OPERATION CONTROL SYSTEM AND A CONTROL METHOD FOR A GAS-STEAM COMBINED CYCLE GENERATOR UNIT

Abstract

An operation control system for a gas-steam combined cycle generator unit-includes a real-time operation data acquisition module, to collect the operation parameter data and power generation data of a power plant unit, an operation status evaluation index mining module, to mine and analyze the operation parameter data of a power plant unit to get key parameters, an operation status evaluation index extraction module, to obtain characteristic variables, an operation characteristic parameter prediction module, to predict the said characteristic variables and obtain the predicted value and the corresponding change trend, and an operation intelligent control module, to realize the intelligent control of parameters. The operation control system can guide the optimal operation of power plant units, and improve their operation reliability and economy.

Claims

1. An operation control system for a gas-steam combined cycle generator unit, comprising a real-time operation data acquisition module, configured to collect operation parameter data and power generation data of a power plant unit, an operation status evaluation index mining module, configured to mine and analyze the operation parameter data of the power plant unit to get key parameters, an operation status evaluation index extraction module, configured to obtain characteristic variables, an operation characteristic parameter prediction module, configured to predict the said characteristic variables and obtain a predicted value and a corresponding change trend, and an operation intelligent control module, configured to realize intelligent control of parameters.

2. The operation control system according to claim 1, further comprising an operation status evaluation index selection module, configured to screen the said key parameters, obtain operation parameters positively correlated with the said power generation data, and send the said operation parameters to the said operation status evaluation index extraction module for processing.

3. The operation control system according to claim 1, further comprising an operation status evaluation index analysis module, configured to analyze the said key parameters and obtain operating condition stability judgment parameters.

4. The operation control system according to claim 3, further comprising a steady operating condition establishment module, configured to label the said operating condition stability judgment parameters, and establish a stable operating condition database.

5. The operation control system according to claim 1, further comprising a data preprocessing module, configured to preprocess the said operation parameter data of the power plant unit.

6. A method for optimizing and controlling operation parameters of a power plant unit, comprising step 1: obtaining operation parameter data and power generation data of a power plant unit with a real-time operation data acquisition module, and preprocessing the operation parameter data and the power generation data; step 2: mining the said operation parameter data obtained in step 1 with operation status evaluation index mining module to get key parameters; step 3: processing the said key parameters with the said operation status evaluation index extraction module and multiple modules to get corresponding characteristic variables; step 4: predicting the said corresponding characteristic variables with an operation characteristic parameter prediction module to get a predicted value and a corresponding change trend; step 5: analyzing the said key parameters with a steady operating condition establishment module to get operating condition stability judgment parameters, and labeling the said operating condition stability judgment parameters to establish a stable operating condition database; step 6: in combination with the said stable operating condition database, the said predicted value and the corresponding change trend, carrying out comparative control with an operation intelligent control module, to realize intelligent control.

7. The method according to claim 6, wherein the said step 3 further comprises step 31: screening the said key parameters with an operation status evaluation index selection module to get operation parameters positively correlated with the said power generation data; step 32: analyzing the said operation parameters with an operation status evaluation index analysis module to get the operating condition stability judgment parameters.

8. The method according to claim 6, wherein in the said step 2, the operation parameter data is mined by an improved association rule mining method.

9. The method according to claim 7, wherein in the said step 32, a clustering method is used to have a correlation analysis of the operation parameters to get the operating condition stability judgment parameters.

10. The operation control system according to claim 2, further comprising a data preprocessing module, configured to preprocess the said operation parameter data of the power plant unit.

11. The operation control system according to claim 3, further comprising a data preprocessing module, configured to preprocess the said operation parameter data of the power plant unit.

12. The operation control system according to claim 4, further comprising a data preprocessing module, configured to preprocess the said operation parameter data of the power plant unit.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0032] To better describe the embodiment of the present invention or the technical scheme of the prior art, a brief introduction of the accompanying drawings to be used in the descriptions of the embodiment or the prior art is made hereby. Obviously, the drawings below are only the embodiment of the present invention, and for those ordinarily skilled in the art, other drawings based on such drawings can be obtained without making creative endeavors.

[0033] FIG. 1 is a structural principle block diagram of an operation control system for a gas-steam combined cycle generator unit provided by the present invention;

[0034] FIG. 2 is a specific flowchart of a clustering process provided in Embodiment 1 of the present invention;

[0035] FIG. 3 is a specific flowchart of the intelligent control provided by Embodiment 1 of the present invention;

[0036] FIG. 4 is a specific flowchart of the improved association rule mining method provided in Embodiment 1 of the present invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] The technical schemes in the embodiments of the present invention are clearly and completely described below in combination with the drawings of the embodiments of the present invention. Obviously, such embodiments are just a part of embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all the other embodiments obtained by those ordinarily skilled in the art without making creative endeavors shall fall into the scope of protection of the present invention.

Embodiment 1

[0038] Referring to FIG. 1, Embodiment 1 of the present invention provides an operation control system for a gas-steam combined cycle generator unit, which comprises

[0039] A real-time operation data acquisition module 1, to collect the operation parameter data and power generation data of a power plant unit,

[0040] An operation status evaluation index mining module 2, to mine and analyze the operation parameter data of a power plant unit to get key parameters,

[0041] An operation status evaluation index extraction module 3, to obtain characteristic variables,

[0042] An operation characteristic parameter prediction module 4, to predict the said characteristic variables and obtain the predicted value and the corresponding change trend,

[0043] Among them, the operation characteristic parameter prediction module 4 trains the characteristic quantities determined in the power plant unit health status characteristic acquisition module with the LSTM neural network model, and predicts the change trend of the parameters over time to assist the status judgment.

[0044] And an operation intelligent control module 5, to realize the intelligent control of parameters.

[0045] In a specific embodiment, it further comprises an operation status evaluation index selection module 6, to screen the key parameters, obtain operation parameters that are positively correlated with the power generation data, and send the operation parameters to the operation status evaluation index extraction module 3 for processing.

[0046] In a specific embodiment, it further comprises an operation status evaluation index analysis module 7, to analyze the said key parameters and obtain the operating condition stability judgment parameters.

[0047] In a specific embodiment, it further comprises a steady operating condition establishment module 8, to label the operating condition stability judgment parameters, and establish a stable operating condition database.

[0048] In a specific embodiment, it further comprises a data preprocessing module 9, to preprocess the operation parameter data of a power plant unit. The data preprocessing module 9 performs abnormal value processing, missing value processing, discretization processing and normalization processing on the data collected by the power plant unit, preparing for the follow-up data mining and analysis.

[0049] Further, Embodiment 1 of the present invention also provides a method for optimizing and controlling the operation parameters of a power plant unit, which comprises

[0050] Step 1: obtain the operation parameter data and power generation data of a power plant unit with the real-time operation data acquisition module 1, and preprocess the data;

[0051] Step 2: mine the operation parameter data obtained in Step 1 with the operation status evaluation index mining module 2 to get the key parameters;

[0052] Step 3: process the key parameters with the operation status evaluation index extraction module 3 and multiple modules to get corresponding characteristic variables;

[0053] Step 4: predict the characteristic variables with the operation characteristic parameter prediction module 4 to get the predicted value and the corresponding change trend;

[0054] Step 5: analyze the key parameters with the steady operating condition establishment module 8 to get the operating condition stability judgment parameters, and label the operating condition stability judgment parameters to establish a stable operating condition database;

[0055] Step 6: in combination with the stable operating condition database, the predicted value and the corresponding change trend, carry out comparative control with the operation intelligent control module 5, to realize intelligent control.

[0056] In a specific embodiment, the Step 3 further comprises Step 31: screen the key parameters with the operation status evaluation index selection module 6 to get the operation parameters that are positively correlated with the power generation data;

[0057] Step 32: analyze the said operation parameters with the said operation status evaluation index analysis module 7 to get the operating condition stability judgment parameters.

[0058] Referring to FIG. 4, in a specific embodiment, in Step 2, the operation parameter data is mined by using the improved association rule mining method, and the specific process is as follows:

[0059] In FIG. 4, the input is a transaction database D composed of n transactions. Each transaction contains in sub-items. There is a set of membership functions, and the j.sup.th (j=1, 2, 3 . . . , m) item in the i.sup.th (i=1, 2, 3 . . . , n) transaction data can be described by the k.sup.th membership function μ.sub.i(R.sub.js) (s=1, 2,3 . . . , k). The set minimum support threshold is mins and the minimum confidence threshold is minc. The output is a set of quantitative association rules.

[0060] The running process of the improved algorithm is as follows:

[0061] 1) Represent each item (j=1, 2, . . . m) of each transaction data T.sub.i (i=1, 2, . . . , n) in the transaction database D with a given membership function as a quantization interval, and describe the items as a set of quantization intervals using Zadeh notation, as shown in Formula (1):

[00001]$\begin{array}{cc}{f}_i^j=\frac{{\mu }_i\left(R_{j1}\right)}{R_{j1}}+\frac{{\mu }_i\left(R_{j2}\right)}{R_{j2}}+.Math.+\frac{{\mu }_i\left(R_{\mathrm{jk}}\right)}{R_{\mathrm{jk}}}& \left(1\right)\end{array}$

[0062] Wherein, f.sub.i.sup.j and t.sub.i.sup.j are the corresponding quantization interval sets, R.sub.ji is the i.sup.th quantization interval partition of item t.sub.i.sup.j, and μ.sub.i(R.sub.ji) is the membership value on partition R.sub.ji.

[0063] 2) Calculate the weight of the membership degree of each item t.sub.i.sup.j (j=1, 2 , . . . m) in n transaction data Ti (i=1, 2, . . . , n) in the corresponding quantization interval set R.sub.ji (s=1, 2, . . . k). The specific expression is shown in Formula (2):

[00002]$\begin{array}{cc}{\mathrm{weight}}_{\mathrm{js}}=\frac{1}{n}\underset{i=1}{\overset{n}{.Math.}}{\mu }_i\left(R_{\mathrm{js}}\right)& \left(2\right)\end{array}$

[0064] Wherein, weight.sub.js is the weight of membership degree, n is the number of transaction data, and μ.sub.i(R.sub.js) is the membership function.

[0065] 3) For each partition R.sub.ji (1≤j≤m, 1≤s≤k), verify whether the weight of each transaction set is not less than the preset minsupport. If the partition weight.sub.js meets that, then put it into the frequent item set L1, as shown in Formula (3):

L.sub.1={R.sub.js|weight.sub.js≥min support, 1≤j≤m, 1≤s≤m}  (3)

[0066] Wherein, minsupport is a preset minimum weight.

[0067] 4) Set r=1, to calculate the total number of transactions remaining in the items after filtering.

[0068] 5) Generate a candidate item set C.sub.r+i from the frequent item set L.sub.r by Apriori. L.sub.r has r−1 identical items in two item sets, while other items are different and belong to two partitions of the same item and thereby can not appear in the same item of the candidate item set C.sub.r+i at the same time.

[0069] 6) Process each newly generated r+1 item set in the candidate item set C.sub.r−1 as follows:

[0070] a. For each transaction data T.sub.i, calculate the membership value of the item t in the candidate large item set in the total transaction item set, as shown in Formula (4):

μ.sub.it=μ.sub.i(R.sub.t.sub.1){circumflex over ( )}μ.sub.i(R.sub.t.sub.2){circumflex over ( )} . . . μ.sub.i(R.sub.tr+1)   (4)

[0071] Wherein, μ.sub.i(R.sub.t.sub.j) is the membership value of the transaction data T.sub.i on the partition.

[0072] If the minimum operators have intersection, then

μ.sub.it=Min μ.sub.i(R.sub.t.sub.j)   (5)

[0073] b. Solve the weight in each sub-item.

[00003]$\begin{array}{cc}{\mathrm{weight}}_t=\frac{1}{n}\underset{i=1}{\overset{n}{.Math.}}{\mu }_{\mathrm{it}}& \left(6\right)\end{array}$

[0074] c. If weight.sub.t is not less than the previously set threshold minsupport, put item t=t.sub.1, t.sub.2. . . , t.sub.r°1) into L.sub.r+1.

[0075] 7) If it is null, go to the next step; otherwise set r=r+1 and repeat Steps 5-7.

[0076] 8) Establish solution rules for all large q (q≥2) item sets t with item 1.

[0077] 9) According to the formula, add the interestingness as a new measure, and the interestingness function is

[00004]$I\left(A.Math.B\right)=\frac{1-P\left(B\right)}{\left(1-P\left(B\right)\right)*\left(1-P\left(A.Math.B\right)\right)}$

[0078] The larger the value of the interestingness I, the more valuable this association rule is. The larger the minimum I value is set, the less the mining results will be, and vice versa.

[0079] The effective association rules with quantization interval attributes mined by the above quantization interval association rule mining algorithm have high reference value for the setting of the final operation parameters.

[0080] In a specific embodiment, the specific screening process of Step 31 is as follows:

[0081] Analyze the correlation between key parameters by the correlation analysis method, select the operation parameters that are significantly positively correlated with the power generation data (i.e. the combined cycle power of gas-steam combined cycle generator set), calculate the corresponding Pearson correlation coefficient, and mine the key characteristic parameters that meet the requirements through stability judgment, extreme value standardization, membership degree setting, quantization interval division and minimum support and minimum confidence value adjustment.

[0082] In a specific embodiment, in Step 32, a clustering method is used to perform a correlation analysis of the operation parameters to get the operating condition stability judgment parameters. The specific process is as follows:

[0083] Determine the stability judgment index of the critical value that may lead to abnormal operation combining the practical production experience, and further screen the preprocessed data within the limited range of multiple critical values to find the data that meets all the restrictive conditions. The screening result is used as input data for clustering.

[0084] Referring to FIG. 2, a flowchart of the K-means cluster analysis of the health status of the unit, when performing K-nearest neighbor cluster analysis on the data to be clustered after data mining and extraction in the relational database, it is necessary to set the number of clusters K and the maximum number of iterations n, and then randomly select K data points as the starting centroid. By calculating the distance from each data point to the centroid, the data points are assigned to the cluster with the smallest distance value, and the centroid of each cluster is updated repeatedly by the mean value. The clustering is ended until the cluster of data points shows no changes or the maximum number of iterations n is reached, and results are output. In the practical application process, technicians can set several categories according to the determined optimization characteristic parameters of the power plant units, define the characteristics of each category, classify the data after feature mining, and demarcate the stable and unstable status.

[0085] In a specific embodiment, in Step 5, the steady operating condition establishment module 8 is used to label the operating condition stability judgment parameters, and to establish the stable operating condition database. The specific process is as follows:

[0086] Complete the category labeling of the existing operating condition records of power plant units according to the definition of data status in cluster analysis, set the steady and unsteady category labels as 0 and 1 respectively, extract the stable operating condition, and establish the stable operating condition database. As shown in FIG. 3 (process of establishing the stable operating condition mode library of the unit), an operating condition contains controllable variables x, stability characterization variables y and category labels. For each operating condition, calculate the distance between the parameters in x and the existing operating conditions in the library. If the distance is zero, it is considered that the operating condition already exists in the library and will not be recorded again. Otherwise, add the time tag to the operating condition and store it in the stable operating condition library in the form of vector.

[0087] Referring to FIG. 3, the specific process of judging whether the operation of the unit is stable in Step 6 is as follows:

[0088] 1) In case of any abnormal parameters in the stability index, the program, if started, will search for the control target from the stability mode library, and return the point closest to the current status as the candidate operating condition.

[0089] 2) Compare the current status and the candidate operating condition, and count the parameters that need to be controlled when the current status is adjusted to the target, the control range and the number of parameters to be controlled. Determine a control target from the candidate operating conditions from the three dimensions. The control target shall be determined in such a way that ensures the minimum number of control parameters and the minimum control range.

[0090] 3) After determining the target, adjust the controllable variables according to the set adjustment range until the parameters reach the target values. During the control process, the change trend of the stable index will be monitored. If it does not return to normal, the control will be stopped at any time and the manual regulation will be started.

Embodiment 2

[0091] The system and method provided in Embodiment 1 of the present invention are put into operation in a combined cycle power plant, and the historical operation parameters of a generator unit in the plant in a year are selected for analysis and processing. When the interval value of the combined cycle power is high, the optimal value range of flue gas temperature at the inlet of waste heat boiler is [881.2431, 888.4293]K, and the optimal value range of steam temperature at the outlet of reheater is [831.3329, 838.2954]K. The parameter value of operation optimization is selected in the optimal interval. For convenience, the central value of the interval is taken as the optimal value in this embodiment. The obtained optimal value of flue gas temperature at the inlet of waste heat boiler is 884.2362K and the optimal value of steam temperature at the outlet of reheater is 834.8142K.

[0092] Following the above steps, the optimal values of the operation parameters of the power plant unit under specific operating conditions are obtained for all the valid association rules of data mining. Table 1 shows the comparison of the target values of some controllable operation parameters determined by the traditional method and the improved association rule method. In the table, 1-10 represent the waste heat boiler outlet flue gas temperature/K, waste heat boiler outlet flue gas pressure/Mpa, reheater outlet steam temperature/K, high pressure cylinder exhaust pressure/Mpa, high pressure cylinder exhaust temperature/K, low pressure cylinder inlet steam temperature/K, low pressure cylinder inlet steam pressure/Kpa and combustion chamber inlet flue gas temperature/K.

TABLE-US-00001 TABLE 1 Comparison of original and optimized values of unit operation parameters Original setting Optimized Parameter value value Increase/% Waste heat boiler outlet flue gas 345.0850 355.3850 3.000 temperature/K Waste heat boiler outlet flue gas  73.8000 104.0770 41.000  pressure/Mpa Reheater outlet steam 304.4000 434.8142 42.800  temperature/K High pressure cylinder exhaust 1970.3426  1985.2138  0.755 pressure/Mpa High pressure cylinder exhaust 615.4865 632.2643 2.730 temperature/K Low pressure cylinder inlet steam 398.5642 434.7157 9.070 temperature/K Low pressure cylinder inlet steam 2733.5624  2788.8509  2.020 pressure/Kpa Combustion chamber inlet flue gas 344.6541 357.8299 3.715 temperature/K

[0093] As can be seen from Table 1, the original parameters are generally lower than the optimized values based on the optimal parameter setting values obtained by the improved association rule mining method, which indicates that the parameter values are controlled within the safe range during the operation of power plant units to reduce the occurrence of accidents.

[0094] Each embodiment in this specification is described in a progressive manner, focusing on its differences from other embodiments, and the same and similar parts between embodiments can be referred to mutually. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and reference can be made to the description of the method section when needed.

[0095] The above description of the disclosed embodiments enables those skilled in the art to practice or use the present invention. Modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the essence or scope of the present invention. Accordingly, the present invention will not be limited to the embodiments described herein, but will cover the widest scope consistent with the principles and novel features disclosed herein.