CN117031950A - Modeling method and device for deep peak-shaving thermal power generating unit control system - Google Patents
Modeling method and device for deep peak-shaving thermal power generating unit control system Download PDFInfo
- Publication number
- CN117031950A CN117031950A CN202310989293.XA CN202310989293A CN117031950A CN 117031950 A CN117031950 A CN 117031950A CN 202310989293 A CN202310989293 A CN 202310989293A CN 117031950 A CN117031950 A CN 117031950A
- Authority
- CN
- China
- Prior art keywords
- data
- thermal power
- deep peak
- period
- unit
- 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 157
- 230000001105 regulatory effect Effects 0.000 claims abstract description 107
- 230000008569 process Effects 0.000 claims abstract description 70
- 238000003303 reheating Methods 0.000 claims abstract description 59
- 238000002485 combustion reaction Methods 0.000 claims abstract description 51
- 238000010801 machine learning Methods 0.000 claims abstract description 29
- 238000012544 monitoring process Methods 0.000 claims abstract description 20
- 238000012360 testing method Methods 0.000 claims description 50
- 238000012549 training Methods 0.000 claims description 48
- 238000004422 calculation algorithm Methods 0.000 claims description 28
- 238000013528 artificial neural network Methods 0.000 claims description 26
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 21
- 238000010977 unit operation Methods 0.000 claims description 18
- 239000003245 coal Substances 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 11
- 230000007774 longterm Effects 0.000 claims description 11
- 125000004122 cyclic group Chemical group 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 claims description 4
- 238000003860 storage Methods 0.000 claims description 4
- 238000013461 design Methods 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 abstract description 2
- 238000004364 calculation method Methods 0.000 description 19
- 210000002569 neuron Anatomy 0.000 description 10
- 238000013499 data model Methods 0.000 description 9
- 210000004027 cell Anatomy 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 150000003839 salts Chemical class 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 238000009825 accumulation Methods 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 239000010802 sludge Substances 0.000 description 4
- 238000011835 investigation Methods 0.000 description 3
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000004146 energy storage Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000006386 neutralization reaction Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011020 pilot scale process Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a method and a device for modeling a control system of a deep peak shaving thermal power unit, and relates to the technical field of design optimization; the method comprises the steps of modeling, namely obtaining historical operation data from a plant-level monitoring information system of a unit, constructing a boiler combustion system, a turbine speed regulating system, a reheat system, a turbine system and a unit cold end system, combining to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after regulating the level in a future period, wherein a prediction result is used as a basis to provide a reference for a control process in an actual operation process; the device comprises a modeling module, wherein the modeling module is based on five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, and is used for combining to obtain a deep peak shaving thermal power unit system model based on data driving and machine learning, so that modeling of a deep peak shaving thermal power unit control system is realized.
Description
Technical Field
The invention relates to the technical field of design optimization, in particular to a method and a device for modeling a deep peak shaving thermal power unit control system.
Background
The writer searches for (tacd= (boiler AND combustion turbine AND speed regulation AND reheat AND cold)), obtaining a closer prior art solution as follows.
The authorized bulletin number is CN115899660B, and the name is a peak shaving system and a peak shaving method of a coal-fired unit. The system comprises a sludge dryer, a coal-fired boiler, a steam turbine, a generator, a thermochemical energy storage device, a first heat exchanger, a second heat exchanger, a third heat exchanger, a first molten salt tank and a second molten salt tank, wherein the temperature of molten salt in the first molten salt tank is lower than that of molten salt in the second molten salt tank, and sludge to be dried in the sludge dryer is dried and peak shaving is carried out on a coal-fired unit through the thermochemical energy storage device, the first molten salt tank and the second molten salt tank. The provided scheme can improve the flexibility of the unit while realizing sludge drying.
The application publication number is CN115935624A, and the name is a dynamic process thermal power unit primary frequency modulation capability assessment method. Firstly, a single decoupling model of the steam turbine is established, and parameter identification is carried out on the single decoupling model according to historical data; then, calculating the whole heat accumulation of the boiler according to the cold fluid working medium heat accumulation of the boiler outlet and the metal heat accumulation of the boiler to obtain a dynamic response process equation of the primary frequency modulation performance of the steam turbine; and finally, combining the heat accumulation increment with a steam turbine model to obtain primary frequency modulation capacity prediction.
In combination with the two patent documents and the prior art, the inventors analyzed the prior art as follows.
The new energy power generation ratio is continuously improved under the background of carbon peak and carbon neutralization, the large-scale integration of the new energy has the characteristics of randomness, intermittence and volatility, and the construction of a novel power system taking the new energy as a main body puts higher requirements on the power supply regulation capability. The adjustment of the power system at the present stage mainly depends on the traditional thermal power, but the current grid-related technical index system, test and verification method under the deep peak regulation running state of the thermal power unit lacks system research and unified requirements, and the flexible running capability lacks accurate estimation. And establishing a thermal power unit control system model under the deep peak regulation working condition, and performing regulation performance monitoring and evaluation is an urgent requirement for power system stability.
Under the deep peak regulation operation working condition of the thermal power generating unit, the automatic power generation control AGC, primary frequency modulation and other grid-related adjustment capabilities of the unit are reduced, and the safe and stable operation of a power grid is affected. The control system which influences the network-related regulation performance of the unit has the advantages of large quantity and complex structure, large difficulty and poor accuracy by only manually judging the performance of the control system, and is difficult to realize comprehensive and accurate assessment of the regulation capacity of the thermal power unit under the deep peak regulation working condition; therefore, the performance change prediction evaluation method of the key control systems such as the main steam pressure, the main steam temperature, the reheat steam parameters and the like which influence the deep regulation capability of the unit is researched, and a related data model for monitoring and predicting the performance of the unit control system is established, so that the method has important significance for improving the regulation capability of the thermal power unit and the safety and stability of a power grid.
Problems and considerations in the prior art:
how to solve the technical problem of modeling of the deep peak-shaving thermal power generating unit control system.
Disclosure of Invention
The invention provides a modeling method and device for a deep peak shaving thermal power unit control system, which solve the technical problem of modeling of the deep peak shaving thermal power unit control system.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
the modeling method comprises the step of modeling, wherein the modeling step comprises the steps of obtaining historical operation data from a plant-level monitoring information system SIS of the unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after a regulation stage in a future period.
The further technical proposal is that: the modeling step specific division includes the steps of,
S1, extracting corresponding historical operation data from a plant-level monitoring information system SIS of a unit, wherein the historical operation data comprise historical operation data of main steam pressure and main steam temperature and deep peak regulation working conditions;
s2, sorting the historical operation data, and removing missing values; carrying out standardization processing on the data set, adjusting an input data structure in the historical data into three dimensions, and adjusting an output data structure into two dimensions;
s3, respectively constructing and obtaining five subsystem modules of a boiler combustion system, a turbine speed regulation system, a reheating system, a turbine system and a unit cold end system, respectively inputting historical operation data for training, and moving target data on time sequence according to the prediction duration and the prediction target;
s4, combining five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system to obtain a deep peak shaving thermal power unit system model based on data driving and machine learning, and realizing short-term prediction of pressure and load data after adjusting a stage in a future period, wherein the prediction result is used as a basis to realize control coordination work of the unit under a deep peak shaving working condition;
s5, after the unit generates new operation data, the generated real-time data is used as input quantity to be input into a deep peak shaving thermal power unit system model, and historical data with the same length are deleted on the basis, so that the real-time updating of the input quantity by the system is realized.
The further technical proposal is that: the method is characterized in that the building of five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system is based on a cyclic neural network (RNN) algorithm in machine learning, error magnitudes of a test set and a predicted value of three algorithms of a common RNN, an LSTM long-term artificial neural network and a GRU gate control unit are respectively compared, and an algorithm with the smallest error and the best predicted effect is selected as a data modeling mode.
The further technical proposal is that: the method comprises the steps of constructing a boiler combustion system prediction module based on a GRU (grid-control unit), taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a boiler combustion process, and advancing a target amount of main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is a period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
The method comprises the steps of constructing a steam turbine speed regulating system prediction module based on an LSTM long-term artificial neural network, taking a specific valve regulating instruction, main steam pressure and main steam temperature in the operation history data of the deep peak regulating thermal power unit as input quantity, taking the pressure after regulating stage in the operation history data of the deep peak regulating thermal power unit as target quantity, and constructing the input quantity and output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a turbine speed regulating system, and advancing the pressure of a target quantity regulation stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
constructing a reheating system prediction module based on a GRU gate control unit, taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak shaving thermal power unit as input amounts, taking the reheating steam pressure and the reheating steam temperature in the operation history data of the deep peak shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The further technical proposal is that: the method comprises the steps of constructing a steam turbine system prediction module based on a GRU gate control unit, taking the pressure after regulation, the reheat steam pressure and the reheat steam temperature in the operation history data of the deep peak shaving thermal power unit as input quantities, taking the unit load in the operation history data of the deep peak shaving thermal power unit as a target quantity, and constructing the input quantities and the output quantities into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target unit load by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
the method comprises the steps of constructing a unit cold end prediction module based on an LSTM long-term artificial neural network, taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate water temperature, ambient temperature and fan air quantity in the operation history data of the deep peak shaving thermal power unit as input quantities, taking the cold end heat exchange coefficient K in the operation history data of the deep peak shaving thermal power unit as a target quantity, and constructing the input quantities and the output quantities into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The modeling device of the deep peak regulation thermal power unit control system comprises a modeling module and a modeling module, wherein the modeling module is used for obtaining historical operation data from a plant level monitoring information system SIS of a unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak regulation thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after a regulation stage in a future period, and providing a reference for a control process in an actual operation process based on a prediction result, thereby realizing control coordination of the unit under a deep peak regulation working condition.
The further technical proposal is that: the system also comprises a boiler combustion system prediction module, a steam turbine speed regulation system prediction module and a reheating system prediction module,
the boiler combustion system prediction module is used for taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a boiler combustion process, and advancing a target amount of main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is a period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
The turbine speed regulating system prediction module is used for taking a specific valve regulating instruction, main steam pressure and main steam temperature in the operation history data of the deep peak regulating thermal power unit as input quantity, taking the regulated pressure in the operation history data of the deep peak regulating thermal power unit as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a turbine speed regulating system, and advancing the pressure of a target quantity regulation stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
the reheating system prediction module is used for taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the reheating steam pressure and the reheating steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The further technical proposal is that: the system also comprises a steam turbine system prediction module and a unit cold end prediction module,
the steam turbine system prediction module is used for taking the pressure after the regulation stage, the reheat steam pressure and the reheat steam temperature in the deep peak regulation thermal power unit operation history data as input quantity, taking the unit load in the deep peak regulation thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target unit load by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
the unit cold end prediction module is used for taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate temperature, ambient temperature and fan air quantity in the deep peak shaving thermal power unit operation history data as input quantity, taking the cold end heat exchange coefficient K in the deep peak shaving thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The device for modeling the deep peak-shaving thermal power unit control system comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor realizes corresponding steps in the method when executing the computer program.
The device for modeling the deep peak shaving thermal power generating unit control system comprises a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes corresponding steps in the method when being executed by a processor.
The beneficial effects of adopting above-mentioned technical scheme to produce lie in:
firstly, a modeling method of a deep peak-shaving thermal power unit control system comprises a modeling step, wherein the modeling step comprises the steps of obtaining historical operation data from a plant-level monitoring information system SIS of a unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after a regulating level in a future period of time, and providing a reference for a control process in an actual operation process based on a prediction result, thereby realizing control coordination of the unit under a deep peak-shaving working condition. According to the technical scheme, the deep peak shaving thermal power unit system model based on data driving and machine learning is obtained by combining five subsystem modules, namely a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, so that modeling of a deep peak shaving thermal power unit control system is realized.
Second, a device for modeling a deep peak-shaving thermal power unit control system comprises a modeling module, wherein the modeling module is used for obtaining historical operation data from a plant-level monitoring information system SIS of the unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after a regulating level in a future period, and providing a reference for a control process in an actual operation process based on a prediction result, thereby realizing control coordination of the unit under a deep peak-shaving working condition. According to the technical scheme, the deep peak shaving thermal power unit system model based on data driving and machine learning is obtained by combining five subsystem modules, namely a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, so that modeling of a deep peak shaving thermal power unit control system is realized.
See the description of the detailed description section.
Drawings
FIG. 1 is a data flow diagram of data modeling;
FIG. 2 is a block diagram of an LSTM neuron;
FIG. 3 is a block diagram of GRU neurons;
fig. 4 is a graph comparing predicted values with actual values.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. The following description of at least one exemplary embodiment is merely exemplary in nature and is in no way intended to limit the application, its application, or uses. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application, but the present application may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present application is not limited to the specific embodiments disclosed below.
Example 1:
as shown in fig. 1, the application discloses a modeling method of a deep peak-shaving thermal power unit control system, which comprises the steps of modeling, wherein the modeling step comprises the steps of obtaining historical operation data from a plant-level monitoring information system SIS of the unit, constructing and obtaining five subsystem modules of a boiler combustion system, a turbine speed regulation system, a reheat system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after regulating a stage in a future period, and providing a reference for a control process in an actual operation process based on a prediction result, thereby realizing control coordination of the unit under a deep peak-shaving working condition.
The modeling step specific division includes the steps of,
s1, extracting corresponding historical operation data from a plant-level monitoring information system SIS of a unit, wherein the historical operation data comprise historical operation data of main steam pressure and main steam temperature and deep peak regulation working conditions;
s2, sorting the historical operation data, and removing missing values; carrying out standardization processing on the data set, adjusting an input data structure in the historical data into three dimensions, and adjusting an output data structure into two dimensions;
s3, respectively constructing and obtaining five subsystem modules of a boiler combustion system, a turbine speed regulation system, a reheating system, a turbine system and a unit cold end system, respectively inputting historical operation data for training, and moving target data on time sequence according to the prediction duration and the prediction target;
s4, combining five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system to obtain a deep peak shaving thermal power unit system model based on data driving and machine learning, and realizing short-term prediction of pressure and load data after adjusting a stage in a future period, wherein the prediction result is used as a basis to realize control coordination work of the unit under a deep peak shaving working condition;
S5, after the unit generates new operation data, the generated real-time data is used as input quantity to be input into a deep peak shaving thermal power unit system model, and historical data with the same length are deleted on the basis, so that the real-time updating of the input quantity by the system is realized.
The method comprises the steps of establishing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, respectively comparing error magnitudes of a test set and a predicted value of three algorithms of a common RNN, an LSTM long-term artificial neural network and a GRU gate control unit based on a cyclic neural network RNN algorithm in machine learning, and selecting the algorithm with the smallest error and the best predicted effect as a data modeling mode.
The method comprises the steps of constructing a boiler combustion system prediction module based on a GRU (grid-control unit), taking the coal amount, a specific valve regulating instruction, water supply amount and air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a boiler combustion process, and advancing a target amount of main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is a period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The method comprises the steps of constructing a steam turbine speed regulating system prediction module based on an LSTM long-term artificial neural network, taking a specific valve regulating instruction, main steam pressure and main steam temperature in the operation history data of the deep peak regulating thermal power unit as input quantity, taking the pressure after regulating stage in the operation history data of the deep peak regulating thermal power unit as target quantity, and constructing the input quantity and output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a turbine speed regulating system, and advancing the pressure of a target quantity regulation stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Constructing a reheating system prediction module based on a GRU gate control unit, taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak shaving thermal power unit as input amounts, taking the reheating steam pressure and the reheating steam temperature in the operation history data of the deep peak shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The method comprises the steps of constructing a steam turbine system prediction module based on a GRU gate control unit, taking the pressure after regulation, the reheat steam pressure and the reheat steam temperature in the operation history data of the deep peak shaving thermal power unit as input quantities, taking the unit load in the operation history data of the deep peak shaving thermal power unit as a target quantity, and constructing the input quantities and the output quantities into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target unit load by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The method comprises the steps of constructing a unit cold end prediction module based on an LSTM long-term artificial neural network, taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate water temperature, ambient temperature and fan air quantity in the operation history data of the deep peak shaving thermal power unit as input quantities, taking the cold end heat exchange coefficient K in the operation history data of the deep peak shaving thermal power unit as a target quantity, and constructing the input quantities and the output quantities into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Example 2:
the invention discloses a device for modeling a deep peak-shaving thermal power unit control system, which comprises a modeling module, a boiler combustion system prediction module, a turbine speed regulation system prediction module, a reheating system prediction module, a turbine system prediction module and a unit cold end prediction module.
The modeling module is used for obtaining historical operation data from a plant-level monitoring information system SIS of the unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after regulating the level in a future period, taking the prediction result as a basis, providing a reference for a control process in an actual operation process, and further realizing control coordination of the unit under a deep peak-shaving working condition.
The boiler combustion system prediction module is used for taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a boiler combustion process, and advancing a target amount of main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is a period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The turbine speed regulating system prediction module is used for taking a specific valve regulating instruction, main steam pressure and main steam temperature in the operation history data of the deep peak regulating thermal power unit as input quantity, taking the regulated pressure in the operation history data of the deep peak regulating thermal power unit as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a turbine speed regulating system, and advancing the pressure of a target quantity regulation stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The reheating system prediction module is used for taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the reheating steam pressure and the reheating steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The steam turbine system prediction module is used for taking the pressure after the regulation stage, the reheat steam pressure and the reheat steam temperature in the deep peak regulation thermal power unit operation history data as input quantity, taking the unit load in the deep peak regulation thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target unit load by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
The unit cold end prediction module is used for taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate temperature, ambient temperature and fan air quantity in the deep peak shaving thermal power unit operation history data as input quantity, taking the cold end heat exchange coefficient K in the deep peak shaving thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Example 3:
the application discloses a device for modeling a deep peak-shaving thermal power unit control system, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the memory and the processor form an electronic terminal, and the processor realizes the steps of the embodiment 1 when executing the computer program.
Example 4:
the present application discloses a computer-readable storage medium storing a computer program which, when executed by a processor, implements the steps of embodiment 1.
Compared with the above embodiment, the program modules may be hardware modules made by using the existing logic operation technology, so as to implement the corresponding logic operation steps, communication steps and control steps, and further implement the corresponding steps, where the logic operation unit is not described in detail in the prior art.
The technical contribution of the application is as follows:
1. the modeling method of the deep peak shaving thermal power unit control system based on data driving and machine learning comprises the following steps:
s1, corresponding historical operation data, such as main steam pressure, main steam temperature and other historical operation data, are extracted from a plant-level monitoring information system SIS of the unit, wherein the historical operation data comprise typical working conditions such as deep peak shaving and the like.
S2, the historical operation data are arranged, and missing values are removed. And carrying out standardization processing on the data set, adjusting the input data structure in the historical data into three dimensions, and adjusting the output data structure into two dimensions.
S3, respectively constructing five subsystem modules of a boiler combustion system, a turbine speed regulation system, a reheating system, a turbine system and a unit cold end system, respectively inputting historical operation data for specific lower functions of the five subsystem modules for training, moving target data on time sequence according to the prediction duration and the prediction target, and constructing the model.
S4, combining the boiler combustion system, the turbine speed regulating system, the reheating system, the turbine system and the unit cold end system to obtain a deep peak shaving thermal power unit system model based on data driving and machine learning, and realizing short-term prediction of pressure, load and other data after adjusting the level in a future period, wherein the prediction result is based on the realization of control coordination work of the unit under the deep peak shaving working condition.
S5, after the unit generates new operation data, the generated real-time data is used as input quantity to be input into a unit system data model, and historical data with the same length are deleted on the basis, so that the real-time updating function of the system on the input quantity is realized.
2. The plant-level monitoring information system of the unit can update and record operation data in real time along with the operation of the unit so as to ensure timeliness and accuracy of the predicted data.
3. The data needs to be standardized, so that the data accords with normal distribution, and the influence of data dimension on modeling calculation of the data is eliminated.
4. The building of five subsystems of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system is based on a cyclic neural network RNN algorithm in machine learning, error magnitudes of a test set and a predicted value of three algorithms of a common RNN, an LSTM long-short-term artificial neural network and a GRU gating unit are respectively compared, and specific error analysis indexes can be selected as Root Mean Square Error (RMSE), mean Absolute Percentage Error (MAPE) and fitting goodness (R) of a predicted data set and an actual data set 2 The specific calculation formula of the weighted average error evaluation index is shown as follows.
K=α*RMSE+β*MAPE+γ*R 2 (4)
In the formula (1), true is the actual value in the test set, pred predicted value.
In equation (3), true_mean is the average of the actual values in the test set.
In the formula (4), α, β, γ are weighted parameters, and K is a weighted average error evaluation index. And selecting an algorithm with the smallest weighted average error, namely the best prediction effect, as a data modeling mode.
5. A boiler combustion system prediction module based on the GRU gating unit is constructed.
The method comprises the steps of taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak regulating thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak regulating thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence. The method is characterized by comprising the steps of realizing a data modeling process of a boiler combustion process, and advancing target quantity main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted.
Different model parameters are set, the model is iterated for a plurality of times on the premise of not changing the data of the training set and the test set, and the optimal value of the model parameters is determined.
6. And constructing a steam turbine speed regulating system prediction module based on the LSTM long-term artificial neural network.
The specific valve regulating instruction, the main steam pressure and the main steam temperature in the operation history data of the deep peak regulating thermal power unit are taken as input quantities, the pressure after the regulating stage in the operation history data of the deep peak regulating thermal power unit is taken as a target quantity, and the input quantities and the output quantities are constructed into two-dimensional time sequence data with the same time sequence. The method is characterized by comprising the steps of realizing the data modeling process of the turbine speed regulating system, and advancing the pressure of the target quantity after the regulating stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted.
Different model parameters are set, the model is iterated for a plurality of times on the premise of not changing the data of the training set and the test set, and the optimal value of the model parameters is determined.
7. And constructing a reheating system prediction module based on the GRU gating unit.
Taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak regulating thermal power unit as input amounts, taking the reheat steam pressure and the reheat steam temperature in the operation history data of the deep peak regulating thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence. The method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted.
Different model parameters are set, the model is iterated for a plurality of times on the premise of not changing the data of the training set and the test set, and the optimal value of the model parameters is determined.
8. A turbine system prediction module based on the GRU gating unit is constructed.
Taking the pressure after the regulation stage, the reheat steam pressure and the reheat steam temperature in the deep peak regulation thermal power unit operation history data as input quantity, taking the unit load in the deep peak regulation thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence. The method is characterized by comprising the steps of realizing the data modeling process of the steam turbine system, and advancing the load of the target unit by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted.
Different model parameters are set, the model is iterated for a plurality of times on the premise of not changing the data of the training set and the test set, and the optimal value of the model parameters is determined.
9. And constructing a unit cold end prediction module based on the LSTM long-term artificial neural network.
Taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate water temperature, ambient temperature, fan air quantity and the like in the operation history data of the deep peak shaving thermal power generating unit as input quantity, taking the cold end heat exchange coefficient K in the operation history data of the deep peak shaving thermal power generating unit as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence. The method comprises the steps of realizing the data modeling process of the steam turbine system, and advancing the target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted.
Different model parameters are set, the model is iterated for a plurality of times on the premise of not changing the data of the training set and the test set, and the optimal value of the model parameters is determined.
10. The input data structure is (Z, T, Y), the output data structure is (Z, N), Z is the number of data groups corresponding to the input data and the output data, the data group is determined by the dimension of the historical data, T is the time dimension to be predicted, Y is the characteristic dimension of the input data, and N is the characteristic dimension of the output data. The data model algorithm adopts LSTM and GRU in a cyclic neural network algorithm, wherein the neural network calculation process of the LSTM comprises three parts of an input gate, a forgetting gate and an output gate.
f t =σ(W f ·[h t-1 ,x t ]+b f ) (5)
i t =σ(W i ·[h t-1 ,x t ]+b i ) (6)
o t =σ(W o ·[h t-1 ,x t ]+b o ) (9)
h t =o t ·tanh(C t ) (10)
Wherein equation (5) is the calculation of the forgetting gate, sigma is the activation function, and the output is controlled to be [0,1 ]]Between 0 means all drops and 1 means all reservations. X is x t For input quantity, h t-1 For the output of the last neuron, W is input as input at this layer f ,W i B f ,b i ,b c ,b o For the parameter matrix, the hidden layer computes and discards the data by these parameters. Formulas (6), (7), (8) are calculations of the input gates, where i t C is the output of the input gate t To characterize the vector of cell states. the tanh layer creates a new vectorFor updating the current cell state. Updating old cell status in formula (8), C t-1 Updated to C t . Associating old state with f t Multiplication forgets a part of information. Then add +.>This is the new candidate created in the previous step, from which the degree of update for each state is determined. Equation (9), equation (10) is the output gate calculation, o t To output the result of the gate, the cell state is again calculatedCalculating through a tanh layer to obtain [ -1,1]And outputs the value of (2) to the gate result o t Multiplying and finally outputting a result h t . Through multiple times of hidden layer calculation, the reservation and forgetting of the past operation result are realized, and the continuous data are learned and predicted.
The GRU is a variant of the LSTM network, is simpler than the LSTM network in structure, and only comprises two gate structures of an update gate and a reset gate, and the calculation process is as follows:
z t =σ(W z ·[h t-1 ,x t ]) (11)
r t =σ(W r ·[h t-1 ,x t ]) (12)
wherein equation (11) is the calculation of the update gate, equation (12) is the calculation of the reset gate, W z ,W r W is a parameter matrix, z t ,r t The calculation results of the update gate and the reset gate, h t-1 For the output of the last neuron,to the vector calculated by the tanh layer, finally the vector is calculated by the tanh layer and the vector is calculated by the tanh layer t-1 Calculating the output quantity h of the neuron t . The results of the model under historical operating data training are compared to determine which algorithm each module uses.
11. And according to valve regulating instructions and other operation data, completing data modeling of five subsystems of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, and completing a deep peak regulating thermal power unit control system modeling process after combining.
The technical scheme is as follows:
the invention provides a data modeling method of a deep peak shaving thermal power unit control system based on data driving and machine learning.
In order to achieve the above, the present invention proposes the following scheme:
corresponding historical operation data, such as the main steam pressure, the main steam temperature and the like under typical working conditions of deep peak shaving and the like, are extracted from a plant-level monitoring information system SIS of the unit. The plant-level monitoring information system of the unit can update and record operation data in real time along with the operation of the unit so as to ensure timeliness and accuracy of the predicted data.
And (5) sorting the historical operation data to remove the missing value. And carrying out standardization processing on the data set so that the data accords with normal distribution, eliminating the influence of data dimension on modeling calculation of the data, adjusting an input data structure in the historical data into three dimensions, and adjusting an output data structure into two dimensions.
Five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system are respectively constructed, historical operation data are respectively input for specific functions of the boiler combustion system, the turbine speed regulating system, the reheating system, the turbine system and the unit cold end system for training, target data are moved in time sequence according to prediction duration and a prediction target, and a model is built by using a proper cyclic neural network algorithm.
Specifically, compared with three algorithms of common RNN, LSTM, GRU, the prediction module of the boiler combustion system based on the GRU gating unit is selected and established. The method comprises the steps of taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak regulating thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak regulating thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence. The method is characterized by comprising the steps of realizing a data modeling process of a boiler combustion process, and advancing target quantity main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted. And setting different model parameters, and carrying out repeated iteration on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Specifically, compared with three algorithms of common RNN, LSTM, GRU, a steam turbine speed regulation system prediction module based on an LSTM long-short-term artificial neural network is selected and established. The specific valve regulating instruction, the main steam pressure and the main steam temperature in the operation history data of the deep peak regulating thermal power unit are taken as input quantities, the pressure after the regulating stage in the operation history data of the deep peak regulating thermal power unit is taken as a target quantity, and the input quantities and the output quantities are constructed into two-dimensional time sequence data with the same time sequence. The method is characterized by comprising the steps of realizing the data modeling process of the turbine speed regulating system, and advancing the pressure of the target quantity after the regulating stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted. And setting different model parameters, and carrying out repeated iteration on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Specifically, compared with three algorithms of common RNN, LSTM, GRU, a reheat system prediction module based on the GRU gating unit is selected and established. Taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak regulating thermal power unit as input amounts, taking the reheat steam pressure and the reheat steam temperature in the operation history data of the deep peak regulating thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence. The method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted. And setting different model parameters, and carrying out repeated iteration on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Specifically, compared with three algorithms of common RNN, LSTM, GRU, a turbine system prediction module based on the GRU gating unit is selected and established. Taking the pressure after the regulation stage, the reheat steam pressure and the reheat steam temperature in the deep peak regulation thermal power unit operation history data as input quantity, taking the unit load in the deep peak regulation thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence. The method is characterized by comprising the steps of realizing the data modeling process of the steam turbine system, and advancing the load of the target unit by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted. And setting different model parameters, and carrying out repeated iteration on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Specifically, compared with three algorithms of common RNN, LSTM, GRU, the method selects and establishes a unit cold end prediction module based on an LSTM long-short-term artificial neural network. Taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate water temperature, ambient temperature, fan air quantity and the like in the operation history data of the deep peak shaving thermal power generating unit as input quantity, taking the cold end heat exchange coefficient K in the operation history data of the deep peak shaving thermal power generating unit as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence. The method comprises the steps of realizing the data modeling process of the steam turbine system, and advancing the target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted. And setting different model parameters, and carrying out repeated iteration on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
Specifically, the data model of the deep peak shaving thermal power generating unit control system based on data driving and machine learning is built with an input data structure of (Z, T, Y), an output data structure of (Z, N), Z is the number of data sets corresponding to input and output data, the number is determined by the dimension of historical data, T is the time dimension to be predicted, Y is the characteristic dimension of the input data, and N is the characteristic dimension of the output data.
Specifically, the data model algorithm adopts LSTM and GRU in a cyclic neural network algorithm, wherein the neural network operation process of the LSTM comprises three parts of an input gate, a forgetting gate and an output gate as shown below.
f t =σ(W f ·[h t-1 ,x t ]+b f ) (5)
i t =σ(W i ·[h t-1 ,x t ]+b i ) (6)
o t =σ(W o ·[h t-1 ,x t ]+b o ) (9)
h t =o t ·tanh(C t ) (10)
Specifically, equation (5) is a calculation of forgetting gate, σ is an activation function, and the output is controlled to be [0,1]Between 0 means all drops and 1 means all reservations. X is x t For input quantity, h t-1 For the output of the last neuron, W is input as input at this layer f ,W i B f ,b i ,b c ,b o For the parameter matrix, the hidden layer computes and discards the data by these parameters. Formulas (6), (7), (8) are calculations of the input gates, where i t C is the output of the input gate t To characterize the vector of cell states. the tanh layer creates a new vectorFor updating the current cell state. Update old cell state, mix C t-1 Updated to C t . Associating old state with f t Multiplication forgets a part of information. Then add +.>This is the new candidate created in the previous step, from which the degree of update for each state is determined. Equation (9), equation (10) is the output gate calculation, o t To output the output result of the gate, the cell state is calculated by the tanh layer to obtain [ -1,1]And associate it with the inputResults of going out o t Multiplying and finally outputting a result h t . Through multiple times of hidden layer calculation, the reservation and forgetting of the past operation result are realized, and the continuous data are learned and predicted.
Specifically, the GRU is a variant of the LSTM network, which is simpler than the LSTM network in structure, and only comprises two gate structures of an update gate and a reset gate, and the operation process is as follows:
z t =σ(W z ·[h t-1 ,x t ]) (11)
r t =σ(W r ·[h t-1 ,x t ]) (12)
wherein equation (11) is the calculation of the update gate, equation (12) is the calculation of the reset gate, W z ,W r W is a parameter matrix, z t ,r t The calculation results of the update gate and the reset gate, h t-1 For the output of the last neuron,to the vector calculated by the tanh layer, finally the vector is calculated by the tanh layer and the vector is calculated by the tanh layer t-1 Calculating the output quantity h of the neuron t . The results of the model under historical operating data training are compared to determine which algorithm each module uses.
Specifically, five subsystems of a boiler combustion system, a turbine speed regulation system, a reheating system, a turbine system and a unit cold end system are combined to obtain a deep peak shaving thermal power unit system model based on data driving and machine learning, short-term prediction of pressure, load and other data after regulation in a future period is realized, and the prediction result is used for realizing control coordination work of the unit under the deep peak shaving working condition.
Specifically, after the unit generates new operation data, the generated real-time data is input into a unit system data model as input quantity, and historical data with the same length is deleted on the basis, so that the real-time updating function of the system on the input quantity is realized.
Compared with the prior art, the method provided by the invention utilizes the algorithm which is excellent in data prediction and is a cyclic neural network, utilizes a data driving mode to discover the historical operation data of a massive thermal power unit which is rarely utilized in the past, discovers the internal connection between a valve regulating instruction and each key parameter of unit operation, realizes the prediction of the key parameters, and simulates and guides the control process of the thermal power unit based on the prediction.
Specifically, compared with the traditional data modeling method using only one machine learning mode, the method simplifies the operation process by dividing different unit modules, compares the performance of various circulating neural networks in different modules, and selects the optimal mode to finish the data modeling process. And simultaneously, the parameter selection of the data model is determined in a certain range in an iterative comparison error mode, so that the data model of the finally combined deep peak shaving thermal power unit control system is ensured to have minimized error to a certain extent.
Specifically, the data can be connected with a DCS system of the thermal power generating unit in real time, the latest unit operation data can be extracted in real time, and the data updating process is completed. And the prediction of the key control data with timeliness is completed by means of the updated real-time data, and the control process in the actual engineering is optimized according to the prediction. The data modeling is more time-efficient than the past control models, and therefore superior in accuracy.
As shown in fig. 1, a data modeling flow chart.
As shown in fig. 2, LSTM neuron structure diagram.
As shown in fig. 3, the gre neuron structure diagram.
As shown in fig. 4, a graph of predicted values versus actual values is shown, taking the main vapor pressure as an example, with the horizontal axis representing the data sequence and the vertical axis representing the main vapor pressure.
The modeling method of the deep peak shaving thermal power generating unit control system based on data driving and machine learning comprises the following steps:
step one, corresponding historical operation data, such as the historical operation data of main steam pressure, main steam temperature and the like, are extracted from a plant-level monitoring information system SIS of the unit, wherein the historical operation data comprise typical working conditions such as deep peak shaving and the like.
And step two, the historical operation data are arranged, and the missing value is removed. And carrying out standardization processing on the data set, adjusting the input data structure in the historical data into three dimensions, and adjusting the output data structure into two dimensions. The input data structure is (Z, T, Y), the output data structure is (Z, N), Z is the number of data groups corresponding to the input data and the output data, the number is determined by the dimension of the historical data, T is the time dimension to be predicted, Y is the characteristic dimension of the input data, and N is the characteristic dimension of the output data.
The data normalization processing formula is as follows:
in the formula (15), X norm Is normalized data, X, X mean 、X std Raw data values, mean and standard deviation, respectively.
And thirdly, respectively constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, respectively inputting historical operation data for specific lower functions of the five subsystem modules for training, moving target data on time sequence according to the prediction duration and the prediction target, and constructing the model.
The movement rule of the data is as follows: outputting the data structure up-shift prediction timing sequence value:
u is input data, P is output data, t is a time sequence value to be predicted
And step four, combining five subsystems of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system to obtain the deep peak shaving thermal power unit system model based on data driving and machine learning.
Specifically, the optimal algorithm is determined for different subsystems by comparing different cyclic neural network algorithms, and the optimal model parameters are determined in an iterative mode, so that the goodness of fit R of each algorithm in different subsystems is determined 2 As shown in table 1.
Table 1: goodness of fit table
The model parameters were iteratively selected as shown in table 2.
Table 2: model parameter iteration selection table
Each subsystem needs to iterate 3 x 2 = 54 times, a combination of parameters of the mean absolute percentage error map is selected as the final model parameter.
The parameters of each subsystem were selected as shown in table 3:
table 3: parameter selection table for each subsystem
And fifthly, after the unit generates new operation data, the generated real-time data is used as input quantity to be input into a unit system data model, and historical data with the same length is deleted on the basis, so that the real-time updating function of the system on the input quantity is realized.
And step six, inputting the operation data updated in real time and part of key control parameters into a coupling model to realize short-term prediction of the operation data such as pressure, load and the like after the regulation stage in a future period of time and the key control parameters, wherein the prediction result is based on the realization of control coordination work of the unit under the variable working condition or even the deep peak regulation working condition of the unit.
After the application runs for a period of time internally, the feedback of field technicians is beneficial in that:
The invention provides a method for carrying out data modeling on a thermal power unit control system under a deep peak regulation working condition by using a data driving and machine learning tool, which extracts corresponding historical operation data from a plant-level monitoring information system SIS of a unit and constructs five subsystem modules of a boiler combustion system, a turbine speed regulation system, a reheating system, a turbine system and a unit cold end system. Combining five subsystems to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, realizing short-term prediction of pressure, load and other data after adjusting the level in a future period, taking the prediction result as a basis, providing a reference for a control process in an actual operation process, and realizing control coordination work of the unit under a deep peak-shaving working condition.
At present, the technical scheme of the invention has been subjected to pilot-scale test, namely, smaller-scale test of products before large-scale mass production; after the pilot test is completed, the use investigation of the user is performed in a small range, and the investigation result shows that the user satisfaction is higher; now, the preparation of the formal production of the product for industrialization (including intellectual property risk early warning investigation) is started.
Claims (10)
1. A modeling method of a deep peak-shaving thermal power generating unit control system is characterized by comprising the following steps of: the modeling step comprises the steps of obtaining historical operation data from a plant-level monitoring information system SIS of a unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after regulating the level in a future period, wherein a prediction result is used as a basis, a reference is provided for a control process in an actual operation process, and further control coordination of the unit under a deep peak-shaving working condition is realized.
2. The method for modeling a deep peak shaver thermal power plant control system according to claim 1, wherein: the modeling step specific division includes the steps of,
s1, extracting corresponding historical operation data from a plant-level monitoring information system SIS of a unit, wherein the historical operation data comprise historical operation data of main steam pressure and main steam temperature and deep peak regulation working conditions;
s2, sorting the historical operation data, and removing missing values; carrying out standardization processing on the data set, adjusting an input data structure in the historical data into three dimensions, and adjusting an output data structure into two dimensions;
s3, respectively constructing and obtaining five subsystem modules of a boiler combustion system, a turbine speed regulation system, a reheating system, a turbine system and a unit cold end system, respectively inputting historical operation data for training, and moving target data on time sequence according to the prediction duration and the prediction target;
s4, combining five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system to obtain a deep peak shaving thermal power unit system model based on data driving and machine learning, and realizing short-term prediction of pressure and load data after adjusting a stage in a future period, wherein the prediction result is used as a basis to realize control coordination work of the unit under a deep peak shaving working condition;
S5, after the unit generates new operation data, the generated real-time data is used as input quantity to be input into a deep peak shaving thermal power unit system model, and historical data with the same length are deleted on the basis, so that the real-time updating of the input quantity by the system is realized.
3. The method for modeling a deep peak shaver thermal power plant control system according to claim 1, wherein: the method is characterized in that the building of five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system is based on a cyclic neural network (RNN) algorithm in machine learning, error magnitudes of a test set and a predicted value of three algorithms of a common RNN, an LSTM long-term artificial neural network and a GRU gate control unit are respectively compared, and an algorithm with the smallest error and the best predicted effect is selected as a data modeling mode.
4. A method of modeling a deep peak shaver thermal power plant control system according to claim 3, wherein: the method comprises the steps of constructing a boiler combustion system prediction module based on a GRU (grid-control unit), taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a boiler combustion process, and advancing a target amount of main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is a period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
The method comprises the steps of constructing a steam turbine speed regulating system prediction module based on an LSTM long-term artificial neural network, taking a specific valve regulating instruction, main steam pressure and main steam temperature in the operation history data of the deep peak regulating thermal power unit as input quantity, taking the pressure after regulating stage in the operation history data of the deep peak regulating thermal power unit as target quantity, and constructing the input quantity and output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a turbine speed regulating system, and advancing the pressure of a target quantity regulation stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
constructing a reheating system prediction module based on a GRU gate control unit, taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak shaving thermal power unit as input amounts, taking the reheating steam pressure and the reheating steam temperature in the operation history data of the deep peak shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
5. A method of modeling a deep peak shaver thermal power plant control system according to claim 3, wherein: the method comprises the steps of constructing a steam turbine system prediction module based on a GRU gate control unit, taking the pressure after regulation, the reheat steam pressure and the reheat steam temperature in the operation history data of the deep peak shaving thermal power unit as input quantities, taking the unit load in the operation history data of the deep peak shaving thermal power unit as a target quantity, and constructing the input quantities and the output quantities into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target unit load by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
the method comprises the steps of constructing a unit cold end prediction module based on an LSTM long-term artificial neural network, taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate water temperature, ambient temperature and fan air quantity in the operation history data of the deep peak shaving thermal power unit as input quantities, taking the cold end heat exchange coefficient K in the operation history data of the deep peak shaving thermal power unit as a target quantity, and constructing the input quantities and the output quantities into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
6. The device for modeling the deep peak-shaving thermal power unit control system is characterized in that: the modeling module is used for obtaining historical operation data from a plant-level monitoring information system SIS of the unit, constructing five subsystem modules of a boiler combustion system, a turbine speed regulating system, a reheating system, a turbine system and a unit cold end system, combining the five subsystem modules to obtain a deep peak-shaving thermal power unit system model based on data driving and machine learning, and performing short-term prediction on pressure and load data after regulating the level in a future period, providing a reference for a control process in an actual operation process based on a prediction result, and further achieving control coordination of the unit under a deep peak-shaving working condition.
7. The device for modeling a deep peak shaver thermal power plant control system according to claim 6, wherein: the system also comprises a boiler combustion system prediction module, a steam turbine speed regulation system prediction module and a reheating system prediction module,
the boiler combustion system prediction module is used for taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the main steam pressure and the main steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a boiler combustion process, and advancing a target amount of main steam pressure and main steam temperature by a period of time in time sequence, wherein the advanced period of time is a period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
The turbine speed regulating system prediction module is used for taking a specific valve regulating instruction, main steam pressure and main steam temperature in the operation history data of the deep peak regulating thermal power unit as input quantity, taking the regulated pressure in the operation history data of the deep peak regulating thermal power unit as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a turbine speed regulating system, and advancing the pressure of a target quantity regulation stage by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
the reheating system prediction module is used for taking the coal amount, a specific valve regulating instruction, the water supply amount and the air supply amount in the operation history data of the deep peak-shaving thermal power unit as input amounts, taking the reheating steam pressure and the reheating steam temperature in the operation history data of the deep peak-shaving thermal power unit as target amounts, and constructing the input amounts and the output amounts into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a reheating system, and advancing target quantity reheat steam pressure and reheat steam temperature by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
8. The device for modeling a deep peak shaver thermal power plant control system according to claim 6, wherein: the system also comprises a steam turbine system prediction module and a unit cold end prediction module,
the steam turbine system prediction module is used for taking the pressure after the regulation stage, the reheat steam pressure and the reheat steam temperature in the deep peak regulation thermal power unit operation history data as input quantity, taking the unit load in the deep peak regulation thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target unit load by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out repeated iteration on the model on the premise of not changing data of a training set and a testing set to determine the optimal value of the model parameters;
the unit cold end prediction module is used for taking the exhaust steam temperature, condenser vacuum, condensate flow, condensate temperature, ambient temperature and fan air quantity in the deep peak shaving thermal power unit operation history data as input quantity, taking the cold end heat exchange coefficient K in the deep peak shaving thermal power unit operation history data as target quantity, and constructing the input quantity and the output quantity into two-dimensional time sequence data with the same time sequence; the method comprises the steps of realizing a data modeling process of a steam turbine system, and advancing a target cold end heat exchange coefficient K by a period of time in time sequence, wherein the advanced period of time is the period of time to be predicted; setting model parameters, and carrying out multiple iterations on the model on the premise of not changing the data of the training set and the test set to determine the optimal value of the model parameters.
9. An apparatus for modeling a deep-peak-shaving thermal power generating unit control system, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, characterized in that: the processor, when executing a computer program, carries out the respective steps of the method of any one of claims 1 to 5.
10. An apparatus for modeling a deep pitch thermal power generating unit control system, comprising a computer readable storage medium storing a computer program, characterized in that: which computer program, when being executed by a processor, carries out the respective steps of the method of any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310989293.XA CN117031950A (en) | 2023-08-08 | 2023-08-08 | Modeling method and device for deep peak-shaving thermal power generating unit control system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310989293.XA CN117031950A (en) | 2023-08-08 | 2023-08-08 | Modeling method and device for deep peak-shaving thermal power generating unit control system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117031950A true CN117031950A (en) | 2023-11-10 |
Family
ID=88636645
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310989293.XA Pending CN117031950A (en) | 2023-08-08 | 2023-08-08 | Modeling method and device for deep peak-shaving thermal power generating unit control system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117031950A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117387056A (en) * | 2023-12-13 | 2024-01-12 | 华能济南黄台发电有限公司 | Thermal power plant depth peak regulation state monitoring method and system |
CN117869930A (en) * | 2024-01-31 | 2024-04-12 | 中国电力工程顾问集团有限公司 | Multi-variable control method and device for stable combustion of coal-fired boiler in wide load range |
CN117970986A (en) * | 2024-04-01 | 2024-05-03 | 广东热矩智能科技有限公司 | Temperature and humidity control method, device and medium of cold and hot system |
CN117991639A (en) * | 2024-01-31 | 2024-05-07 | 中国电力工程顾问集团有限公司 | Multi-target combustion optimization control method and device for coal-fired power plant based on machine learning |
-
2023
- 2023-08-08 CN CN202310989293.XA patent/CN117031950A/en active Pending
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117387056A (en) * | 2023-12-13 | 2024-01-12 | 华能济南黄台发电有限公司 | Thermal power plant depth peak regulation state monitoring method and system |
CN117387056B (en) * | 2023-12-13 | 2024-03-08 | 华能济南黄台发电有限公司 | Thermal power plant depth peak regulation state monitoring method and system |
CN117869930A (en) * | 2024-01-31 | 2024-04-12 | 中国电力工程顾问集团有限公司 | Multi-variable control method and device for stable combustion of coal-fired boiler in wide load range |
CN117991639A (en) * | 2024-01-31 | 2024-05-07 | 中国电力工程顾问集团有限公司 | Multi-target combustion optimization control method and device for coal-fired power plant based on machine learning |
CN117869930B (en) * | 2024-01-31 | 2024-06-07 | 中国电力工程顾问集团有限公司 | Multi-variable control method and device for stable combustion of coal-fired boiler in wide load range |
CN117970986A (en) * | 2024-04-01 | 2024-05-03 | 广东热矩智能科技有限公司 | Temperature and humidity control method, device and medium of cold and hot system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN117031950A (en) | Modeling method and device for deep peak-shaving thermal power generating unit control system | |
CN106920006B (en) | Subway station air conditioning system energy consumption prediction method based on ISOA-LSSVM | |
CN105955026B (en) | Fuzzy PID control method, device and system | |
CN102129259A (en) | Neural network proportion integration (PI)-based intelligent temperature control system and method for sand dust environment test wind tunnel | |
CN105634018B (en) | A kind of Load Flow Solution method of random optimum containing wind-powered electricity generation based on stochastic response surface and interior point method | |
CN104036328A (en) | Self-adaptive wind power prediction system and prediction method | |
CN115764870A (en) | Multivariable photovoltaic power generation power prediction method and device based on automatic machine learning | |
CN116070769A (en) | Ultra-short-term wind power plant power multi-step interval prediction modularization method and device thereof | |
CN111931436A (en) | Burner nozzle air quantity prediction method based on numerical simulation and neural network | |
CN115795992A (en) | Park energy Internet online scheduling method based on virtual deduction of operation situation | |
Wang et al. | Multitarget stochastic configuration network and applications | |
CN113759709A (en) | Method and device for training strategy model, electronic equipment and storage medium | |
CN113222263A (en) | Photovoltaic power generation power prediction method based on long-term and short-term memory neural network | |
CN111413864A (en) | 600MW supercritical thermal power generating unit modeling and control method | |
Kaipbergenov et al. | Modeling of the Technological Process of Ammonization of Soda Production Under Conditions of Changing Initial Data | |
CN115018156A (en) | Short-term wind power prediction method | |
Chen et al. | Variation-cognizant probabilistic power flow analysis via multi-task learning | |
Wu et al. | Neural Network Based Fea sible Region Approximation Model for Optimal Operation of Integrated Electricity and Heating System | |
CN111897203A (en) | Heater economic regulation method based on data driving and mechanism modeling state monitoring | |
Yan et al. | Multi-state probability prediction of wind power based on deep learning model | |
Li et al. | A wind power prediction model based on optimized N-BEATS network with multivariate inputs | |
Guo et al. | Neural Network Modeling and Predictive Control of Low Oxygen Combustion System in Petrochemical Heating Furnace | |
CN115660324B (en) | Power grid multi-section out-of-limit regulation and control method and system based on graph reinforcement learning | |
CN118246352B (en) | Machine learning calculation method, product and equipment for thermal resistance of dirt of seawater heat exchanger | |
CN114723128A (en) | Static voltage stability margin online prediction method based on convolutional neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |