CN113611375A - Data determination method, device, equipment and storage medium in thermal power plant system - Google Patents

Data determination method, device, equipment and storage medium in thermal power plant system Download PDF

Info

Publication number
CN113611375A
CN113611375A CN202110907440.5A CN202110907440A CN113611375A CN 113611375 A CN113611375 A CN 113611375A CN 202110907440 A CN202110907440 A CN 202110907440A CN 113611375 A CN113611375 A CN 113611375A
Authority
CN
China
Prior art keywords
data
group
nitrogen oxide
initial data
goodness
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.)
Granted
Application number
CN202110907440.5A
Other languages
Chinese (zh)
Other versions
CN113611375B (en
Inventor
廖强
刘正一
王向勇
陈俊
李辰
段斌
曾金龙
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chengdu Jiahua Chain Cloud Technology Co ltd
Original Assignee
Chengdu Jiahua Chain Cloud Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chengdu Jiahua Chain Cloud Technology Co ltd filed Critical Chengdu Jiahua Chain Cloud Technology Co ltd
Priority to CN202110907440.5A priority Critical patent/CN113611375B/en
Publication of CN113611375A publication Critical patent/CN113611375A/en
Application granted granted Critical
Publication of CN113611375B publication Critical patent/CN113611375B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Chemical & Material Sciences (AREA)
  • Medical Informatics (AREA)
  • Mathematical Physics (AREA)
  • Evolutionary Biology (AREA)
  • Feedback Control In General (AREA)
  • Treating Waste Gases (AREA)

Abstract

The application provides a data determination method, a data determination device, data determination equipment and a storage medium in a thermal power plant system, wherein the method comprises the following steps: inputting reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, wherein the nitrogen oxide emission data sets comprise a plurality of nitrogen oxide emission data; calculating a group goodness degree corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, wherein the group goodness degree is used for evaluating the goodness and the badness of the initial data group; and if the maximum group goodness of the plurality of group goodness of comments does not meet the target condition, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtaining a target data group until the maximum group goodness of comment meets the target condition.

Description

Data determination method, device, equipment and storage medium in thermal power plant system
Technical Field
The application relates to the technical field of thermal power plants, in particular to a data determination method, device, equipment and storage medium in a thermal power plant system.
Background
In the prior art, prediction of nitrogen oxide emission data of a thermal power plant is generally realized by establishing an ARIMA model, predicting the nitrogen oxide emission data through the established ARIMA model, and adjusting ammonia steam inlet flow introduced into a Selective Catalytic Reduction (SCR) reactor according to the predicted nitrogen oxide emission data at a chimney inlet after obtaining the nitrogen oxide emission data at the chimney inlet through the ARIMA model prediction. For example, when the predicted nitrogen oxide emission data of the chimney inlet is serious in environmental pollution, the flow of ammonia steam introduced into the SCR reactor is increased to reduce the nitrogen oxide emission data of the chimney inlet.
Although the emission data of the nitrogen oxides at the inlet of the chimney is predicted through an ARIMA model, when the emission data of the nitrogen oxides at the inlet of the chimney is too large or too small, manual adjustment is still needed, the efficiency is low, and the adjustment to a proper value is difficult.
Disclosure of Invention
Based on the above, a data determination method, device, equipment and storage medium in a thermal power plant system are provided to solve the problem of low manual regulation efficiency in the prior art.
In a first aspect, a method for determining data in a thermal power plant system is provided, including:
generating a plurality of initial data sets for an SCR reactor in a target system, the initial data sets comprising a plurality of initial data, the initial data comprising an ammonia vapor inlet flow rate, the target system comprising a thermal power plant system;
acquiring reactor data corresponding to the SCR reactor;
inputting reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, wherein the nitrogen oxide emission data sets comprise a plurality of nitrogen oxide emission data;
calculating a group goodness degree corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, wherein the group goodness degree is used for evaluating the goodness and the badness of the initial data group;
and if the maximum group goodness of the plurality of group goodness of comments does not meet the target condition, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtaining a target data group until the maximum group goodness of comment meets the target condition.
The data determining method in the thermal power plant system generates a plurality of initial data groups for the SCR reactor in the target system, inputs the initial data groups and reactor data corresponding to the SCR reactor into a nitrogen oxide prediction model, predicts nitrogen oxide discharge data through the nitrogen oxide prediction model, calculates group goodness according to the predicted nitrogen oxide discharge data, namely calculates goodness of the initial data groups, and obtains the target data group when the maximum group goodness meets the target condition. Therefore, through the mode, continuous adjustment is not needed, and only the optimal target data set is selected through a mode of calculating the group goodness under the condition that the reactor data is determined, so that the adjustment efficiency is improved compared with a manual adjustment mode, and in addition, the optimal target data set (ammonia steam inlet flow rate) can be quickly obtained through the mode.
In one embodiment, the calculating a group goodness score for each of the nox emission data sets based on a plurality of nox emission data sets for the target system includes: calculating a first-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and first reference nitrogen oxide emission data; calculating a second-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and second reference nitrogen oxide emission data; calculating the data consumption loss corresponding to the target nitrogen oxide emission data set according to the initial data set and the reference data set corresponding to the target nitrogen oxide emission data set; calculating the corresponding balance degree of the target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system; and obtaining the group goodness corresponding to the target nitrogen oxide emission data group according to the first-level emission limit loss, the second-level emission limit loss, the data loss and the balance degree corresponding to the target nitrogen oxide emission data group.
In the embodiment, the group goodness evaluation is generated by combining the four losses, so that the accuracy of the group goodness evaluation is improved.
In one embodiment, the generating a plurality of initial data sets for a next input to the nox prediction model based on the plurality of initial data sets comprises: determining the combination probability corresponding to the initial data group according to the group goodness of evaluation corresponding to the initial data group; and generating a plurality of initial data sets which are input into the nitrogen oxide prediction model next time according to the combined probability corresponding to each initial data set.
The above embodiments provide a method for generating a plurality of initial data sets, and particularly, a plurality of initial data sets to be input into the nox prediction model next time are generated according to the group goodness of evaluation, it can be understood that the initial data set with the high group goodness of evaluation will have a higher probability of being used for generating the initial data set to be input into the nox prediction model next time, thereby increasing the speed of obtaining the target data set.
In one embodiment, the determining the combined probability corresponding to the initial data group according to the size of the group goodness corresponding to the initial data group includes: adding the group goodness corresponding to the plurality of initial data groups to obtain a total goodness; and dividing the group goodness corresponding to the initial data group by the total goodness to obtain the combined probability corresponding to the initial data group.
The above embodiment provides a method for determining a combination probability, and the method can simply obtain the combination probability by solving a mode of calculating the ratio of the group goodness to the total goodness.
In one embodiment, the generating a plurality of initial data sets for the next input to the nox prediction model based on the combined probability corresponding to each of the initial data sets comprises: combining the plurality of initial data groups pairwise according to the combination probability corresponding to each initial data group to obtain a plurality of initial data group pairs; crossing two initial data sets in the plurality of initial data set pairs to obtain a plurality of crossed data set pairs; and (3) carrying out variation on two initial data sets in the crossed data set pair to obtain a plurality of initial data sets which are input into the nitrogen oxide prediction model next time.
The above-described embodiment generates a plurality of initial data sets for the next input nox prediction model by combining data pairs, cross data pairs, and variations.
In one embodiment, the reactor data includes operating condition data, time data, derivative data, and time delay data.
In the embodiment, not only a plurality of initial data sets but also a plurality of types of SCR reactor data are input in the process of obtaining the target data set, so that the finally obtained target data set is more accurate.
In one embodiment, a plurality of SCR reactors are arranged in the target system, and the nitrogen oxide prediction model comprises a prediction neural network and a sequence-to-sequence model corresponding to each SCR reactor.
In the embodiment, when a plurality of SCR reactors are provided in the target system, a sequence-to-sequence model is established for each SCR reactor to predict the outlet data (nitrogen oxide emission data) of each SCR reactor outlet, so as to improve the accuracy of the finally predicted nitrogen oxide emission data.
In a second aspect, a data determination apparatus in a thermal power plant system is provided, including:
a generating module configured to generate a plurality of initial data sets for an SCR reactor in a target system, the initial data sets including a plurality of initial data, the initial data including an ammonia vapor inlet flow rate, the target system including a thermal power plant system;
the acquisition module is used for acquiring reactor data corresponding to the SCR reactor;
an obtaining module, configured to input reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model, so as to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, where the nitrogen oxide emission data sets include a plurality of nitrogen oxide emission data;
the calculation module is used for calculating the group goodness corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, and the group goodness is used for evaluating the goodness of the initial data group;
and the circulation module is used for generating a plurality of initial data sets input into the nitrogen oxide prediction model next time according to the plurality of initial data sets if the maximum group goodness of the plurality of group goodness does not meet the target condition, and obtaining a target data set until the maximum group goodness of the plurality of initial data sets meets the target condition.
In one embodiment, the calculation module is specifically configured to: calculating a first-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and first reference nitrogen oxide emission data; calculating a second-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and second reference nitrogen oxide emission data; calculating the data consumption loss corresponding to the target nitrogen oxide emission data set according to the initial data set and the reference data set corresponding to the target nitrogen oxide emission data set; calculating the corresponding balance degree of the target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system; and obtaining the group goodness corresponding to the target nitrogen oxide emission data group according to the first-level emission limit loss, the second-level emission limit loss, the data loss and the balance degree corresponding to the target nitrogen oxide emission data group.
In one embodiment, the loop module is specifically configured to: determining the combination probability corresponding to the initial data group according to the group goodness of evaluation corresponding to the initial data group; and generating a plurality of initial data sets which are input into the nitrogen oxide prediction model next time according to the combined probability corresponding to each initial data set.
In one embodiment, the loop module is specifically configured to: adding the group goodness corresponding to the plurality of initial data groups to obtain a total goodness; and dividing the group goodness corresponding to the initial data group by the total goodness to obtain the combined probability corresponding to the initial data group.
In one embodiment, the loop module is specifically configured to: combining the plurality of initial data groups pairwise according to the combination probability corresponding to each initial data group to obtain a plurality of initial data group pairs; crossing two initial data sets in the plurality of initial data set pairs to obtain a plurality of crossed data set pairs; and (3) carrying out variation on two initial data sets in the crossed data set pair to obtain a plurality of initial data sets which are input into the nitrogen oxide prediction model next time.
In one embodiment, the reactor data includes operating condition data, time data, derivative data, and time delay data.
In one embodiment, a plurality of SCR reactors are arranged in the target system, and the nitrogen oxide prediction model comprises a prediction neural network and a sequence-to-sequence model corresponding to each SCR reactor.
In a third aspect, a computer device is provided, which is characterized by comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the data determination method in the thermal power plant system as described above when executing the computer program.
In a fourth aspect, a computer readable storage medium is provided, having stored therein computer program instructions, which when read and executed by a processor, perform the steps of the data determination method in the thermal power plant system as described above.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a schematic flow chart illustrating an implementation of a data determination method in a thermal power plant system according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a sequence-to-sequence model in an embodiment of the present application;
FIG. 3 is a schematic diagram of a NOx prediction model in an embodiment of the present application;
FIG. 4 is a schematic diagram of a crossover in the present application;
FIG. 5 is a schematic representation of variations in the present application;
fig. 6 is a schematic structural diagram of a data determination device in a thermal power plant system according to an embodiment of the present application;
fig. 7 is a block diagram of an internal structure of a computer device in the embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In one embodiment, a method of data determination in a thermal power plant system is provided. The execution subject of the data determination method in the thermal power plant system according to the embodiment of the present invention is a computer device capable of implementing the data determination method in the thermal power plant system according to the embodiment of the present invention, and the computer device may include, but is not limited to, a terminal and a server. The terminal comprises a desktop terminal and a mobile terminal, wherein the desktop terminal comprises but is not limited to a desktop computer and a vehicle-mounted computer; mobile terminals include, but are not limited to, cell phones, tablets, laptops, and smartwatches. The server includes a high performance computer and a cluster of high performance computers.
In one embodiment, as shown in fig. 1, there is provided a data determination method in a thermal power plant system, including:
step 100, generating a plurality of initial data sets for an SCR reactor in a target system, wherein the initial data sets comprise a plurality of initial data, the initial data comprises ammonia steam inlet flow, and the target system comprises a thermal power plant system.
Under the action of a catalyst, reducing agent ammonia vapor in the SCR reactor selectively reduces nitric oxide and nitrogen dioxide into nitrogen at a certain temperature, and oxidation reaction of the reducing agent ammonia vapor and oxygen hardly occurs, so that the selectivity of the reducing agent to the nitric oxide and the nitrogen dioxide is improved, and the consumption of the reducing agent ammonia vapor is reduced.
And the ammonia steam inlet flow rate reflects the flow rate of the reducing agent ammonia steam entering the SCR reactor, and the larger the ammonia steam inlet flow rate is, the more the ammonia steam entering the SCR reactor in unit time is, the smaller the ammonia steam inlet flow rate is, and the less the ammonia steam entering the SCR reactor in unit time is.
Thermal power plant refers to thermal power plant.
The number of SCR reactors in the target system may be one, or two or more. When the number of the SCR reactors in the target system is one, each initial data group includes a plurality of initial data each inputted to the unique SCR reactor, for example, there are 100 initial data groups each having 5 initial data, and the ith initial data group is [ Ai1, Ai2, Ai3, Ai4, Ai5], and all of the 100 initial data groups are inputted to the unique SCR reactor in the target system; when the number of the SCR reactors in the target system is two or more, each of the initial data groups includes a plurality of initial data required to be input to each SCR reactor in the target system, for example, there are 100 initial data groups, there are 2 SCR reactors in total in the target system, respectively, the SCR reactor 1 and the SCR reactor 2, and the ith initial data group is [ Ai1, Ai2, Ai3, Ai4, Ai5, Ai6, Ai7, Ai8, Ai9, Ai 6326 ], Ai1, Ai2, Ai3, Ai4, Ai5 in the 100 initial data groups are input to the SCR reactor 1, and Ai6, Ai 6866, Ai8, Ai9, Ai10 in the 100 initial data groups are input to the SCR reactor 2.
Illustratively, generating a plurality of initial data sets includes: a plurality of initial data sets are randomly generated.
When the range of each data in the target data set is unclear, a plurality of initial data sets can be randomly generated, and the method is quite suitable for novice operators.
Illustratively, generating a plurality of initial data sets includes: acquiring a reference data range of each initial data in the initial data group; a plurality of initial data sets are generated according to the reference data range of each initial data in the initial data sets.
For example, the initial data groups include 4 initial data, the reference data range of the first initial data is [ P1, P2], the reference data range of the second initial data is [ P3, P4], the reference data range of the third initial data is [ P5, P6], the reference data range of the fourth initial data is [ P7, P8], and thus the first initial data in each generated initial data group needs to be within [ P1, P2], the second initial data needs to be within [ P3, P4], the third initial data needs to be within [ P5, P6], and the fourth initial data needs to be within [ P7, P8 ].
Because the reference data range is preset, the target data set is obtained in the reference data range, and compared with the random generation of the initial data set, the optimization range can be reduced, so that the iteration times are reduced, and the speed of obtaining the target data set is accelerated to a certain extent.
And 200, acquiring reactor data corresponding to the SCR reactor.
Reactor data, data relating to an SCR reactor. For example, the flue gas temperature at the inlet of the SCR reactor.
When only one SCR reactor is arranged in a target system, reactor data corresponding to the SCR reactor is acquired; when two or more SCR reactors are arranged in the target system, reactor data corresponding to each SCR reactor is acquired.
Step 300, inputting the reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, where the nitrogen oxide emission data sets include a plurality of nitrogen oxide emission data.
A nitrogen oxide prediction model which is a model capable of predicting nitrogen oxide emission data; nitrogen oxide emission data, which is data reflecting the nitrogen oxide emission situation, for example, nitrogen oxide emission data corresponding to a target system, refers to the concentration of nitrogen oxide at the inlet of a chimney; the nitrogen oxide emission data group comprises a plurality of nitrogen oxide emission data, and one initial data group correspondingly generates one nitrogen oxide emission data group.
Step 400, calculating a group goodness corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, wherein the group goodness is used for evaluating the goodness of the initial data group.
And the corresponding group goodness of the nitrogen oxide emission data group is used for evaluating the goodness of the nitrogen oxide emission data group, and the goodness of the nitrogen oxide emission data group reflects the goodness of the initial data group as one initial data group correspondingly generates one nitrogen oxide emission data group. The higher the group goodness corresponding to the nitrogen oxide emission data set, the more excellent the nitrogen oxide emission data set is, i.e. the more excellent the initial data set is, and the lower the group goodness corresponding to the nitrogen oxide emission data set is, the worse the nitrogen oxide emission data set is, i.e. the worse the initial data set is.
Because each nitrogen oxide emission data in the nitrogen oxide emission data group can pollute the air, the more the emission is, the larger the air pollution is, the less the emission is, and the smaller the air pollution is, therefore, each nitrogen oxide emission data in the nitrogen oxide emission data group can be subjected to pollution analysis, and the group goodness corresponding to the nitrogen oxide emission data group is determined according to the result of the pollution analysis, for example, the pollution analysis result is that the pollution is smaller, the more excellent the nitrogen oxide emission data group is, and the larger the group goodness corresponding to the nitrogen oxide emission data group is; the pollution analysis result is that the larger the pollution is, the worse the nitrogen oxide emission data set is, and the better the group corresponding to the nitrogen oxide emission data set is.
Providing an example for determining the group goodness of the nitrogen oxide emission data group, acquiring a reference emission limit loss, subtracting the reference emission limit loss from each nitrogen oxide emission data in the nitrogen oxide emission data group to obtain the data goodness of each nitrogen oxide emission data in the nitrogen oxide emission data group, and adding the data goodness of each nitrogen oxide emission data in the nitrogen oxide emission data group to obtain the group goodness of the nitrogen oxide emission data group. For example, the nox emission data set is [ D1, D2, D3, D4, D5], the reference emission limit loss is Dc, and thus the group goodness corresponding to the nox emission data set is: (D1-Dc) + (D2-Dc) (D3-Dc) + (D4-Dc) + (D5-Dc).
And 500, if the maximum group goodness of the plurality of group goodness of scores does not meet the target condition, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtaining a target data group until the maximum group goodness of scores meets the target condition.
A target condition, which is a preset condition for determining the superiority and inferiority of the maximum group goodness, for example, the maximum group goodness is greater than the preset goodness, wherein the preset goodness is a preset goodness, and a value of the preset goodness can be set to be relatively greater, so that the obtained target data set is more excellent (for example, the pollution to the environment is smaller), and correspondingly, if the maximum group goodness in the plurality of group goodness does not satisfy the target condition, a plurality of initial data sets input into the nox prediction model next time are generated according to the plurality of initial data sets, including: if the maximum group goodness evaluation in the plurality of group goodness evaluations is smaller than a preset goodness evaluation, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups; for another example, if the target condition is that the number of calculations corresponding to the largest group goodness of the plurality of group goodness of merits reaches a predetermined number of times, where the predetermined number of times is a preset number of times, and correspondingly, if the largest group goodness of the plurality of group goodness of merits does not satisfy the target condition, a plurality of initial data groups to be input into the nox prediction model next time are generated according to the plurality of initial data groups, including: if the calculation times corresponding to the maximum group goodness in the plurality of group goodness is less than the preset times, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and increasing the calculation times corresponding to the maximum group goodness in the plurality of group goodness by 1 every time step 400 is executed; for another example, if the target condition is that the absolute value of the difference between the maximum group goodness evaluation and the last maximum group goodness evaluation in the plurality of group goodness evaluations is smaller than a preset absolute value, and correspondingly, if the maximum group goodness evaluation in the plurality of group goodness evaluations does not satisfy the target condition, a plurality of initial data groups input into the nox prediction model next time are generated according to the plurality of initial data groups, including: if the absolute value of the difference between the maximum group goodness of the plurality of group goodness of the group and the last "maximum group goodness of the plurality of group goodness of the group" is greater than or equal to the absolute value of a preset value, then generating a plurality of initial data groups to be input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, for example, assuming that the maximum group goodness of the plurality of group goodness of the initial data groups and the reactor data corresponding to the SCR reactor is currently calculated as a according to the plurality of initial data groups and the reactor data corresponding to the SCR reactor, and the maximum group goodness of the plurality of group goodness of the SCR reactor is calculated as B last time, then the last "maximum group goodness of the plurality of group goodness of the group goodness" is B, and if | a-B | is greater than or equal to the preset absolute value C, then generating a plurality of initial data groups to be input into the nitrogen oxide prediction model next time, and obtaining a target data group until the absolute value of the difference between the maximum group goodness in the plurality of group goodness and the last 'maximum group goodness in the plurality of group goodness' is less than C.
Acquiring the maximum group goodness of the group corresponding to the multiple nitrogen oxide discharge data groups, if the maximum group goodness of the group is still smaller than the preset goodness of the group, generating multiple initial data groups input into a nitrogen oxide prediction model next time according to the multiple initial data groups, and repeatedly executing the steps 300 to 500 until the maximum group goodness of the group meets the target condition to obtain a target data group, wherein the target data group is the initial data group capable of obtaining a better nitrogen oxide discharge data group.
In the above embodiment, a plurality of initial data sets are generated for the SCR reactor in the target system, then the initial data sets and reactor data corresponding to the SCR reactor are input into the nitrogen oxide prediction model, nitrogen oxide discharge data is predicted by the nitrogen oxide prediction model, then group goodness is calculated according to the predicted nitrogen oxide discharge data, that is, goodness of the initial data sets is calculated, and when the maximum group goodness meets the target condition, the target data set is obtained. Therefore, through the mode, continuous adjustment is not needed, and only the optimal target data set is selected through a mode of calculating the group goodness under the condition that the reactor data is determined, so that the adjustment efficiency is improved compared with a manual adjustment mode, and in addition, the optimal target data set (ammonia steam inlet flow rate) can be quickly obtained through the mode.
In one embodiment, the reactor data includes: condition data, time data, derivative data, and delay data.
And the working condition data is data reflecting the working condition of the SCR reactor, and comprises SCR inlet pressure, SCR inlet-outlet pressure difference, SCR inlet flue gas temperature, SCR inlet oxygen and boiler load. Wherein the SCR inlet pressure is the pressure of the flue gas at the inlet of the SCR reactor; the pressure difference between the SCR inlet and the SCR outlet is high and is the difference between the pressure of the flue gas at the SCR reactor outlet and the pressure of the flue gas at the SCR reactor inlet; the SCR inlet flue gas temperature is the temperature of the flue gas at the inlet of the SCR reactor; SCR inlet oxygen, reflecting the condition of oxygen entering the SCR reactor; the boiler load, which is the amount of power generated by the boiler, also affects the nitrogen oxides output by the SCR reactor.
Time data, which is time corresponding to each data, for example, data of initial data [ Ai1, Ai2, Ai3, Ai4, Ai5]Then the time corresponding to the initial data is [ T ]Ai1，TAi2，TAi3，TAi4，TAi5]As another example, the data is boiler load, assuming boiler load is [ G1, G2, G3, G4, G5]And the time corresponding to the boiler load is [ T ]G1，TG2，TG3，TG4，TG5]。
The derivative data is derived for data, for example, for boiler load, and by deriving the boiler load, it is possible to determine whether the trend of change of the boiler load is large or small.
The time delay data is nitrogen oxide emission data (nitrogen oxide concentration) at the outlet of the SCR reactor within a current period of time, for example, when the current time is 10 o 'clock 30 min 00 sec, and the period of time is 5 min, [ S1, S2, S3, S4, S5], where S1 is nitrogen oxide emission data at the outlet of the SCR reactor at 10 o' clock 25 min 00 sec, S2 is nitrogen oxide emission data at the outlet of the SCR reactor at 10 o 'clock 26 min 00 sec, S3 is nitrogen oxide emission data at the outlet of the SCR reactor at 10 o' clock 27 min 00 sec, S4 is nitrogen oxide emission data at the outlet of the SCR reactor at 10 o 'clock 28 min 00 sec, and S5 is nitrogen oxide emission data at the outlet of the SCR reactor at 10 o' clock 29 min 00 sec.
In the embodiment, not only a plurality of initial data sets but also a plurality of types of SCR reactor data are input in the process of obtaining the target data set, so that the finally obtained target data set is more accurate.
In one embodiment, a plurality of SCR reactors are arranged in the target system, and the nitrogen oxide prediction model comprises a prediction neural network and a sequence-to-sequence model corresponding to each SCR reactor; step 300, inputting the reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, including:
step 301, for each SCR reactor of the plurality of SCR reactors, inputting a sequence corresponding to the SCR reactor to a sequence model according to operating condition data, time data, derivative data corresponding to the SCR reactor and initial data corresponding to the SCR reactor in the plurality of initial data groups to obtain a plurality of outlet data groups of an outlet of the SCR reactor, where each outlet data group includes a plurality of outlet data, and the outlet data includes nitrogen oxide emission data.
A sequence-to-sequence model, composed of an encoder and a decoder, the encoder is used for encoding the input to obtain the encoding characteristics, the decoder is used for decoding according to the encoding characteristics to obtain the desired output, as shown in fig. 2, x1 to x5 compose the input of the sequence-to-sequence model, v1 to v5 compose the output of the sequence-to-sequence model, since hi (i > -2) is generated by hi-1 and xi, in order to keep the overall structure consistent, h0 is set to generate h1 when i ═ 1, and likewise, since qi (i > -2) is generated by qi-1 and C, in order to keep the overall structure consistent, q0 is set to generate q1 when i ═ 1. In the embodiment of the invention, x1 to x5 are data (including initial data, operating condition data, time data and derivative data) at different times, and v1 to v5 are outputs (nitrogen oxide emission data) at different times.
The nitrogen oxide emission data at the outlet of the SCR reactor is an intermediate result of a nitrogen oxide prediction model, and the nitrogen oxide emission data at the inlet of a chimney needs to be obtained by combining the nitrogen oxide emission data at the outlets of all the SCR reactors in a target system.
For example, in fig. 3, there are 2 SCR reactors, SCR reactor 1 and SCR reactor 2, and the nox prediction model includes: sequence-to-sequence model 1 corresponding to SCR reactor 1, and sequence-to-sequence model 2 corresponding to SCR reactor 2, for the ith initial data group, operating condition data 1, time data 1, derivative data 1 corresponding to SCR reactor 1, and initial data [ Ai1, Ai2, Ai3, Ai4, Ai5] corresponding to SCR reactor 1 in the ith initial data group are input into sequence model 1 to obtain a plurality of outlet data groups [ Mi1, Mi2, Mi3, Mi4, Mi5] corresponding to SCR reactor 2, time data 2, derivative data 2, and initial data [ Ai6, Ai7, Ai 638, Ai9, Ai10] corresponding to SCR reactor 2 in the ith initial data group, and initial data [ Ai 59629, Ai7, Ai 596 2, Ai 638, Mi9, Mi10] corresponding to SCR reactor 2 in the ith initial data group, and SCR reactor 2 are input into sequence model 2 to obtain a plurality of outlet data [ Mi 596 3, Mi9, Mi 3642 ] of SCR reactor 2.
Step 302, inputting the plurality of outlet data sets at the outlet of each SCR reactor and the time delay data corresponding to each SCR reactor into the prediction neural network, so as to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system.
And the prediction neural network is used for obtaining the nitrogen oxide emission data of the chimney inlet by combining the nitrogen oxide emission data and the time delay data of all SCR reactor outlets in the target system.
As shown in fig. 3, a plurality of outlet data sets [ Mi1, Mi2, Mi3, Mi4, Mi5] at the outlet of the SCR reactor 1, a plurality of outlet data sets [ Mi6, Mi7, Mi8, Mi9, Mi10] at the outlet of the SCR reactor 2, and a plurality of delay data sets [ Mi6, Mi7, Mi8, Mi9, Mi10] at the outlet of the SCR reactor 2 are input into the prediction neural network, so as to obtain an ith nitrogen oxide emission data set [ Di1, Di2, Di3, Di4, Di5] corresponding to the target system.
In the embodiment, when a plurality of SCR reactors are provided in the target system, a sequence-to-sequence model is established for each SCR reactor to predict the outlet data (nitrogen oxide emission data) of each SCR reactor outlet, so as to improve the accuracy of the finally predicted nitrogen oxide emission data.
In one embodiment, before generating the plurality of initial data sets for the SCR reactor in the target system in step 100, the method further comprises:
and 000, training the nitrogen oxide prediction model.
Acquiring reactor training data corresponding to each SCR reactor in a target system; acquiring an initial training data set corresponding to each SCR reactor in a target system, wherein the initial training data set comprises a plurality of initial training data, and the initial training data comprises ammonia steam inlet training flow; acquiring a real nitrogen oxide emission data set; acquiring a real outlet data group of each SCR reactor outlet; inputting reactor training data and an initial training data set corresponding to each SCR reactor into a nitrogen oxide prediction model to obtain a training outlet data set and a training nitrogen oxide discharge data set of each SCR reactor outlet; calculating the outlet loss corresponding to each SCR reactor according to the training outlet data group of each SCR reactor outlet and the corresponding real outlet data group of the SCR reactor outlet; calculating the emission loss according to the training nitrogen oxide emission data set and the real nitrogen oxide emission data set; summing the discharge loss and the outlet loss corresponding to each SCR reactor to obtain a total loss; and training the nitrogen oxide prediction model according to the total loss until the total loss is less than the preset total loss, and obtaining the trained nitrogen oxide prediction model.
Illustratively, the average absolute error MAE is used to calculate the exit loss and the drain loss.
In order to ensure the training precision, the reliability of each data participating in the training needs to be ensured, and therefore, the obtaining of the reactor training data and the initial training data set corresponding to each SCR reactor in the target system includes: acquiring reactor original training data and an initial original training data set corresponding to each SCR reactor in a target system; removing abnormal values of the reactor original training data and the initial original training data group corresponding to each SCR reactor in the target system to obtain reactor training normal data and an initial training normal data group corresponding to each SCR reactor in the target system; and carrying out missing data filling processing on the reactor training normal data and the initial training normal data group corresponding to each SCR reactor in the target system to obtain the reactor training data and the initial training data group corresponding to each SCR reactor in the target system.
For example, assume that 30 initial training data are in an initial training data set, wherein 2 initial training data have an anomaly (for example, an abnormal initial training is identified in 30 initial training data through a box chart), so the 2 abnormal initial training data are removed to obtain an initial training data set only containing 28 initial training data, and after the abnormal data are removed, the initial training data set only contains 28 initial training data and lacks 2 initial training data, so the initial training data set needs to be subjected to missing data padding processing, for example, a linear interpolation method is adopted to perform missing data padding processing on boiler load to obtain an initial training data set containing 30 initial training data.
The embodiment realizes the training of the nitrogen oxide emission model.
In one embodiment, the step 400 of calculating a group goodness score for each nox emission data set based on a plurality of nox emission data sets corresponding to the target system comprises:
step 401, calculating a first-level limited emission loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and first reference nitrogen oxide emission data.
The first reference nox emission data is preset emission data that may seriously affect the environment, for example, the first reference nox emission data is national required first-class emission limit data.
For example, assuming that the first reference nox emission data is Y1 and the target nox emission data set is [ Di1, Di2, Di3, Di4, Di5], then the first-order limiting emission loss is: (Di1-Y1) + (Di2-Y1) + (Di3-Y1) + (Di4-Y1) + (Di 5-Y1).
And 402, calculating a secondary emission limit loss corresponding to the target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and second reference nitrogen oxide emission data.
The second reference nox emission data is preset emission data that may affect the environment to some extent, for example, the second reference nox emission data is second-level emission limit data required by the country.
For example, assuming that the second reference nox emission data is Y2 and the target nox emission data set is [ Di1, Di2, Di3, Di4, Di5], then the secondary emission limit penalty is: (Di1-Y2) + (Di2-Y2) + (Di3-Y2) + (Di4-Y2) + (Di 5-Y2).
And 403, calculating the data consumption loss corresponding to the target nitrogen oxide emission data set according to the initial data set and the reference data set corresponding to the target nitrogen oxide emission data set.
The reference data group is a data group set in advance for measuring the size of the data of the initial data group. For example, assuming that the initial data set corresponding to the target nox emission data set is [ Ai1, Ai2, Ai3, Ai4, Ai5], the reference data set is [ Ac1, Ac2, Ac3, Ac4, Ac5], that is, the respective values in the reference data set are different, the data consumption loss corresponding to the target nox emission data set is (Ai1-Ac1) + (Ai2-Ac2) + (Ai3-Ac3) + (Ai4-Ac4) + (Ai4-Ac4) + (Ai5-Ac 5); for another example, if the reference data set is [ Ac, Ac ], i.e., if the values in the reference data set are the same, the target nox emissions data set will have a corresponding data loss of (Ai1-Ac) + (Ai2-Ac) + (Ai3-Ac) + (Ai4-Ac) + (Ai4-Ac) + (Ai 5-Ac).
Step 404, calculating a corresponding balance degree of the target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system.
Assume that the target nitrogen oxide emission data set is [ Di1, Di2, Di3, Di4, Di5]Thus, the target nox emission data set corresponds to a degree of balance of: ((Di1-D)2+(Di2-D)2+(Di3-D)2+(Di4-D)2+(Di5-D)2) And/5, wherein D is the mean of Di1, Di2, Di3, Di4 and Di 5.
And 405, obtaining the group goodness corresponding to the target nitrogen oxide emission data group according to the first-level emission limit loss, the second-level emission limit loss, the data loss and the balance degree corresponding to the target nitrogen oxide emission data group.
Acquiring a first weight corresponding to the primary emission limiting loss, a second weight corresponding to the secondary emission limiting loss, a third weight corresponding to the data consumption loss and a fourth weight corresponding to the balance degree; and carrying out weighted summation on the primary limited emission loss, the secondary limited emission loss, the data loss and the balance degree according to the first weight corresponding to the primary limited emission loss, the second weight corresponding to the secondary limited emission loss, the third weight corresponding to the data consumption loss and the fourth weight corresponding to the balance degree to obtain the group goodness corresponding to the target nitrogen oxide emission data group.
For example, if the first weight is a1, the second weight is a2, the third weight is a3, the fourth weight is a4, the first-level limit loss is j1, the second-level limit loss is j2, the data consumption loss is j3, and the balance is j4, then the initial group goodness scores are: a1 × j1+ a2 × j2+ a3 × j3+ a4 × j4+ a5 × j 5; and obtaining the group goodness according to the initial group goodness.
In the embodiment, the group goodness evaluation is generated by combining the four losses, so that the accuracy of the group goodness evaluation is improved.
In one embodiment, the step 500 of generating a plurality of initial data sets for a next input to the nox prediction model based on the plurality of initial data sets comprises:
step 501, determining the combination probability corresponding to the initial data group according to the group goodness corresponding to the initial data group.
The greater the group goodness corresponding to the initial data group is, the greater the combination probability corresponding to the initial data group is determined to be; the smaller the group goodness corresponding to the initial data group is, the smaller the combination probability corresponding to the initial data group is determined to be. For example, there are 4 initial data sets in total, and the group goodness scores corresponding to the 4 initial data sets are respectively: k1, K2, K3 and K4, and K1< K2< K3< K4, then the combined probabilities for the 4 initial data sets are determined to be R1, R2, R3 and R4, and R1< R2< R3< R4.
And 502, generating a plurality of initial data sets input into the nitrogen oxide prediction model next time according to the combined probability corresponding to each initial data set.
The higher the combined probability, the more times the initial data set is used to generate the next input nox prediction model; the lower the combined probability, the fewer times the initial data set is used to generate the next input nox prediction model. In this way, excellent initial data sets can be combined to obtain a plurality of initial data sets to be input into the nox prediction model next time, and the speed of obtaining the target initial data set can be increased because the excellent initial data sets are combined.
The above embodiments provide a method for generating a plurality of initial data sets, and particularly, a plurality of initial data sets to be input into the nox prediction model next time are generated according to the group goodness of evaluation, it can be understood that the initial data set with the high group goodness of evaluation will have a higher probability of being used for generating the initial data set to be input into the nox prediction model next time, thereby increasing the speed of obtaining the target data set.
In one embodiment, the step 501 of determining the combined probability corresponding to the initial data group according to the size of the group goodness corresponding to the initial data group includes:
step 501A, adding the group goodness corresponding to the plurality of initial data groups to obtain a total goodness.
For example, there are 4 initial data sets in total, and the group goodness scores corresponding to the 4 initial data sets are respectively: k1, K2, K3 and K4, so the overall goodness score was K1+ K2+ K3+ K4.
And step 501B, dividing the group goodness corresponding to the initial data group by the total goodness to obtain the combination probability corresponding to the initial data group.
As another example, the combined probability for the 4 initial data sets is: r1 ═ K1/(K1+ K2+ K3+ K4), R2 ═ K2/(K1+ K2+ K3+ K4), R ═ K3/(K1+ K2+ K3+ K4), and R4 ═ K4/(K1+ K2+ K3+ K4).
The above embodiment provides a method for determining a combination probability, and the method can simply obtain the combination probability by solving a mode of calculating the ratio of the group goodness to the total goodness.
In one embodiment, the step 502 of generating a plurality of initial data sets to be input into the nox prediction model next time according to the combined probability corresponding to each of the initial data sets comprises:
step 502A, combining the plurality of initial data sets pairwise according to the combination probability corresponding to each initial data set to obtain a plurality of initial data set pairs.
For example, there are 6 initial data sets, U1, U2, U3, U4, U5, and U6, and the magnitude relationship of the corresponding combined probabilities is: r1< R2< R3< R4< R5< R6, since the larger the combination probability, the more the number of times of participation in combination is, the more the combination frequency of the initial data group corresponding to R6 is set as the combination frequency of the initial data group corresponding to R5 is set as the combination frequency of the initial data group corresponding to R4 is set as the combination frequency of the initial data group corresponding to R3 is set as the combination frequency of the initial data group corresponding to R2 is set as the combination frequency of the initial data group corresponding to R1, and thus, 3 initial data group pairs (U6, U5), (U6, U4), (U5, U3) can be obtained, and of course, 3 initial data group pairs (U6, U5), (U6, U4), (U6, U3) can be obtained as long as the magnitude relation of the above-mentioned combination frequency is satisfied.
Step 502B, intersecting two of the plurality of initial data group pairs to obtain a plurality of intersecting data group pairs.
For example, the initial data group pair is (U6, U5), where U6 is [ a61, a62, a63, a64, a65, a66, a67, a68, a69, a610], U5 is [ a51, a52, a53, a54, a55, a56, a57, a58, a59, a510], and each of the initial data groups U6 and U5 is converted into binary data, for example, binary data representing Aij with Aij '(for example, assuming a61, binary data a 61' of a61 is 01010100), and then the data corresponding to the initial data group U6 is obtained as: a61 ' a62 ' a63 ' a64 ' a65 ', the initial data set U5 corresponds to binary data: a51 ' a52 ' a53 ' a54 ' a55 ', representing a61 ' a62 ' a63 ' a64 ' a65 ' and a51 ' a52 ' a53 ' a54 ' a55 ' in a binary manner, are:
010101000101011001011100011101001101010001011100，
010101010111011101001100001100100100110101111101。
and crossing the two initial data sets to obtain a crossed data set pair. Fig. 4 provides an example of interleaving, and it can be seen that, since the initial data groups are already represented as binary data, interleaving of two initial data groups can be implemented by interleaving the binary data, and as shown in fig. 4, interleaving of two initial data groups can be implemented by one interleaving or multiple interleaving, resulting in an interleaved data group pair.
And 502C, performing variation on two initial data sets in the crossed data set pair to obtain a plurality of initial data sets input into the nitrogen oxide prediction model next time.
Fig. 5 shows the mutation process for two initial datasets in a crossing dataset pair for one crossover, e.g., two mutations at 010101000101011001011100011101001101110001011100 to yield 000100000101011001011100011101001101110001011100 and one mutation at 010101010111011101001100001100100100010101111101 to yield 000101010111011101001100001100100100010101111101.
The above-described embodiment generates a plurality of initial data sets for the next input nox prediction model by combining data pairs, cross data pairs, and variations.
In one embodiment, there is provided a data determination apparatus 600 in a thermal power plant system, including:
a generating module 601, configured to generate a plurality of initial data sets for an SCR reactor in a target system, where the initial data sets include a plurality of initial data, the initial data includes an ammonia vapor inlet flow rate, and the target system includes a thermal power plant system;
an obtaining module 602, configured to obtain reactor data corresponding to the SCR reactor;
an obtaining module 603, configured to input reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model, so as to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, where the nitrogen oxide emission data sets include a plurality of nitrogen oxide emission data;
a calculating module 604, configured to calculate a group goodness corresponding to each nox emission data group according to a plurality of nox emission data groups corresponding to the target system, where the group goodness is used to evaluate the goodness of the initial data group;
and a circulation module 605, configured to, if the largest group goodness evaluation of the plurality of group goodness evaluations does not meet a target condition, generate a plurality of initial data groups to be input to the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtain a target data group until the largest group goodness evaluation meets the target condition.
In an embodiment, the calculating module 604 is specifically configured to: calculating a first-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and first reference nitrogen oxide emission data; calculating a second-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and second reference nitrogen oxide emission data; calculating the data consumption loss corresponding to the target nitrogen oxide emission data set according to the initial data set and the reference data set corresponding to the target nitrogen oxide emission data set; calculating the corresponding balance degree of the target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system; and obtaining the group goodness corresponding to the target nitrogen oxide emission data group according to the first-level emission limit loss, the second-level emission limit loss, the data loss and the balance degree corresponding to the target nitrogen oxide emission data group.
In one embodiment, the loop module 605 is specifically configured to: determining the combination probability corresponding to the initial data group according to the group goodness of evaluation corresponding to the initial data group; and generating a plurality of initial data sets which are input into the nitrogen oxide prediction model next time according to the combined probability corresponding to each initial data set.
In one embodiment, the loop module 605 is specifically configured to: adding the group goodness corresponding to the plurality of initial data groups to obtain a total goodness; and dividing the group goodness corresponding to the initial data group by the total goodness to obtain the combined probability corresponding to the initial data group.
In one embodiment, the loop module 605 is specifically configured to: combining the plurality of initial data groups pairwise according to the combination probability corresponding to each initial data group to obtain a plurality of initial data group pairs; crossing two initial data sets in the plurality of initial data set pairs to obtain a plurality of crossed data set pairs; and (3) carrying out variation on two initial data sets in the crossed data set pair to obtain a plurality of initial data sets which are input into the nitrogen oxide prediction model next time.
In one embodiment, the reactor data includes operating condition data, time data, derivative data, and time delay data.
In one embodiment, a plurality of SCR reactors are arranged in the target system, and the nitrogen oxide prediction model comprises a prediction neural network and a sequence-to-sequence model corresponding to each SCR reactor.
In one embodiment, as shown in fig. 7, a computer device is provided, which may be a terminal or a server in particular. The computer device comprises a processor, a memory and a network interface which are connected through a system bus, wherein the memory comprises a nonvolatile storage medium and an internal memory, the nonvolatile storage medium of the computer device stores an operating system, and also stores a computer program, and when the computer program is executed by the processor, the processor can realize the data determination method in the thermal power plant system. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM). The internal memory may also have stored therein a computer program that, when executed by the processor, causes the processor to perform a method of determining data in a thermal power plant system. Those skilled in the art will appreciate that the architecture shown in fig. 7 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
The data determination method in the thermal power plant system provided by the present application may be implemented in the form of a computer program that is executable on a computer device as shown in fig. 7. The memory of the computer device may store therein various program templates that make up the data determination device in the thermal power plant system. For example, the generating module 601, the obtaining module 602, and the obtaining module 603.
A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of:
generating a plurality of initial data sets for an SCR reactor in a target system, the initial data sets comprising a plurality of initial data, the initial data comprising an ammonia vapor inlet flow rate, the target system comprising a thermal power plant system;
acquiring reactor data corresponding to the SCR reactor;
inputting reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, wherein the nitrogen oxide emission data sets comprise a plurality of nitrogen oxide emission data;
calculating a group goodness degree corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, wherein the group goodness degree is used for evaluating the goodness and the badness of the initial data group;
and if the maximum group goodness of the plurality of group goodness of comments does not meet the target condition, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtaining a target data group until the maximum group goodness of comment meets the target condition.
In one embodiment, a computer readable storage medium is provided, storing a computer program that, when executed by a processor, causes the processor to perform the steps of:
generating a plurality of initial data sets for an SCR reactor in a target system, the initial data sets comprising a plurality of initial data, the initial data comprising an ammonia vapor inlet flow rate, the target system comprising a thermal power plant system;
acquiring reactor data corresponding to the SCR reactor;
inputting reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, wherein the nitrogen oxide emission data sets comprise a plurality of nitrogen oxide emission data;
calculating a group goodness degree corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, wherein the group goodness degree is used for evaluating the goodness and the badness of the initial data group;
and if the maximum group goodness of the plurality of group goodness of comments does not meet the target condition, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtaining a target data group until the maximum group goodness of comment meets the target condition.
It should be noted that the data determining method in the thermal power plant system, the data determining apparatus in the thermal power plant system, the computer device and the computer readable storage medium described above belong to a general inventive concept, and the contents in the data determining method in the thermal power plant system, the data determining apparatus in the thermal power plant system, the computer device and the computer readable storage medium embodiment may be mutually applicable.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form. In addition, units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment. Furthermore, the functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part. In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims (10)

1. A method of data determination in a thermal power plant system, comprising:
generating a plurality of initial data sets for an SCR reactor in a target system, the initial data sets comprising a plurality of initial data, the initial data comprising an ammonia vapor inlet flow rate, the target system comprising a thermal power plant system;
acquiring reactor data corresponding to the SCR reactor;
inputting reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, wherein the nitrogen oxide emission data sets comprise a plurality of nitrogen oxide emission data;
calculating a group goodness degree corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, wherein the group goodness degree is used for evaluating the goodness and the badness of the initial data group;
and if the maximum group goodness of the plurality of group goodness of comments does not meet the target condition, generating a plurality of initial data groups input into the nitrogen oxide prediction model next time according to the plurality of initial data groups, and obtaining a target data group until the maximum group goodness of comment meets the target condition.
2. The method of claim 1, wherein calculating a group goodness score for each nox emission data group based on a plurality of nox emission data groups for the target system comprises:
calculating a first-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and first reference nitrogen oxide emission data;
calculating a second-level emission limit loss corresponding to a target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system and second reference nitrogen oxide emission data;
calculating the data consumption loss corresponding to the target nitrogen oxide emission data set according to the initial data set and the reference data set corresponding to the target nitrogen oxide emission data set;
calculating the corresponding balance degree of the target nitrogen oxide emission data set according to the target nitrogen oxide emission data set corresponding to the target system;
and obtaining the group goodness corresponding to the target nitrogen oxide emission data group according to the first-level emission limit loss, the second-level emission limit loss, the data loss and the balance degree corresponding to the target nitrogen oxide emission data group.
3. The method of claim 1, wherein said generating a plurality of initial data sets for a next input to said nox prediction model based on said plurality of initial data sets comprises:
determining the combination probability corresponding to the initial data group according to the group goodness of evaluation corresponding to the initial data group;
and generating a plurality of initial data sets which are input into the nitrogen oxide prediction model next time according to the combined probability corresponding to each initial data set.
4. The method of claim 3, wherein determining the combined probability corresponding to the initial data set according to the group goodness of fit corresponding to the initial data set comprises:
adding the group goodness corresponding to the plurality of initial data groups to obtain a total goodness;
and dividing the group goodness corresponding to the initial data group by the total goodness to obtain the combined probability corresponding to the initial data group.
5. The method of claim 3, wherein said generating a plurality of initial data sets for a next input to said NOx predictive model based on a combined probability associated with each of said initial data sets comprises:
combining the plurality of initial data groups pairwise according to the combination probability corresponding to each initial data group to obtain a plurality of initial data group pairs;
crossing two initial data sets in the plurality of initial data set pairs to obtain a plurality of crossed data set pairs;
and (3) carrying out variation on two initial data sets in the crossed data set pair to obtain a plurality of initial data sets which are input into the nitrogen oxide prediction model next time.
6. The method of claim 1, wherein the reactor data comprises operating condition data, time data, derivative data, and time delay data.
7. The method of claim 6, wherein a plurality of SCR reactors are provided in the target system, and wherein the NOx prediction model comprises a prediction neural network and a sequence-to-sequence model corresponding to each of the SCR reactors.
8. A data determination apparatus in a thermal power plant system, comprising:
a generating module configured to generate a plurality of initial data sets for an SCR reactor in a target system, the initial data sets including a plurality of initial data, the initial data including an ammonia vapor inlet flow rate, the target system including a thermal power plant system;
the acquisition module is used for acquiring reactor data corresponding to the SCR reactor;
an obtaining module, configured to input reactor data corresponding to the SCR reactor and the plurality of initial data sets into a nitrogen oxide prediction model, so as to obtain a plurality of nitrogen oxide emission data sets corresponding to the target system, where the nitrogen oxide emission data sets include a plurality of nitrogen oxide emission data;
the calculation module is used for calculating the group goodness corresponding to each nitrogen oxide emission data group according to a plurality of nitrogen oxide emission data groups corresponding to the target system, and the group goodness is used for evaluating the goodness of the initial data group;
and the circulation module is used for generating a plurality of initial data sets input into the nitrogen oxide prediction model next time according to the plurality of initial data sets if the maximum group goodness of the plurality of group goodness does not meet the target condition, and obtaining a target data set until the maximum group goodness of the plurality of initial data sets meets the target condition.
9. A computer arrangement comprising a memory, a processor and a computer program stored in said memory and executable on said processor, said processor implementing the steps of the data determination method in the thermal power plant system according to any one of claims 1 to 7 when executing said computer program.
10. A computer readable storage medium, having stored thereon computer program instructions, which when read and executed by a processor, perform the steps of the data determination method in the thermal power plant system as claimed in any one of claims 1 to 7.
CN202110907440.5A 2021-08-09 2021-08-09 Method, device, equipment and storage medium for determining data in thermal power plant system Active CN113611375B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110907440.5A CN113611375B (en) 2021-08-09 2021-08-09 Method, device, equipment and storage medium for determining data in thermal power plant system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110907440.5A CN113611375B (en) 2021-08-09 2021-08-09 Method, device, equipment and storage medium for determining data in thermal power plant system

Publications (2)

Publication Number Publication Date
CN113611375A true CN113611375A (en) 2021-11-05
CN113611375B CN113611375B (en) 2023-10-10

Family

ID=78307603

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110907440.5A Active CN113611375B (en) 2021-08-09 2021-08-09 Method, device, equipment and storage medium for determining data in thermal power plant system

Country Status (1)

Country Link
CN (1) CN113611375B (en)

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060047607A1 (en) * 2004-08-27 2006-03-02 Boyden Scott A Maximizing profit and minimizing losses in controlling air pollution
US20060047564A1 (en) * 2004-08-27 2006-03-02 Alstom Technology Ltd. Estimating an economic parameter related to a process for controlling emission of a pollutant into the air
CN104880536A (en) * 2015-03-30 2015-09-02 北京京能电力股份有限公司石景山热电厂 Multi-point-location nitrogen oxide continuous monitoring device and continuous monitoring method for SCR reactors
CN105224730A (en) * 2015-09-17 2016-01-06 西南林业大学 The original NO of a kind of high pressure common rail electric-controlled diesel engine 2forecasting of discharged quantity method
CN211042859U (en) * 2019-11-14 2020-07-17 上海明华电力科技有限公司 Flue gas grid sampling monitoring device in inverted trapezoidal flue behind denitration reactor
CN111832222A (en) * 2020-06-28 2020-10-27 成都佳华物链云科技有限公司 Pollutant concentration prediction model training method, prediction method and device
CN111897373A (en) * 2020-08-05 2020-11-06 海南创实科技有限公司 Model prediction-based ammonia injection flow adjusting method for SCR denitration device
CN111921377A (en) * 2020-09-25 2020-11-13 大唐环境产业集团股份有限公司 SCR denitration ammonia injection control method and system based on mechanism and data driving
CN111931426A (en) * 2020-09-25 2020-11-13 大唐环境产业集团股份有限公司 Method and equipment for determining influence factors of concentration of nitrogen oxides at inlet of SCR (Selective catalytic reduction) reactor
CN112085277A (en) * 2020-09-10 2020-12-15 南京工业大学 SCR denitration system prediction model optimization method based on machine learning
CN113009890A (en) * 2021-01-25 2021-06-22 南京工业大学 Thermal power plant SCR flue gas denitration control system optimization control method based on BMO-LSSVM

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060047607A1 (en) * 2004-08-27 2006-03-02 Boyden Scott A Maximizing profit and minimizing losses in controlling air pollution
US20060047564A1 (en) * 2004-08-27 2006-03-02 Alstom Technology Ltd. Estimating an economic parameter related to a process for controlling emission of a pollutant into the air
CN104880536A (en) * 2015-03-30 2015-09-02 北京京能电力股份有限公司石景山热电厂 Multi-point-location nitrogen oxide continuous monitoring device and continuous monitoring method for SCR reactors
CN105224730A (en) * 2015-09-17 2016-01-06 西南林业大学 The original NO of a kind of high pressure common rail electric-controlled diesel engine 2forecasting of discharged quantity method
CN211042859U (en) * 2019-11-14 2020-07-17 上海明华电力科技有限公司 Flue gas grid sampling monitoring device in inverted trapezoidal flue behind denitration reactor
CN111832222A (en) * 2020-06-28 2020-10-27 成都佳华物链云科技有限公司 Pollutant concentration prediction model training method, prediction method and device
CN111897373A (en) * 2020-08-05 2020-11-06 海南创实科技有限公司 Model prediction-based ammonia injection flow adjusting method for SCR denitration device
CN112085277A (en) * 2020-09-10 2020-12-15 南京工业大学 SCR denitration system prediction model optimization method based on machine learning
CN111921377A (en) * 2020-09-25 2020-11-13 大唐环境产业集团股份有限公司 SCR denitration ammonia injection control method and system based on mechanism and data driving
CN111931426A (en) * 2020-09-25 2020-11-13 大唐环境产业集团股份有限公司 Method and equipment for determining influence factors of concentration of nitrogen oxides at inlet of SCR (Selective catalytic reduction) reactor
CN113009890A (en) * 2021-01-25 2021-06-22 南京工业大学 Thermal power plant SCR flue gas denitration control system optimization control method based on BMO-LSSVM

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
PEIRAN XIE等: "Dynamic modeling for NOx emission sequence prediction of SCR system outlet based on sequence to sequence long short-term memory network", ENERGY, vol. 190, pages 1 - 12 *
冯建荣等: "火电厂SCR脱硝系统控制逻辑方案的优化", 硫酸工业, no. 10, pages 26 - 28 *
李瞳: "基于SCR反应器NOx均匀度测试的脱硝系统超低排放改造", 环境工程学报, vol. 12, no. 4, pages 1112 - 1119 *

Also Published As

Publication number Publication date
CN113611375B (en) 2023-10-10

Similar Documents

Publication Publication Date Title
Bayón et al. The exact solution of the environmental/economic dispatch problem
Luo et al. A novel Riccati equation grey model and its application in forecasting clean energy
CN111921377B (en) SCR denitration ammonia injection control method and system based on mechanism and data driving
Epiney et al. Economic analysis of a nuclear hybrid energy system in a stochastic environment including wind turbines in an electricity grid
CN104102782A (en) Method for controlling temperatures of reactors of pressurized water reactor nuclear power stations by aid of RBF (radial basis function) neural networks
CN109670625A (en) NOx emission concentration prediction method based on Unscented kalman filtering least square method supporting vector machine
CN110165698B (en) Wind power plant smooth grid-connection method for realizing prospective error asset conversion
Zhao et al. Toward carbon-neutral electric power systems in the New York State: A novel multi-scale bottom-up optimization framework coupled with machine learning for capacity planning at hourly resolution
Wang et al. Modelling and optimization for a selective assembly process of parts with non-normal distribution
CN110941801A (en) Method and system for predicting reliability of single in-service pressurized water reactor nuclear power unit
CN116599050A (en) Photovoltaic prediction method and related device based on self-attention mechanism
CN110867852B (en) Microgrid energy storage optimization configuration method and device considering whole life cycle cost
Wu et al. Nonlinear dynamic soft-sensing modeling of NOx emission of a selective catalytic reduction denitration system
CN114971048A (en) Zero-carbon park comprehensive energy planning simulation method and system and computer equipment
CN113611375A (en) Data determination method, device, equipment and storage medium in thermal power plant system
CN110674460B (en) E-Seq2Seq technology-based data driving type unit combination intelligent decision method
CN102521510B (en) A kind of electrical network resists concentration power exfoliation ability appraisal procedure
CN108805323B (en) Data prediction method and device
Mohammadi Neural network for univariate and multivariate nonlinearity tests
Zhou et al. Double-tank liquid level control based on genetic algorithm
Liu et al. Application of the main steam temperature control based on sliding multi-level multi-model predictive control
CN112003281A (en) Optimal configuration method, device and equipment for dynamic voltage restorer
CN111859807A (en) Initial pressure optimizing method, device, equipment and storage medium for steam turbine
Tang et al. Multi‐model predictive control of SCR flue gas denitrification system in coal‐fired power plant based on kernel fuzzy c‐means clustering and integrated model
CN114862287B (en) Risk benefit analysis method, system, terminal and medium for cascade power station group scheduling

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
GR01 Patent grant
GR01 Patent grant