Identification method and system for nonlinear relationship between input and output of main steam regulating valve

Through the trend extraction algorithm and data segment fusion technology combined with plane segmentation algorithm and least squares method, the problem of large time consumption and estimation deviation of the nonlinear relationship identification of the main steam regulating valve in the thermal power unit is solved, and efficient and accurate flow characteristic identification is achieved, which improves the Unit control accuracy and safety.

CN113987057BActive Publication Date: 2024-11-05SHANDONG UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111272023.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-29
Publication Date
2024-11-05
Estimated Expiration
2041-10-29

AI Technical Summary

Technical Problem

The nonlinear relationship identification method for the input and output of the main steam regulating valve in the thermal power set has problems such as large time consumption, large estimation deviation and lack of theoretical explanation of the results, which affects the unit's control accuracy and safety.

Method used

Trend extraction algorithm and data segment fusion technology are used to obtain trend information of historical data, and the plane segmentation algorithm and least squares method are used to estimate the model parameters under different operating conditions, so as to identify the segmented nonlinear function curve of the valve.

Benefits of technology

It realizes efficient and rapid extraction of target information from a large amount of historical data, accurately obtains the actual flow characteristics of the regulating valve, and improves the precise control ability of the unit load.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113987057B_ABST
    Figure CN113987057B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for identifying the nonlinear relationship between the input and output of a main steam regulating valve, including: obtaining historical main steam regulating valve data and dividing it into different data segments; obtaining trend information of each data segment; fusing adjacent data segments in the same change trend; obtaining a steady-state value sample set and a change amplitude sample set corresponding to each variable in the original data; dividing the steady-state value sample set or the change amplitude sample set of the change amount distributed on different planes due to different working conditions into multiple groups; respectively estimating the model parameters corresponding to the steady-state value sample set or the change amplitude sample set of the change amount on different planes, thereby identifying the piecewise linear function curve of the valve. The present invention can efficiently and quickly extract the required target information from a large amount of historical data, and then identify the corresponding model parameters of the unit under different working conditions, and accurately obtain the actual flow characteristics of the regulating valve.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of main steam regulating valve control, and in particular to a method and system for identifying a nonlinear relationship between an input and an output of a main steam regulating valve. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In daily industrial production, thermal power units adjust the opening of the main steam regulating valve according to the grid instructions and pre-set control parameters, control the main steam flow entering the turbine and adjust the actual power generation, so that the power generation closely follows the changes in the grid load instructions. However, during the production process, the overhaul of the unit and daily wear and tear will cause the valve opening input of the regulating valve and the main steam flow output to have a nonlinear relationship, which is not conducive to the control of the turbine. The turbine valve flow characteristic curve is inconsistent with the actual flow characteristic curve, which will reduce the control accuracy of the unit and pose a threat to industrial production safety.

[0004] Therefore, in order to improve the controllability and safety of unit operation, the optimization of the flow characteristics of the turbine control valve is crucial for the stable operation of the power grid. In order to achieve accurate control of the unit load, the actual flow characteristic curve of the control valve must be obtained first. In recent years, the theoretical research on valve flow characteristics by scholars mainly includes: valve nonlinear identification method based on field tests, artificial intelligence algorithm based on neural network parameter identification, and Hammerstein model nonlinear identification method.

[0005] The main drawbacks of the existing methods are that it is difficult to identify the nonlinear characteristic curve of the valve with the help of field tests, which requires a lot of time for researchers and affects the normal operation of the unit; due to the need to identify global parameters, the nonlinear identification method based on the Murstein model will have a large estimation deviation when the unit operating conditions change; the neural network is a black box model, and the results obtained often have no specific theoretical explanation, and the prediction results are easily questioned. In addition, the massive historical data stored during the operation of the unit contains all the operating status information of the unit, but these historical data are often not effectively extracted. Summary of the invention

[0006] In order to solve the above problems, the present invention proposes a method and system for identifying the nonlinear relationship between the input and output of the main steam control valve, uses a trend extraction algorithm to extract rough trend information of each related variable, proposes a data segment fusion technology to obtain accurate trend information of historical data, and collects steady-state and transient sample data sets. The plane segmentation algorithm (RANSAC) is used to group the sample data sets belonging to different working conditions, and the model parameters corresponding to different groups of sample sets are estimated respectively by using the least squares method, so as to identify the model parameters of the piecewise nonlinear function curve of the valve.

[0007] In some embodiments, the following technical solutions are adopted:

[0008] A method for identifying a nonlinear relationship between input and output of a main steam regulating valve, comprising:

[0009] Obtain the historical main steam control valve data of a time series as the original data, divide the original data into different data segments; obtain the trend information of each data segment;

[0010] Merge adjacent data segments with the same change trend;

[0011] Obtain a sample set of steady-state values ​​and a sample set of amplitude of change corresponding to each variable in the original data;

[0012] The plane segmentation algorithm is used to divide the steady-state value sample sets or the change amplitude sample sets distributed on different planes due to different working conditions into multiple groups, and the scattered point set of each group corresponds to one working condition;

[0013] The least squares method is used to estimate the model parameters corresponding to the steady-state value sample sets or the change amplitude sample sets on different planes, so as to identify the piecewise linear function curve of the valve and obtain the nonlinear relationship between the input and output of the main steam control valve.

[0014] In other embodiments, the following technical solutions are adopted:

[0015] A main steam regulating valve input-output nonlinear relationship identification system, comprising:

[0016] The data segmentation module is used to obtain the historical main steam control valve data of a period of time as the original data, divide the original data into different data segments, and obtain the trend information of each data segment;

[0017] A data fusion module is used to fuse adjacent data segments with the same change trend;

[0018] The data segmentation module is used to obtain the steady-state value sample set and the change value sample set corresponding to each variable in the original data; the steady-state value sample set or the change amplitude sample set distributed on different planes due to different working conditions are divided into multiple groups by using the plane segmentation algorithm, and the scattered point set of each group corresponds to one working condition;

[0019] The valve parameter identification module is used to estimate the model parameters corresponding to the steady-state value sample sets or the change amplitude sample sets on different planes using the least squares method, so as to identify the piecewise linear function curve of the valve, obtain the nonlinear relationship of the main steam control valve, and then obtain the nonlinear relationship between the input and output of the main steam control valve.

[0020] In other embodiments, the following technical solutions are adopted:

[0021] A terminal device comprises a processor and a memory, wherein the processor is used to implement various instructions; the memory is used to store a plurality of instructions, wherein the instructions are suitable for being loaded by the processor and executing the above-mentioned main steam regulating valve input-output nonlinear relationship identification method.

[0022] In other embodiments, the following technical solutions are adopted:

[0023] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the above-mentioned main steam regulating valve input-output nonlinear relationship identification method.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] (1) The present invention adopts a piecewise linear representation method to extract trend information of time series, proposes a data segment fusion technology to locate steady-state and transient target data segments from a large amount of historical data, and establishes the sample data set required by the method.

[0026] Combined with the data mining algorithm, the present invention can efficiently and quickly extract the required target information from a large amount of historical data, and then identify the corresponding model parameters of the unit under different working conditions, accurately obtain the actual flow characteristics of the regulating valve, which is of great significance for achieving precise control of the unit load.

[0027] (2) The present invention proposes a new approach to identify the input-output nonlinear relationship characteristic curve of the regulating valve, using a plane segmentation algorithm to distinguish sample sets corresponding to different working conditions, and using a least squares algorithm to obtain the model parameters of each set of samples under the corresponding working conditions. Unlike other nonlinear identification methods, the flow curve finally identified by the proposed method has a large number of historical data segments as data support in different segments, and the accuracy of the identification result can be intuitively verified with the help of the trend change information of the data segment.

[0028] Other features and advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flow chart of a method for identifying the input-output nonlinear relationship of a main steam regulating valve based on special data segment mining in a specific implementation example of the present invention;

[0030] Figure 2 It is a schematic diagram of data segment fusion technology in a specific implementation example of the present invention;

[0031] Figure 3 It is part of the sampled data of the signal in the specific implementation example of the present invention;

[0032] Figure 4 It is a PLR segmentation result diagram of the sampling data in the specific implementation example of the present invention;

[0033] Figure 5 It is a trend fusion result diagram of sampling data in a specific implementation example of the present invention;

[0034] 6(a)-(d) are respectively sample data of steady-state section of industrial data in a specific implementation example of the present invention;

[0035] 7(a)-(d) are sample data of industrial data change segments in a specific implementation example of the present invention;

[0036] 8(a)-(f) are three-dimensional scatter plots of a sample data set of a steady state segment and a sample data set of a changing segment in a specific implementation example of the present invention;

[0037] Fig. 9 It is a result diagram of the nonlinear relationship characteristic curve of the valve identified by industrial data in a specific implementation example of the present invention;

[0038] Fig.10 It is a density diagram of the valve opening range of industrial data in a specific implementation example of the present invention. DETAILED DESCRIPTION

[0039] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0040] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0041] Embodiment 1

[0042] In one or more embodiments, a method for identifying a nonlinear relationship between a main steam control valve input and output is disclosed, referring to Figure 1 ,include:

[0043] (1) Obtaining historical main steam control valve data of a period of time as raw data;

[0044] Specifically, the original data includes:

[0045] Time series of actual power generation For example, select a suitable historical main steam control valve data as the original data, as shown in the attached Figure 3 As shown, the change pattern, noise level, working status, etc. in this data should be relatively comprehensive.

[0046] (2) Using the optimal segmentation number determination method based on confidence interval to obtain the optimal segmentation number, the original data is divided into different data segments; and the trend information of each data segment is obtained;

[0047] In this embodiment, combined with Figure 4 , the piecewise linear representation (PLR) method can convert the time series of the actual power P Divided into k short sequence segments Taking the kth data segment as an example, the data points of the kth data segment are fitted by the linear model shown in formula (1):

[0048]

[0049] Parameters of formula (1) and They represent the intercept and slope of the linear model respectively, and their parameter values ​​can be obtained by equations (2) and (3).

[0050]

[0051]

[0052] In the above formula is the average value of the amplitude of the sequence points of power P in the kth data segment, is the average value of the coordinates of the sequence points n.

[0053] Therefore, the time series of the actual power P Transformed into a sequence of linear models.

[0054] The optimal number of segments k is a prerequisite parameter, which is estimated based on the method for determining the optimal number of segments based on confidence intervals. The core idea of ​​the method for determining the optimal number of segments based on confidence intervals is that when fitting a time series into segments, the more segments there are, the smaller the noise variance of the linear fit and the narrower the confidence interval; on the other hand, the more segments there are, the fewer data points the trend segment contains, the larger the noise variance and the wider the confidence interval; therefore, a balance coefficient is constructed for each segment:

[0055]

[0056] Region A represents the confidence interval of the fitting result, region B represents the convex hull formed by the envelope of the data points in this segment, and A∩B represents the intersection of region A and region B. Then the weighted average of the balance coefficients of all trend segments shown in formula (5) is obtained:

[0057]

[0058] Among them, the physical meaning of η(k) is that when η(k) is maximized, the confidence interval can contain as many points as possible and make the fitting residual as small as possible, thereby determining the optimal number of segments.

[0059]

[0060] (3) Merge adjacent data segments with the same change trend;

[0061] In this embodiment, a significant amplitude change threshold is introduced to compare with the change amplitude of each trend segment, and the historical data is converted into a trend string consisting of 1, 0, and -1 characters. A trend segment fusion technology is proposed to re-divide the different trend change segments of the historical data.

[0062] Specifically, the trend segment fusion technology is to find the complete change process of the data segment with the same change trend through a series of operations such as character string segmentation and merging. is segmented and fitted into multiple straight line segments, where the starting point coordinates of the kth straight line segment and the end point coordinates are all known. Then the change amplitude of the kth straight line segment is:

[0063]

[0064] Similarly, the duration of the kth straight line segment is:

[0065]

[0066] Then, the amplitude significant change threshold A0 is introduced to compare with the change amplitude of the kth straight line segment to determine its increasing, decreasing, or flat change trend. Thus, the time series is converted into a string consisting of three primitives: 1, -1, and 0. The symbol allocation formula is shown in formula (9):

[0067]

[0068] The physical meaning of the amplitude significant change threshold A0 is whether the change amplitude of the data trend is significant relative to the noise change of the data sequence point. Based on the goodness of fit and noise variance, the threshold A0 can be obtained by formula (10):

[0069]

[0070] in, Indicates the goodness of fit, and its value is 0.8 here. represents the approximate noise variance. The calculation formula is as follows:

[0071]

[0072] The letter k represents the total number of segments, n k+1 -n k represents the number of sequence points in the kth segment, p(n) represents the true value of the data, is the approximate value after model fitting.

[0073] Next, we will introduce the algorithm flow of trend segment fusion technology. Assuming the actual power Converted as attached Figure 2 The primitive sequence shown is {1, 1, -1, 0, 0, 1, -1, 0, -1, 0, 1, 1}. Set a time threshold T th , if the duration of the 0 primitive in the kth segment is Δt k >T th , then the trend of the data segment is considered to be flat, otherwise this segment is too short and this segment is attributed to an indeterminate trend segment. Figure 2 The 4th and 5th segments are flat trend segments, and the 8th and 10th segments are uncertain trend segments. The time series is split at the flat trend segments, and the string is converted into: {1, 1, -1}, {0, 0}, {1, -1, 0, -1, 0, 1, 1}. {0, 0} is judged as a flat trend, and the remaining two segments {1, 1, -1}, {1, -1, 0, -1, 0, 1, 1} are transferred to the next step. Remove the '0' primitive with short change time and unclear trend in the previous step primitive sequence {1, -1, 0, -1, 0, 1, 1}, that is, segments 8 and 10.

[0074] The string continues to be converted into subsequence segments: {1, 1, -1}, {0, 0}, {1, -1, -1, 1, 1}.

[0075] Find the turning point where the trend suddenly changes in the previous primitive sequence, and segment out the continuous and monotonically changing data segments. For example, if the first two segments of the string {1, 1, -1} have an increasing trend and the third segment has a decreasing trend, the criterion for the segmentation point is:

[0076] abs|(1+1-1)| <abs|(1+1)| (12)

[0077] The criteria for the remaining segmentation points are similar, and the string is divided into 6 groups of data segments: {1, 1}, {-1}, {0, 0}, {1}, {-1, -1}, {1, 1}. The 0 primitive in the middle of the same changing trend is added and merged with the previous and next segments into one segment, such as adding the 8th segment in segments 7, 8, and 9. The 0 primitive in the middle of the different trends shown in segment 10 is further divided into the fourth trend: uncertain trend. The primitive sequence is finally divided into seven groups of data segments corresponding to different trends: {1, 1}, {-1}, {0, 0}, {1}, {-1, 0, -1}, {0}, {1, 1}. In this way, the data trend is reintegrated so as to find the complete target data segment in different trends. The final result is shown in the attached figure. Figure 5 shown.

[0078] (4) Obtain a sample set of steady-state values ​​and a sample set of changes corresponding to each variable in the original data;

[0079] In this embodiment, the character v is used to represent a variable. It is assumed that a control process involves a total of r variables, and the variable v r Finally, it was divided into m r If the character S is used to represent the trend character (1, 0, -1, uncertain trend) of each segment of data, then the different trend character combinations of all variables can be expressed by formula (13):

[0080]

[0081] If the i-th trend of variable v1, the j-th trend of variable v2, and the r The trend of the hth segment is flat:

[0082]

[0083] And the trend S i , S j ,…,S h There is overlap in time, and the time intersection part is [n start ,n end ]:

[0084]

[0085] The data segment within this interval is the variable v1, v2, ..., v r At the same time, if the letter d is used to represent the mean of each target data segment in a steady state, then the variables v1, v2, …, v r A set of steady-state mean samples is {d i (v1),d j (v2),…,d h (v r )},d h (v r ) is solved by formula (16):

[0086]

[0087] If the i-th trend of variable v1, the j-th trend of variable v2, the variable v r The trends of the hth segment have changed significantly:

[0088]

[0089] And the trend S i , S j , S h There is overlap in time, and the starting point n of the time intersection part start and end point n end It is given in formula (15), the interval [n start ,n end ] are variables v1, v2, …, v r Data segments that have significant changes at the same time. r Represents the variable v r The amplitude change of variables v1, v2, …, v r A set of amplitude variation samples is {Δv1, Δv2,…, Δv r}, Δv r The solution formula is shown in formula (18):

[0090]

[0091] Thus we get variables v1, v2, ..., v r The data segments that are both in steady state and dynamic state are further obtained by obtaining the steady state value sample set ∑{d i (v1),d j (v2),…,d h (v r )} and the sample set of variation ∑{Δv1,Δv2,…,Δv r}, as shown in Figures 6(a)-(d) and 7(a)-(d).

[0092] (5) Using a plane segmentation algorithm, the steady-state value sample sets or the change amplitude sample sets distributed on different planes due to different working conditions are divided into multiple groups, and each group of scattered point sets corresponds to one working condition;

[0093] In this embodiment, a plane segmentation algorithm (RANSAC) is introduced to divide the sample data sets corresponding to different working conditions and group the sample sets distributed in different planes. In three-dimensional space, the three-dimensional scattered points of the steady-state value sample set or the change amplitude sample set of the change amount will be distributed on different planes, as shown in Figures 8(a)-(f). We need to use the plane segmentation algorithm (RANSAC) to separate the scattered points w distributed on different planes and extract multiple scattered point sets {w1,w2,…w q}. Where q represents different planes. When q is a fixed value, the scattered point set w q The scattered points in correspond to the same working condition, and these scattered points can be used to estimate the model parameters under this operating condition. The calculation process of the required RANSAC plane segmentation algorithm is as follows:

[0094] (1) Repeatedly randomly select three points to form a plane.

[0095] (2) Select the plane containing the most points as the first target plane;

[0096] (3) Then discard the points in the known target plane in step 2 from the sample space.

[0097] (4) Repeat process (1)-(3) to find all target planes until the number of sample space points is less than the threshold N. th , the algorithm stops.

[0098] Thus, the steady-state value sample sets or the change amplitude sample sets distributed on different working planes will be separated. If the steady-state value sample set or the change amplitude sample set has q planes, it corresponds to q groups of scattered point sets and q working conditions.

[0099] (6) The least squares method is used to estimate the model parameters corresponding to the steady-state value sample sets or the change amplitude sample sets on different planes, so as to identify the piecewise linear function curve of the valve and further analyze the nonlinear relationship between the input and output of the main steam control valve.

[0100] Specifically, the slope and intercept parameters of each segment of the piecewise function curve and the valve opening range corresponding to each segment are solved; the model used in the present invention involves three variables, namely, valve opening u1, main steam pressure u2, and actual power p.

[0101] 1) Assuming that the nonlinear relationship characteristic of the valve is a piecewise linear curve, the valve opening u1 and the main steam flow x can be obtained by combining the theoretical mechanism. LL The steady-state value relationship formula between:

[0102] x LLss =K F ·u 1ss +b (19)

[0103] In formula (19), the character b represents the intercept of one segment of the piecewise linear function, and the intercept b has different values ​​for different segments.

[0104] K F = valve opening u1 to main steam flow x LL The main steam flow rate x can also be deduced from the theoretical mechanism LL , the steady-state relationship formula between the main steam pressure u2 and the actual power p:

[0105] p ss =x LLss ·K G1 +(u 2ss -c0)·K G2 (20)

[0106] K in formula (20) G1 Represents the main steam flow x LL The gain to the actual power p, K G2 represents the gain from the main steam pressure u2 to the actual power p, and c0 represents the initial value of the main steam pressure. Substituting formula (19) into (20) can obtain the steady-state relationship formula between the valve opening u1, the main steam pressure u2 and the actual power p:

[0107] p ss =K F K G1 ·u 1ss +K G2 ·u 2ss +b·K G1 -c0·K G2 (twenty one)

[0108] When multiple groups of steady-state value sample sets ∑{d i (u1),d j (u2),d h When (p)} is known, substituting a set of samples corresponding to the same working condition into the linear equation shown in formula (22) can obtain the gain and intercept parameters of the corresponding model:

[0109]

[0110] Combining the least squares algorithm with formula (23) can solve the value of parameter θ:

[0111]

[0112] Gain of the model and constant value can be solved. c0 is the initial value of the main steam pressure, which is a known constant. The intercept It can also be further calculated that, at this point, the slope and intercept of each segment of the valve piecewise function have been calculated.

[0113] 2) Based on the theoretical mechanism, the relationship between the change amount of variables u1, u2, and p shown in formula (24) can be deduced:

[0114]

[0115] When the sample set of the change amplitudes of variables u1, u2, and p is known, the gain of the model can be obtained by substituting the sample sets of different groups into the linear equation shown in formula (25) in combination with the least squares algorithm.

[0116]

[0117] Formula (26) is the formula for solving the model gain K using the least squares algorithm.

[0118]

[0119] The variation amplitude sample set can only be used to calculate the slope and intercept of the piecewise function. However, after the variation amplitude sample set of the variation corresponding to a certain segment of the piecewise function is known, the valve opening range corresponding to the piecewise function can be depicted according to the data variation process of the valve opening u1. The final result is shown in the attached figure. Fig. 9 and attached Fig.10 shown.

[0120] The piecewise linear representation (PLR) method is used to divide the different trend segments of historical data, and the optimal segment number determination method based on confidence interval is used to obtain the segment number of historical data; a trend segment fusion technology is proposed to reintegrate the preliminary trend information obtained by the PLR ​​method, and further merge trend segments with the same trend and different change rates to obtain complete trend information; extract the data intervals in which each relevant variable is in a steady state or dynamic state at the same time, calculate their mean and change amplitude respectively, and collect the steady-state value and amplitude change amplitude sample set; use the plane segmentation algorithm to distinguish the steady-state value and amplitude change amplitude sample set distributed on different planes due to different working conditions; use the least squares algorithm to estimate the model parameters corresponding to the sample sets on different planes respectively, and effectively identify the piecewise function curve of the valve;

[0121] The embodiments of the present invention can efficiently and quickly extract the required target information from a large amount of historical data, establish a sample data set, and then identify the model parameters corresponding to the unit under different working conditions, accurately obtain the actual flow characteristics of the regulating valve, which is of great significance for achieving precise control of the unit load.

[0122] Embodiment 2

[0123] In one or more embodiments, a main steam control valve input-output nonlinear relationship identification system is disclosed, comprising:

[0124] The data segmentation module is used to obtain historical data of a time series as raw data, divide the raw data into different data segments, and obtain trend information of each data segment;

[0125] A data fusion module is used to fuse adjacent data segments with the same change trend;

[0126] The data segmentation module is used to obtain the steady-state value sample set and the change value sample set corresponding to each variable in the original data; the steady-state value sample set or the change amplitude sample set distributed on different planes due to different working conditions are divided into multiple groups by using the plane segmentation algorithm, and the scattered point set of each group corresponds to one working condition;

[0127] The valve parameter identification module is used to estimate the model parameters corresponding to the steady-state value sample set or the change amplitude sample set on different planes using the least squares method, so as to identify the piecewise linear function curve of the valve and obtain the nonlinear relationship between the input and output of the main steam control valve.

[0128] It should be noted that the specific implementation methods of the above modules have been described in Example 1 and will not be described in detail here.

[0129] Embodiment 3

[0130] In one or more embodiments, a terminal device is disclosed, including a server, the server including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the main steam control valve input-output nonlinear relationship identification method in Example 1 when executing the program. For the sake of brevity, it will not be described in detail here.

[0131] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0132] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0133] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software.

[0134] Embodiment 4

[0135] In one or more embodiments, a computer-readable storage medium is disclosed, in which a plurality of instructions are stored, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the main steam control valve input-output nonlinear relationship identification method described in Example 1.

[0136] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for identifying the nonlinear relationship between input and output of a main steam control valve, characterized in that: include: Obtain the historical main steam control valve data of a period of time as the original data, and divide the original data into different data segments; Get trend information for each data segment; Merge adjacent data segments with the same change trend; The fusing of adjacent data segments in the same change trend specifically includes: Compare the change amplitude of each data segment with the set amplitude significant change threshold to determine the change trend of the data segment; wherein the change trend includes increase, decrease or flatness, and each change trend corresponds to a primitive, thereby converting each data segment into a primitive representation, and the entire data segment sequence is converted into a string composed of primitives; Determine whether the duration of the data segment with a flat trend exceeds a set time threshold. If so, the trend of the data segment is flat; otherwise, the data segment is an uncertain trend segment. Segment the data segments with flat trends and remove the segments with uncertain trends. For the remaining data segment sequences, segment the data segments with continuous and monotonically changing trends, and finally obtain multiple groups of data segments corresponding to different trends. Obtain a sample set of steady-state values ​​and a sample set of amplitude of change corresponding to each variable in the original data; The plane segmentation algorithm is used to divide the steady-state value sample sets or the change amplitude sample sets distributed on different planes due to different working conditions into multiple groups, and the scattered point set of each group corresponds to one working condition; The plane segmentation algorithm is used to divide the steady-state value sample sets or the change amplitude sample sets distributed on different planes due to different working conditions into multiple groups, including: In the three-dimensional space sample, three points are randomly selected to form a plane; the plane containing the most points is selected as the first target plane; the points in the known target plane are discarded from the sample space; Repeat the above process until the number of point sets in the sample space is less than the set threshold, and the algorithm stops, thus finding all target planes; The least squares method is used to estimate the model parameters corresponding to the steady-state value sample sets or the change amplitude sample sets on different planes, so as to identify the piecewise linear function curve of the valve and obtain the nonlinear relationship between the input and output of the main steam control valve.

2. A main steam control valve input-output nonlinear relationship identification method as claimed in claim 1, characterized in that: The original data is divided into different data segments, including: The optimal number of segments of the original data is obtained by using the optimal number of segments determination method based on confidence interval; Based on the optimal number of segments, the data is divided using a bottom-up PLR segmentation method to obtain different data segments.

3. A main steam control valve input-output nonlinear relationship identification method as claimed in claim 1, characterized in that: Obtain the steady-state value sample set and the change amplitude sample set corresponding to each variable in the original data, including: The character v represents a variable. Assume that a control process involves a total of r variables, and the variable v r Finally, it was divided into m r part; If the i-th trend of variable v1, the j-th trend of variable v2, ..., the variable v r The trends of the hth segment are all flat, and the trends overlap in a certain time interval. The data segment in this time interval is the variable v1, v2, [, v r The target data segment is in a steady state at the same time; If the i-th trend of variable v1, the j-th trend of variable v2, ..., the variable v r The trends of the hth segments of have all changed significantly, and the trends overlap in a certain time interval. The data segments in this time interval are variables v1, v2, …, v r Target data segments that have undergone significant changes; Further, a steady-state value sample set and a change amplitude sample set of the change amount corresponding to each variable are obtained.

4. A main steam control valve input-output nonlinear relationship identification method as claimed in claim 1, characterized in that: The least squares method is used to estimate the model parameters corresponding to the steady-state value sample sets on different planes, including: Based on the steady-state value relationship between valve opening and main steam flow, and the steady-state value relationship between main steam flow, main steam pressure and actual power, the steady-state value relationship between valve opening, main steam pressure and actual power is obtained, and the steady-state value models of piecewise linear equations under different working conditions are established respectively; The model is solved using the least square method to obtain the gain, constant value, and slope and intercept of each segment of the model.

5. A method for identifying the nonlinear relationship between input and output of a main steam regulating valve according to claim 1, characterized in that: The least squares method is used to estimate the model parameters corresponding to the change amplitude sample sets of the change quantities on different planes, including: Obtain the variation relationship between valve opening, main steam pressure and actual power, and establish the variation model of piecewise linear equations under different working conditions; The model is solved by the least square method to obtain the slope and intercept of the model, and the valve opening range corresponding to the model is depicted according to the data change process of the valve opening.

6. A main steam control valve input-output nonlinear relationship identification system, characterized in that: include: A data segmentation module is used to obtain a period of historical main steam control valve data as raw data and divide the raw data into different data segments; Get trend information for each data segment; A data fusion module is used to fuse adjacent data segments with the same change trend; The fusing of adjacent data segments in the same change trend specifically includes: Compare the change amplitude of each data segment with the set amplitude significant change threshold to determine the change trend of the data segment; wherein the change trend includes increase, decrease or flatness, and each change trend corresponds to a primitive, thereby converting each data segment into a primitive representation, and the entire data segment sequence is converted into a string composed of primitives; Determine whether the duration of the data segment with a flat trend exceeds a set time threshold. If so, the trend of the data segment is flat; otherwise, the data segment is an uncertain trend segment. Segment the data segments with flat trends and remove the segments with uncertain trends. For the remaining data segment sequences, segment the data segments with continuous and monotonically changing trends, and finally obtain multiple groups of data segments corresponding to different trends. The data segmentation module is used to obtain the steady-state value sample set and the change value sample set corresponding to each variable in the original data; the steady-state value sample set or the change amplitude sample set distributed on different planes due to different working conditions are divided into multiple groups by using the plane segmentation algorithm, and the scattered point set of each group corresponds to one working condition; The plane segmentation algorithm is used to divide the steady-state value sample sets or the change amplitude sample sets distributed on different planes due to different working conditions into multiple groups, including: In the three-dimensional space sample, three points are randomly selected to form a plane; the plane containing the most points is selected as the first target plane; the points in the known target plane are discarded from the sample space; Repeat the above process until the number of point sets in the sample space is less than the set threshold, and the algorithm stops, thus finding all target planes; The valve parameter identification module is used to estimate the model parameters corresponding to the steady-state value sample sets or the change amplitude sample sets on different planes using the least squares method, so as to identify the piecewise linear function curve of the valve, obtain the nonlinear relationship of the main steam control valve, and then obtain the nonlinear relationship between the input and output of the main steam control valve.

7. A terminal device, comprising a processor and a memory, wherein the processor is used to implement each instruction; and the memory is used to store multiple instructions, characterized in that: The instruction is suitable for being loaded by a processor and executing the main steam control valve input-output nonlinear relationship identification method described in any one of claims 1-5.

8. A computer-readable storage medium storing a plurality of instructions, characterized in that: The instruction is suitable for being loaded by a processor of a terminal device and executing the main steam control valve input-output nonlinear relationship identification method described in any one of claims 1-5.