm, B (Δ SetW, para) is motion search noise after bayesian neural network optimization, Δ SetW is an adjustment amount of full-field power, and Para is a corresponding field-level PID parameter.
6. The automated wind farm parameter tuning method of any of claims 1 to 5, wherein training a pre-learning model based on the simulated farm level target power and the simulated farm level output power comprises:
Constructing a reward function according to a response index of the wind turbine from the current field level output power to the field level target power, wherein the value of the reward function is inversely related to the difference between the current field level output power and the field level target power, and the response index is related to the time spent by the wind turbine and the current output power to be adjusted to the target power;
And outputting the PID control parameter corresponding to the current output power and the target power when the value of the reward function is maximum.
7. The automated wind farm parameter tuning method of claim 6, wherein the expression of the reward function is
When f (delta SetW, para) is the value of the reward function and the target power changes once, the reward function value of the PID parameter corresponding to the target power is taken to be-1 when the response index does not meet the national standard in the current period, 1 is taken when all the response indexes meet the national standard in the current period, delta SetW is the adjustment quantity of the full-field power, para is the corresponding field-level PID parameter, t0 is the lag time required by the national standard, the response time required by tp, the adjustment time required by ts national standard, the overshoot required by ct national standard, the number of times of change of the control target in the calculation period is taken, G is different load levels, and the load levels are the ratio of the theoretical power to the installed capacity of the full-field unit.
8. An automated parameter tuning system for a wind farm, comprising:
the simulation unit is used for outputting the simulation field-level target power to a simulation platform to obtain the simulation field-level output power of the wind turbine output by the simulation platform, and the simulation platform is used for simulating the wind turbine;
The first training unit is used for training a pre-learning model based on the simulation field level target power and the simulation field level output power, wherein the pre-learning model is used for generating PID control parameters, and the PID control parameters are used for controlling the simulation field level output power to change to the simulation field level target power;
The actual target power determining unit is used for transmitting the actual field level target power to the wind turbine generator when determining the actual field level target power of the wind turbine generator;
the second training unit is used for inputting the actual field level output power of the wind turbine generator corresponding to the actual field level target power into the pre-learning model so that the pre-learning model continues training to obtain a learning model after continuing training;
And the control unit is used for acquiring the updated PID control parameters output by the learning model so that the wind turbine generator can adjust the actual output power of the wind turbine generator according to the updated PID control parameters.
9. An automatic parameter adjusting device for a wind farm, comprising:
A memory for storing a computer program;
a processor for implementing the steps of the wind farm automatic parameter tuning method according to any of the claims 1 to 7 when executing said computer program.
10. A wind farm, comprising the wind farm automation parameter adjusting device according to claim 9, and further comprising a plurality of wind turbines connected with the wind farm automation parameter adjusting device, wherein the wind farm automation parameter adjusting device is used for outputting PID parameters of the wind turbines.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410284635.2A CN118157224A (en) | 2024-03-13 | 2024-03-13 | Automatic parameter adjusting method for wind farm and related components |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410284635.2A CN118157224A (en) | 2024-03-13 | 2024-03-13 | Automatic parameter adjusting method for wind farm and related components |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118157224A true CN118157224A (en) | 2024-06-07 |
Family
ID=91299540
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410284635.2A Pending CN118157224A (en) | 2024-03-13 | 2024-03-13 | Automatic parameter adjusting method for wind farm and related components |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118157224A (en) |
-
2024
- 2024-03-13 CN CN202410284635.2A patent/CN118157224A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Coordinated load frequency control of multi-area integrated energy system using multi-agent deep reinforcement learning | |
| Aghaei et al. | Scenario-based dynamic economic emission dispatch considering load and wind power uncertainties | |
| CN111523737A (en) | Automatic optimization-approaching adjusting method for operation mode of electric power system driven by deep Q network | |
| CN113852098B (en) | Automatic power generation control scheduling method based on multi-target dragonfly algorithm | |
| Eslami et al. | Optimal design of PID-based low-pass filter for gas turbine using intelligent method | |
| Yin et al. | Mode-decomposition memory reinforcement network strategy for smart generation control in multi-area power systems containing renewable energy | |
| Chertovskikh et al. | An adaptive PID controller with an online auto-tuning by a pretrained neural network | |
| CN115345380A (en) | New energy consumption electric power scheduling method based on artificial intelligence | |
| CN115795992A (en) | Park energy Internet online scheduling method based on virtual deduction of operation situation | |
| CN115764870A (en) | Multivariable photovoltaic power generation power prediction method and device based on automatic machine learning | |
| CN115764931A (en) | Automatic power generation control method, system, equipment and medium for power system | |
| CN114566971A (en) | Real-time optimal power flow calculation method based on near-end strategy optimization algorithm | |
| CN113885328A (en) | Nuclear power tracking control method based on integral reinforcement learning | |
| CN116755409B (en) | Coal-fired power generation system coordination control method based on value distribution DDPG algorithm | |
| CN111064228B (en) | Wind turbine generator droop control method and system considering wind speed and load change scene and computer equipment | |
| CN117375097A (en) | Photovoltaic coordination autonomous method based on multi-agent coordination control strategy and reinforcement learning | |
| CN118157224A (en) | Automatic parameter adjusting method for wind farm and related components | |
| CN116963461A (en) | Energy saving method and device for machine room air conditioner | |
| CN114841595A (en) | Deep-enhancement-algorithm-based hydropower station plant real-time optimization scheduling method | |
| CN110210113B (en) | Wind power plant dynamic equivalent parameter intelligent checking method based on deterministic strategy gradient | |
| CN114326395A (en) | Intelligent generator set control model online updating method based on working condition judgment | |
| Ahamed et al. | Reinforcement learning controllers for automatic generation control in power systems having reheat units with GRC and dead-band | |
| CN106773685A (en) | A kind of angle PI controller tuning methods for wind power yawing system | |
| CN116300430B (en) | MPC control parameter optimizing method and application thereof in parallel connection platform | |
| CN114167717B (en) | Thermal power unit DEH rotating speed control method based on improved PSO-fuzzy PID |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |