Steam turbine valve nonlinear state quantitative method and system based on manifold theory

By constructing a fitted manifold based on manifold theory and performing mapping analysis, the problem of quantifying the nonlinear state of turbine valves was solved. This enabled accurate detection and quantitative evaluation of the valve's nonlinear state, reduced the false alarm rate, and improved the timeliness and accuracy of fault detection.

CN119025980APending Publication Date: 2024-11-26SHANGHAI ELECTRIC POWER GENERATION EQUIPMENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310594638.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-24
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Existing technologies struggle to detect and quantify the nonlinear state of turbine valves in a timely manner, resulting in high false alarm and false negative rates in fault detection. Furthermore, existing methods are not sufficiently applicable to complex systems and cannot accurately assess the impact of nonlinearity on control loops.

Method used

A manifold theory-based approach is adopted. By constructing a target fitting manifold, obtaining the exponential and logarithmic mapping, performing principal component analysis, calculating the valve's dead zone, hysteresis zone, and jump value, and combining high-order statistical analysis and Wiener filter processing of measurement data to eliminate noise interference, the nonlinear state of the valve is quantified.

Benefits of technology

It effectively reduces the complexity of valve nonlinear quantification, accurately estimates dead zone, hysteresis zone and jump value, reduces false alarm rate, and improves the timeliness and accuracy of fault detection, making it suitable for complex steam turbine systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025980A_ABST
    Figure CN119025980A_ABST
Patent Text Reader

Abstract

The invention provides a manifold theory-based steam turbine valve nonlinear state quantification method and system, and the method comprises the steps: constructing a target fitting manifold; solving the index and logarithm mapping of the target fitting confluence, and further obtaining the distribution of mapping points of the tangent space; performing principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and performing calculation based on the dimension reduction point data to obtain target data; and quantitative data corresponding to the steam turbine valve are obtained through calculation based on the target data, wherein the quantitative data comprise the sum of a dead zone and a lag zone corresponding to the steam turbine valve and a jump value. According to the steam turbine valve nonlinear state quantitative method and system based on the manifold theory, manifold theory knowledge is utilized, an optimal fitting flow shape is constructed, the index and logarithm mapping of the fitting flow shape are obtained, the marker quantity representing the valve nonlinear state intensity is defined, and the method and system can be used for valve curve correction and controller parameter modification.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of industrial fault detection and data processing, and in particular to a method and system for quantifying the nonlinear state of a steam turbine valve based on manifold theory. Background Art

[0002] The performance of valves in steam turbine systems is one of the key factors affecting the operation of steam turbines. Nonlinear phenomena such as hysteresis, dead zone, dead band, jump, etc. that occur during the operation of valves may cause process variable oscillation, resulting in deterioration of control loop performance, equipment shutdown, and production suspension, greatly reducing product quality and economic benefits. In severe cases, it may also cause casualties. At the same time, the process variable oscillation caused by valve nonlinearity is likely to be confused with other reasons, such as high-order integral action, slowly changing mean drift or external high-frequency interference, multi-loop interaction, etc., which also brings inconvenience to the diagnosis and detection of steam turbine valve abnormalities.

[0003] At present, steam turbine plants usually use regular inspection and troubleshooting to detect and maintain valves. This method requires that after a valve failure occurs, the operator goes to the site to maintain and replace the valve when the turbine is shut down or the reactor is stopped. This method cannot detect valve failures in a timely manner and consumes a lot of manpower and material resources. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a method and system for quantifying the nonlinear state of a steam turbine valve based on manifold theory, so as to solve the problem that there is a gap in the quantification of the nonlinear state of the valve in the prior art.

[0005] To achieve the above-mentioned and other related purposes, the present invention provides a method for quantifying the nonlinear state of a steam turbine valve based on manifold theory, the method comprising:

[0006] Construct target fitting manifold;

[0007] Obtaining the exponential and logarithmic mapping of the target fitting manifold, and then obtaining the distribution of the mapping points in the tangent space;

[0008] Performing principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and calculating target data based on the dimension reduction point data;

[0009] Quantitative data corresponding to the steam turbine valve is calculated based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and a jump value.

[0010] In one embodiment of the present invention, the method further comprises collecting measurement data of the steam turbine valve, wherein the measurement data comprises valve controller output data and variable data of the valve controlled process.

[0011] In one embodiment of the present invention, the method further includes performing nonlinear detection based on the valve data, specifically including:

[0012] Sampling based on the measurement data to obtain a set of oscillating discrete signal data;

[0013] Performing high-order statistical analysis based on the discrete signal data to obtain target data, wherein the target data includes non-Gaussian exponents and nonlinear exponents;

[0014] Based on the non-Gaussian index and the nonlinear index, it is determined whether the discrete signal data has nonlinear data, wherein:

[0015] When the non-Gaussian index is less than or equal to a first threshold and the nonlinear index is less than or equal to a second threshold, the discrete signal data does not contain nonlinear data; otherwise, the discrete signal data contains nonlinear data.

[0016] In one embodiment of the present invention, constructing a target fitting manifold specifically includes:

[0017] The target fitting manifold is constructed based on the measurement data using a preset calculation formula, and the calculation formula is as follows:

[0018]

[0019] Among them, x (k) represents the measurement data, m represents the dimension, a ij represents the first-order parameter, b j represents the second-order term parameter, c represents the constant term parameter, i and j are counting parameters, k=1,···,N;

[0020] Solve for the first-order parameter a ij , the second-order parameter b j And the constant term parameter c obtains the target fitting manifold, wherein the target fitting manifold is an ellipsoid.

[0021] In one embodiment of the present invention, obtaining the exponential and logarithmic mapping of the target fitting manifold to obtain the distribution of mapping points in the tangent space specifically includes:

[0022] Solving the exponential and logarithmic mapping of the target fitting manifold based on the center of curvature, the arc length on the ellipsoid surface, and the center of the curvature sphere;

[0023] Based on the logarithmic mapping of the ellipsoid surface, the corresponding mapping point distribution in the target tangent space is solved, and the target tangent space includes the Euclidean tangent space.

[0024] In one embodiment of the present invention, the principal component analysis is performed on the mapping points based on the tangent space to obtain the dimension reduction point data, and the target data is calculated based on the dimension reduction point data, specifically including:

[0025] Performing dimensionality reduction processing based on the mapping points to obtain the reduced-dimensionality point data, wherein the reduced-dimensionality point data is a two-dimensional principal component space point;

[0026] The target data is obtained by calculating the maximum square sum of the quantitative index variance based on the dimension reduction point data.

[0027] In one embodiment of the present invention, the quantitative data corresponding to the steam turbine valve is calculated based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and the jump value, specifically including:

[0028] The maximum value of the controller output is obtained based on the target data And the maximum value of the valve controlled process output

[0029] Based on the target data, the maximum output value of the i-th controller of the i-th closed loop is obtained. and the maximum value of the controlled process output of the ith valve in the ith closed loop

[0030] based on as well as The quantitative data are obtained by calculation.

[0031] To achieve the above-mentioned purpose and other related purposes, the present invention provides a steam turbine valve nonlinear state quantification system based on manifold theory, the system comprising:

[0032] A construction module, used to construct the target fitting manifold;

[0033] An obtaining module, used to obtain the exponential and logarithmic mapping of the target fitting manifold, and then obtain the distribution of the mapping points in the tangent space;

[0034] An analysis module, configured to perform principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and calculate target data based on the dimension reduction point data;

[0035] A calculation module is used to calculate the quantitative data corresponding to the steam turbine valve based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and the jump value.

[0036] To achieve the above-mentioned purpose and other related purposes, the present invention provides a computer-readable storage medium as mentioned above, on which a computer program is stored, and when the program is executed by a processor, the nonlinear state quantification method of a steam turbine valve based on manifold theory is implemented.

[0037] To achieve the above-mentioned purpose and other related purposes, the present invention provides an electronic device as mentioned above, which includes: a processor and a memory; wherein the memory is used to store a computer program, and the processor is used to load and execute the computer program so that the electronic device executes the quantitative method of nonlinear state of turbine valves based on manifold theory.

[0038] As described above, the method and system for quantifying the nonlinear state of a steam turbine valve based on manifold theory of the present invention have the following beneficial effects: first, the complexity of the existing valve viscosity quantification method in estimating the dead zone value plus the hysteresis zone value through iteration is reduced, and the quantitative index proposed by the present invention can directly estimate the size value of the dead zone plus the hysteresis zone through a mapping relationship; second, through a series of preliminary preparations such as preprocessing and selection of real-time valve measurement data, the adverse effects of other factors such as high-order integral action, slowly changing mean drift, external environment or other loop disturbances, and high-frequency noise on the quantification of valve nonlinearity are effectively eliminated; in addition, the present invention, through a series of operations such as manifold construction, calculation of exponential and logarithmic mapping, calculation of tangent space principal components, etc., well reveals the nonlinear relationship and underlying structure of the measurement data in the presence of valve nonlinearity. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the structure of a steam turbine valve control system in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0040] Figure 2 It is a schematic diagram of valve sticking in one embodiment of the nonlinear state quantitative method of a steam turbine valve based on manifold theory of the present invention;

[0041] Figure 3 It is a time sequence diagram of valve viscosity measurement variables in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0042] Figure 4 It is a diagram showing the steps of a valve nonlinear state detection method in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0043] Figure 5 It is a schematic diagram of the structure of a steam turbine power control system in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0044] Figure 6 It is a schematic diagram showing the timing of measurement data of a steam turbine valve nonlinear state quantitative method based on manifold theory in one embodiment of the present invention;

[0045] Figure 7 It is a schematic diagram showing the high-order statistical analysis results of a nonlinear state quantitative method for a steam turbine valve based on manifold theory in one embodiment of the present invention;

[0046] Figure 8 It is a schematic diagram of the steps of the valve nonlinear state quantification process in one embodiment of the steam turbine valve nonlinear state quantification method based on manifold theory of the present invention;

[0047] Fig. 9 It is a schematic diagram of the timing of the measurement data preprocessing in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0048] Fig.10 It is a schematic diagram showing the distribution of the measurement data before preprocessing in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0049] Fig.11 It is a schematic diagram showing the distribution of the measurement data after preprocessing in one embodiment of the nonlinear state quantitative method of a steam turbine valve based on manifold theory of the present invention;

[0050] Fig.12 It is a schematic diagram showing the autocorrelation function of a selected valve data segment in an embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0051] Fig.13 It is a schematic diagram of the optimal matching manifold of the selected valve data segment in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0052] Fig.14 It is a schematic diagram showing the distribution of tangent space data points in one embodiment of the steam turbine valve nonlinear state quantitative method based on manifold theory of the present invention;

[0053] Fig.15A It is a schematic diagram of the ellipsoid focus structure of a steam turbine valve nonlinear state quantitative method based on manifold theory in one embodiment of the present invention;

[0054] Fig. 15B It is a schematic diagram of the ellipsoid mapping structure of a nonlinear state quantitative method of a steam turbine valve based on manifold theory in one embodiment of the present invention;

[0055] Fig.16It is a schematic diagram of the overall system flow framework of a steam turbine valve nonlinear state quantification system based on manifold theory in one embodiment of the present invention.

[0056] Component number description

[0057] Steps S402 to S408

[0058] Steps S802 to S808

[0059] 1 Controller

[0060] 2 Controller

[0061] 3 Control valve

[0062] OP1 Controller 1 command signal

[0063] OP2 Controller 2 command signal

[0064] PV1 regulating stage output pressure signal

[0065] PV2 total power output signal

[0066] 160 Nonlinear State Quantification System of Steam Turbine Valve Based on Manifold Theory

[0067] 161 Construction Modules

[0068] 162 Request Module

[0069] 163 Analysis Module

[0070] 164 Computing Module DETAILED DESCRIPTION

[0071] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0072] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0073] Since catastrophic accidents caused by valve failures occur frequently, timely and accurate judgment of the real-time status of valves is of great significance to ensure the safety of industrial production. Therefore, early diagnosis and quantification of the nonlinearity in valves and the appropriate measures recommended by maintenance personnel are major issues that power plant operators and equipment manufacturers are currently facing and need to solve. The American Electric Power Research Institute (EPRI) held the first meeting on valve fault diagnosis at Duke Power's Marshall Steam Station in 2004, and the International Atomic Energy Association (IAEA) also provided corresponding guidance on valve failures in the nuclear power plant status inspection document released in 2007. Since then, academia and industry have conducted extensive research on valve fault diagnosis technology and have achieved many results.

[0074] In recent years, some research results have been achieved in the academic community on the modeling, detection and diagnosis, quantification and compensation of valve nonlinear states, especially for nonlinear state problems caused by valve static friction. However, the early detection and diagnosis of valve failures are mainly concentrated in industries such as petrochemicals and papermaking, and are still lacking in applications in power plant turbines and other fields. The realization of early detection and diagnosis of faults is exactly the requirement of relevant standards for power plants, and the standards also require that fault detection and diagnosis should be combined with the human factors engineering design of the main control room. From the perspective of human factors engineering, the detection and diagnosis of early signs of faults as early as possible can greatly improve the operator's perception of the power plant status and provide sufficient basis for their decision-making, thereby providing a basis for reducing the operator's workload and simplifying the human-machine interface design of the main control room.

[0075] The nonlinear state of valves can be divided into two categories. One is the static nonlinearity of valves caused by the valve mismatch curve. New steam turbines are equipped with valve correction curves in the control system to compensate for the nonlinearity introduced by the mechanical characteristics of the valves, making the control loop almost linear. However, during the operation of the steam turbine, as the valve usage time increases, the mechanical characteristic curve of the valve changes slowly. If the valve correction curve is not corrected in time, static deviation will be introduced, leading to undesirable conditions such as system oscillation. The other is the dynamic nonlinearity of valves caused by valve sticking at any stage of the valve stroke. The root cause is the high static friction generated by the mechanical system (including the electro-hydraulic system and the valve body), which exceeds the friction during smooth movement. Valve sticking will cause the control valve to get stuck in place until the controller continues to increase the valve set value to overcome the static friction. The detection technology of valve nonlinearity measures the state information of the valve during operation, processes and analyzes the measurement data, and combines the historical state information of the valve to predict the health status of the valve, and then determine the necessary valve maintenance strategy.

[0076] As an important part of the process system of a steam turbine plant, valves do not exist in isolation. Therefore, during the operation of the steam turbine, the operator not only needs to detect the nonlinearity of the valve as a single device, but more importantly, consider the changes in the nonlinear state of the valve and the impact on the entire control loop and even the final system power output. This requires obtaining precise quantitative indicators that represent the strength of the valve nonlinearity while detecting the nonlinear state of the valve, which is used to characterize the degree of influence of the nonlinear state of the valve in a certain period of time on the control loop and give the corresponding alarm threshold. However, although good results have been achieved in valve nonlinearity detection, the precise quantification of valve nonlinearity is still a difficult problem, and it is difficult to use one or several markers to characterize the influence of the valve nonlinear state on the control loop. So far, the existing valve viscosity quantification methods mainly focus on the quantification of dead bands and hysteresis zones and the situation of single variable control loops. However, since the valve viscosity process has different degrees of influence on the performance of the control loop, the quantification problem of valve viscosity has not actually been solved. On the other hand, the starting point of most previous valve nonlinearity detection methods is to assume that there is obvious oscillation in the process variable in the control loop. They rely too much on process data and ignore the acquisition of the internal mechanism model of the valve, which makes it difficult to distinguish failures caused by valves or other abnormal sources during the detection process, inevitably leading to an increase in the false alarm rate and missed alarm rate of detection and diagnosis.

[0077] From a methodological point of view, the quantification of valve stickiness mainly depends on the real-time operating data of the control quantity, the controlled process variable and the valve position. However, smart valves are not applicable in many safety-critical industrial applications (such as devices in the containment building of a nuclear power plant), where the valve position is not available, or for more general multivariable situations, such as cascade control loops, existing quantification methods may not be applicable. In fact, when the jump zone feature appears in the valve, the curve graph of the control quantity and the controlled variable may not appear as an ellipse, but roughly as a rugby ball, which will lead to the existing method to misestimate the valve stickiness feature. Therefore, being able to effectively and accurately detect and quantify the nonlinear state of each key valve in the turbine is one of the keys to ensure the smooth and safe operation of the turbine. Although the existing methods can detect and diagnose the nonlinear state in the valve, there is a gap in the quantification of the nonlinear state of the valve, and it is rarely used in actual production. This is also the motivation for this patent to propose a method for quantifying the nonlinear state of turbine valves based on manifold theory.

[0078] like Figure 1The figure shows the structure diagram of the control system corresponding to the steam turbine valve in this application, wherein the outer loop is the steam turbine speed / power control loop and the inner loop is the valve control loop. The outer loop sends the throttle opening command signal to the inner loop through the steam turbine speed / power controller to control the steam turbine speed and load, while the inner loop receives the throttle opening command signal sent by the outer loop, and uses the valve position signal measured on the oil motor side as feedback to realize the servo control of the valve position using the valve position controller.

[0079] When the valve is in a normal state (no nonlinearity), the valve opening command and the controlled variable are almost linearly related, that is, the size S of the dead zone and the hysteresis zone, and the jump value J are all approximately "0". As the system runs longer, the valve may be affected by factors such as steam quality pollution, rust on moving parts, and the piston rod and valve stem of the oil motor, which may cause the dynamic nonlinearity of the valve, that is, the phenomenon of valve sticking. Once nonlinearity exists in the valve, the command signal from the outer loop controller and the actual valve opening will show Figure 2 The parallelogram relationship shown in the figure shows that the size S of the dead zone and hysteresis zone and the jump value J will increase accordingly as the nonlinearity increases. Figure 1 When the control system shown in FIG. 1 is used, the valve opening command, the controlled variable (turbine power, speed) and other working fluid parameters (such as regulating stage pressure, steam inlet valve pressure) will experience abnormal oscillation fluctuations. The schematic diagram of abnormal fluctuations is shown in FIG. Figure 3 shown.

[0080] Therefore, in one embodiment of the invention, the quantitative method of the nonlinear state of a steam turbine valve based on manifold theory of the present invention includes performing nonlinear detection based on the valve data, specifically referring to Figure 4 , including the following steps:

[0081] Step S402, sampling based on the measurement data to obtain a set of oscillating discrete signal data;

[0082] Step S404, performing high-order statistical analysis based on the discrete signal data to obtain target data, wherein the target data includes a non-Gaussian index and a nonlinear index;

[0083] Step S406, judging whether the discrete signal data contains nonlinear data based on the non-Gaussian index and the nonlinear index;

[0084] Step S408: when the non-Gaussian index is less than or equal to a first threshold and the nonlinear index is less than or equal to a second threshold, the discrete signal data does not contain nonlinear data; otherwise, the discrete signal data contains nonlinear data.

[0085] It should be noted that the measurement data is applied when performing nonlinear detection, so the method also includes collecting measurement data of the steam turbine valve, and the measurement data includes valve controller output data and variable data of the valve controlled process.

[0086] Specifically, when collecting measurement data, since the overall system of the steam turbine is a complex nonlinear thermodynamic system, and the valve, as an important part of the steam turbine system, does not exist in isolation, during the operation of the steam turbine, the operator not only needs to detect the nonlinearity of the valve as a single device, but more importantly, consider the change of the nonlinear state of the valve, and the impact on the entire control loop and even the final system power output. Therefore, the collected measurement data should involve all measurements related to the valve input and output. Generally speaking, the controller output and the controlled process variable should appear in pairs as measured signals. In some cases, if the valve to be detected is not an intelligent valve, the valve stem position data cannot be easily obtained. In order to avoid this influence, this patent allows the measured signal to not contain the valve stem position data.

[0087] Further, refer to Figure 5 , which is a schematic diagram of the structure of a steam turbine power control system, specifically including the inner loop of controller 1, the outer loop of controller 2 and the control valve 3 acting on the regulating stage and the main body of the steam turbine. Controller 2 receives the final total power output value PV2 and generates controller 2 command signal OP2 as the input of controller 1. Controller 1 directly controls the movement of the valve and maintains the regulating stage output pressure signal PV1 to maintain stability. If there is no nonlinear phenomenon in the valve, the linear relationship between the controller command signal and the valve position ensures that the total power output value PV2 can be maintained near the set value. Therefore, the input of the valve is affected by the controller 1 command signal OP1 and the controller 2 command signal OP2, and the output of the valve affects the regulating stage output pressure value PV1 and the total power output signal PV2 that ensure the process variables. Therefore, in this example, the measurement data related to the steam turbine valve includes: controller 1 command signal OP1, controller 2 command signal OP2, regulating stage output pressure signal PV1 and total power output signal PV2, that is, the corresponding measurement data needs to include valve controller output data and variable data of the valve controlled process, and the valve stem position signal MV cannot be obtained in some cases, and is not used as valve-related measurement data in this patent.

[0088] Further, refer to Figure 6 , which is a timing diagram of measurement data related to the steam turbine valve in this embodiment, based on Figure 6The data in the data are used to estimate the nonlinear size of the sticky paper that may exist in the valve. Specifically, for a running steam turbine system, if the process output value of the inner loop where a valve is located is found to have a suspected oscillation phenomenon that is different from the normal operating state, the phenomenon may be caused by valve sticking, high-order integral action, slowly changing mean drift, external environment or other loop disturbances. To this end, the existing high-order statistical analysis method is used to detect whether the oscillation is caused by valve sticking. The specific steps are as follows:

[0089] (1) Assume that in the inner loop where the valve to be tested is located, a set of discrete process output signals with oscillation phenomenon obtained by sampling is: (sampling frequency is "1");

[0090] (2) The frequency domain representation of y1 after discrete Fourier transform is Y1(f), and the bispectrum B(f1, f2) in frequency domain is expressed as:

[0091]

[0092] Where E[Y1(f)] represents the mathematical expectation of Y1(f), and * represents the complex conjugate.

[0093] (3) In order to eliminate the undesirable properties of the estimated bispectral variance, the normalized bicoherence coefficient is obtained as:

[0094]

[0095] Among them, E[|Y1(f)| 2 ] represents the power of Y1(f), and Bic represents the bi-coherence coefficient.

[0096] (4) Calculate the non-Gaussian index NGI and the nonlinear index NLI, namely:

[0097]

[0098]

[0099] in, represents the mean value of the square of the bidirectional interference coefficient in the main frequency domain (0<f1<0.5, f2<f1 and 2f1+f2<1), is the maximum value of the square of the bidirectional interference coefficient, is the standard deviation of the square of the bidirectional coefficient, and Yes (among which is the value of the standard normal distribution when the confidence level is α, L is the total number of samples, and K is the number of dual frequencies in the main frequency domain of the dual spectrum).

[0100] Take the first threshold value as "0.001" and the second threshold value as "0.01". Under the premise of "95% confidence level", if NGI≤0.01 and NLI≤0.01, it means that the oscillation of the discrete signal output by the process may be caused by high-order integral action or interference from the external environment or other circuits, and there is no nonlinear phenomenon in the valve. Otherwise, it is considered that there is nonlinearity in the detected valve, that is, there is nonlinear data corresponding to the discrete signal data, where the non-Gaussianity index NGI (Non-gaussianity index) and the nonlinearity index NLI (Non-linear index).

[0101] against Figure 6 The measured data of the steam turbine valve shown, Figure 7 The results of high-order statistical analysis are shown. It is not difficult to see that the calculated high-order statistics - the bidirectional dry coefficients are not consistently zero, and many of them are far greater than zero, indicating the existence of viscous nonlinearity in the valve.

[0102] Furthermore, in one embodiment of the invention, after obtaining the measurement data, data analysis can be performed based on the measurement data, such as Figure 8 As shown in FIG. 1 , the quantitative method for the nonlinear state of a steam turbine valve based on manifold theory specifically includes the following steps:

[0103] Step S802, constructing a target fitting manifold;

[0104] Step S804, obtaining the exponential and logarithmic mapping of the target fitting manifold, and then obtaining the distribution of the mapping points in the tangent space;

[0105] Step S806, performing principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and calculating target data based on the dimension reduction point data;

[0106] Step S808, obtaining quantitative data corresponding to the steam turbine valve by calculation based on the target data, wherein the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and a jump value.

[0107] It should be noted that after detecting that a valve in the turbine system contains nonlinear phenomena, in order to eliminate the influence of the slowly changing mean drift noise on the subsequent quantification procedure, it is considered to use a Wiener filter with a specific boundary to preprocess the data so that the measurement data presents a more obvious ellipsoid shape or a shape obtained by the intersection of several ellipsoids in the sampling space; and the selection of valve data segments is also particularly important for the subsequent quantification steps. In fact, if there is nonlinearity in the valve, the real-time measurement data related to the valve will show more regular and stable oscillations, while the measurement data oscillations caused by other reasons will be irregular or produce mean drift. Therefore, it is particularly important to select a data segment with regular oscillations for the quantification process of the nonlinear state of the valve. Correspondingly, the measurement data needs to be preprocessed and the data selection steps are required.

[0108] The specific steps are as follows:

[0109] (1) Obtain the time series of real-time measurement data related to the valve to be analyzed (Time series length is N, dimension is m), which includes the controller output time u1 and process output time series y1 of the inner loop where the valve to be analyzed is located;

[0110] (2) normalizing the frequency represented in the frequency domain during the nonlinear state detection process of the valve, and obtaining the frequency (f1, f2) corresponding to the maximum bidirectional interference coefficient;

[0111] (3) Determine the boundary of the Wiener filter as: [ω L =max(0.004,f1-0.05),ω H =min(0.5,f2+0.05)];

[0112] The Wiener filter with the above boundary is used to filter the valve-related real-time measurement data to obtain the filtered data

[0113] Among them, for Figure 6 The measured data of the steam turbine valve shown in the figure uses high-order statistical analysis to detect the valve nonlinearity in the valve. Then, the Wiener filter is used to pre-process the valve measurement data. Fig. 9 The data time series diagram after preprocessing is shown. By observing, it can be seen that the data has obvious oscillation phenomenon. At the same time, the data distribution before and after preprocessing is as follows Fig.10 and Fig.11 As shown in the figure, it can be seen that the data of a steam turbine valve presents an obvious ellipsoid shape after preprocessing, which lays the foundation for the subsequent quantification of valve viscous nonlinearity.

[0114] Furthermore, the above embodiment shows that the selection of valve data segments is also particularly important for the subsequent quantification steps, so it is particularly important to select a data segment with regular oscillations for the quantification process of the valve nonlinear state. To this end, the regularity of whether the autocorrelation function passes through the zero point is used to analyze the process output value, as follows:

[0115] (1) Obtain the process output value of the inner loop where the valve to be tested is located after preprocessing

[0116] (2) Selecting a data segment length N (N < L), and dividing the process output value y1 of the inner loop where the valve is located into a number of data segments of length N;

[0117] (3) For each data segment of length N, the following calculation process is performed:

[0118] a) Calculate the autocorrelation function r corresponding to the data segment yy (k′), that is:

[0119]

[0120] Where y1(t) is the measured value of the inner loop process output of the valve at time t, y1 is the sample average of N samples, and k′ represents the lag number. The desired autocorrelation function r yy The value of (k') is between "-1" and "1".

[0121] b) For the autocorrelation function r yy (k′), (k′=1,2,…,N), calculate each r yy The time interval Δt between the values ​​of (k′) passing through the “0” point i Assuming that the number of times the "0" point is passed is n, the average period of oscillation is:

[0122]

[0123] c) Calculate the statistic r used to test whether the oscillation is regular:

[0124]

[0125] Among them, σT p Represents the time interval Δt i The standard deviation of .

[0126] d) Case 1: When the autocorrelation function r yy When the sample mean of (k′) is far from equal to “0”, it proves that the oscillation of the process output value of the inner loop where the valve is located may be caused by drift and transient;

[0127] Case 2: When the autocorrelation function r yy When the sample mean of (k′) is approximately equal to “0” and the value of the statistic r is less than or equal to “1”, it proves that the oscillation may be caused by high-frequency interference from the external environment or other circuits;

[0128] Case 3: When the autocorrelation function r yy When the sample mean of (k′) is approximately equal to “0” and the value of the statistic r is greater than “1”, it proves that the oscillation of the process output value of the inner loop where the valve is located is a regular oscillation (and the oscillation period is ), which may be caused by nonlinear phenomena inside the valve.

[0129] (4) In all cases where the autocorrelation function r yy (k′) is approximately equal to the data segment of “0”, and its corresponding data segment with the largest statistic r value is obtained. and its oscillation period

[0130] (5) Obtain and select the process output value data segment of the inner loop where the valve is located Other measurement data corresponding to the time series, to obtain the real-time measurement data related to the selected valve

[0131] Among them, for Figure 6 The measured data of the steam turbine valve shown, Fig.12 The figure shows the calculated autocorrelation function diagram of the selected valve data segment. The sample mean of the calculated autocorrelation function is "0.0128", which is very close to zero, indicating that the oscillation of the process output value of the inner loop where the valve is located is not caused by drift and transient. At the same time, the calculated average oscillation period is "24.774s", and the value of the statistic r "3.452" is greater than "1", proving that the oscillation presented by the selected data is a regular oscillation and is caused by the nonlinear phenomenon inside the valve.

[0132] Further, after detecting the presence of sticky behavior in the valve, noting the prior knowledge that the measured data presents a special pattern shaped like a standard ellipsoid or a shape formed by the intersection of multiple ellipsoids, this patent considers constructing a manifold that optimally matches the real-time process data in the following manner, and quantifying the nonlinearity of the valve using exponential and logarithmic mappings established on the fitted manifold.

[0133] The construction of the target fitting manifold is in the preprocessing stage, and the construction of the target fitting manifold specifically includes: constructing the target fitting manifold based on the measurement data using a preset calculation formula, and the correspondingly constructed target fitting manifold is an ellipsoid, and the calculation formula is as follows: Among them, x(k) represents the measurement data, m represents the dimension, a ij represents the first-order parameter, b j represents the second-order term parameter, c represents the constant term parameter, i and j are counting parameters, k = 1, ..., N; solve the first-order term parameter a ij , the second-order parameter b j And the constant term parameter c is used to obtain the target fitting manifold. The exponential and logarithmic mapping of the target fitting manifold is obtained to obtain the distribution of mapping points in the tangent space, specifically including: solving the exponential and logarithmic mapping of the target fitting manifold based on the center of curvature, the arc length on the ellipsoid surface, and the center of the curvature sphere; solving the corresponding mapping point distribution in the target tangent space based on the logarithmic mapping on the ellipsoid surface, and the target tangent space includes the Euclidean tangent space. The principal component analysis based on the mapping points in the tangent space is performed to obtain the reduced dimension point data, and the target data is calculated based on the reduced dimension point data, specifically including: performing dimensionality reduction processing based on the mapping points to obtain the reduced dimension point data, wherein the reduced dimension point data is a two-dimensional principal component space point; and calculating the maximum sum of squares of the quantitative index based on the reduced dimension point data to obtain the target data.

[0134] Further, the quantitative data corresponding to the steam turbine valve is calculated based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and the jump value, specifically including: obtaining the maximum value of the controller output based on the target data And the maximum value of the valve controlled process output Based on the target data, the maximum output value of the i-th controller of the i-th closed loop is obtained. and the maximum value of the controlled process output of the ith valve in the ith closed loop based on as well as The quantitative data are obtained by calculation.

[0135] In one embodiment of the invention, the specific steps of the embodiment of the present application are as follows:

[0136] (1) For the valve to be quantified, the relevant real-time measurement data is obtained after preprocessing through the quantification preparatory work mentioned above. (N is the number of time series, which is "1000" in this example, and m is the dimension of valve data, which is "4" in this example);

[0137] (2) Initialize the number of partition subsets κ = 1;

[0138] (3) For each partitioned subset, find the value that matches the data The optimal matching manifold is:

[0139]

[0140] Obtain the first-order term, second-order term and constant term parameter a of the ellipsoid ij ,b j and c.

[0141] (4) A = [a1, a2, ..., a m ], a i =[a i1 , a i2 , ..., a im ] T , and calculate all eigenvalues ​​of the matrix A.

[0142] (5) For the dataset For each given sample data For k = 1, ..., N, the value of the function ψ(·) is:

[0143]

[0144] And the test statistic is calculated from this:

[0145]

[0146] in, and They are the function values ​​ψ(x (k) )’s sampling mean and sampling variance.

[0147] (6) For each subset of ellipsoids, determine whether the following two points are satisfied at the same time:

[0148] a) All eigenvalues ​​of the matrix A are positive;

[0149] b) Where α is the given confidence level, t N-1 (α / 2) represents the 100(α / 2) percentile value of the t distribution with N-1 degrees of freedom, where N is the selected data segment length.

[0150] If the above two conditions are met at the same time, then proceed to step (7) of this implementation;

[0151] If both conditions cannot be met at the same time, the number of partition subsets is increased to κ = κ + 1, and the criteria for partition subsets are re-determined based on known prior knowledge, for example:

[0152] a) Based on whether a measured variable increases or decreases compared to the previous sampling instant;

[0153] b) According to the magnitude of the increase or decrease in the magnitude of a measured variable;

[0154] c) Based on whether a certain variable increases, decreases, fluctuates, etc. within a few specific sampling moments.

[0155] After repartitioning, return to step (3).

[0156] against Fig. 9 The timing diagram of the valve measurement data after preprocessing, using the above steps proposed in this patent, the optimal matching manifold is obtained as follows Fig.13 As shown (for the convenience of visualization, the results of the three-dimensional optimal matching manifold fitting are shown here). During the calculation process, when the subset into which the data is divided is "1", all the eigenvalues ​​of the quadratic coefficient matrix A of the ellipsoid equation are "15.322", "6.854", "3.903" and "2.775", respectively, which are all greater than the value of "0". In addition, under the premise of a confidence level of "90%", the test statistic "3.493" is greater than its statistical threshold "1.282". It can be seen that the optimal matching manifold fitted based on the standard ellipsoid equation effectively captures the data distribution characteristics when the valve has viscous nonlinearity.

[0157] Furthermore, the steps include:

[0158] (7) For each determined sample subset The manifold that best matches the real-time process data is obtained by intersecting the k-fitting ellipsoids obtained through the k-subsets.

[0159] (8) And according to the logarithmic mapping, the mapping point on the tangent space is obtained

[0160] (9) Perform principal component analysis on the mapping points on the tangent space to obtain the processed point V = [v (1) , v (2) , ..., v (L) ] T ;

[0161] Among them, for Fig.13 The optimal matching manifold is obtained, and the valve data is projected onto the tangent space through the logarithmic mapping on the ellipsoid surface. Then, the collected four-dimensional data is reduced to two dimensions by principal component analysis, and the distribution of the data points after dimensionality reduction is obtained as follows: Fig.14 Based on Fig.14 It can be seen that the obtained data points are roughly distributed around a straight line segment in the two-dimensional principal element space on the tangent space.

[0162] Furthermore, the steps include:

[0163] (10) Calculate the maximum square sum of the quantitative indicators D max To obtain the target data, the formula is as follows:

[0164]

[0165] Among them, Ω k is the kth cluster corresponding to the kth subset in the tangent space, is the number of principal components in tangent space.

[0166] In fact, this index represents the maximum distance between data points after principal component analysis in the tangent space. When this index is larger, it means that the valve data deviates more from the normal operating state, that is, the viscosity of the valve is greater.

[0167] against Fig.14 The data points obtained after dimensionality reduction on the tangent space, the quantitative index D in this embodiment max This calculates to "56.692".

[0168] (11) Assume that within the selected data segment, and They are the maximum values ​​of the controller output value of the inner loop where the valve is located and the output value of the valve controlled process. and are the maximum output of the ith controller of the ith closed loop and the maximum output of the ith valve controlled process of the ith closed loop, respectively, and are defined as:

[0169]

[0170] Then the size S of the dead zone plus the hysteresis zone can be estimated as:

[0171]

[0172] Among them, ρ1 is the minor axis length of the two-dimensional ellipse fitted by only using the controller output u1 and the process variable y1 of the inner loop where the valve is located, and α is the counterclockwise rotation angle of the ellipse relative to the controller output direction, which can be calculated based on the fitting result of the two-dimensional ellipse.

[0173] (12) If the process of the steam turbine unit can be approximated as a first-order or second-order plus time delay (FOPTD or SOPTD) link, the size of the jump value J can be estimated:

[0174]

[0175] Where T is the time constant of the dynamic component of the process transfer function, K c and τ i is the controller transfer function G p The proportional gain and integral time, ω is the oscillation frequency of the selected data segment, and S is the estimate of the deadband plus hysteresis given above.

[0176] At this point, by calculating the maximum square sum of the variances of the quantitative indicators, it can be estimated that the size S of the dead zone plus the hysteresis zone and the jump value are respectively "23.556%" (the displacement range of the turbine valve stem is 0-100%) and "1.614", while under normal circumstances, the size S of the dead zone plus the hysteresis zone and the jump value J are both close to "0", which shows that the turbine valve in this embodiment has serious viscous nonlinearity. Therefore, the operating personnel should adjust and calibrate the PID controller parameters in time, or inspect or maintain the turbine valve, where the PID controller is (Proportion Integration Differentiation. Proportional-Integral-Differentiation controller).

[0177] It should be noted that the above embodiment illustrates the method of obtaining the mapping points on the tangent space based on the logarithmic mapping. For valves with viscosity (dynamic nonlinearity), the valve command signal and the process data are roughly presented as an ellipsoidal manifold, and the ellipsoid is a generalization of the sphere. The "fuller" the ellipsoid is, the greater the degree to which the valve data deviates from the normal operating state, that is, the greater the viscosity of the valve. The difficulty studied in this patent is how to estimate the specific size of the nonlinear viscosity of the valve based on the ellipsoidal optimal matching manifold fitted by the valve data, and the calculation of the viscosity needs to be established in the traditional Euclidean space. The exponential and logarithmic mappings of the manifold are a pair of mutually reversible mappings, which establish a mapping relationship between the manifold and the Euclidean tangent space. Therefore, this patent generalizes the existing exponential and logarithmic mappings of the spherical surface for the first time, and solves the exponential and logarithmic mappings for the ellipsoidal surface by calculating the partial curvature center, the arc length on the ellipsoidal surface, and the center of the curvature sphere. Utilizing the distribution of Euclidean tangent space data points obtained by log-mapping of the ellipsoid surface, a quantitative index is established for quantitative estimation of valve nonlinear viscosity.

[0178] Specifically, it is assumed that the parametric equation of the m-dimensional ellipsoid can be expressed as follows:

[0179]

[0180] where ρ i , i = 1, 2, ..., m represents the semi-axis length of the ellipsoid, θ i , i=1, 2, ..., m-1 respectively represent the i Parameters, then for x i Each θ i Taking partial derivatives, we can find that for i=1, 2, ..., m-1, we have:

[0181]

[0182] And for i=m, we have:

[0183]

[0184] Further, the radius of curvature on the m-dimensional sphere can be expressed as:

[0185]

[0186] Assuming two points p and q on the m-dimensional ellipsoid, the arc length A between p and q can be defined as:

[0187]

[0188] where f i is the partial derivative of the point x between the arc lengths p and q, θ i,p and θ i,q (i=1, 2, ..., m-1) represent the parameters related to points p and q respectively, where reference Fig.15A and Fig. 15B According to the properties of the ellipsoid, given a point x on the n-dimensional ellipsoid and its associated semi-axis pair (ρ i , ρ j ), then the angle β between the focal points F1 and F2 and x can be expressed as:

[0189]

[0190] This angle is bisected by the line R connecting the point x and its associated center of curvature c, that is:

[0191]

[0192] where R is the radius of curvature given above, and is the corresponding semi-axis pair (ρ i , ρ j ) is the two-dimensional elliptical focus. Combining the above theory, similar to the method of obtaining the logarithmic mapping of a sphere, the logarithmic mapping with p as the base point q on the ellipsoid can be obtained, that is:

[0193]

[0194] where A is defined as the arc length between points p and q, is the distance between the center of curvature c of point p and point q, For vector Projection onto the line connecting point p and point c. Since the Riemann logarithmic map defines the maximum star domain centered on p The domain covers the tangent space All the projected points on the manifold, and on the manifold, the geodesic distance between every two points can be obtained by integrating the norm of the instantaneous velocity vector at each point along the geodesic. The value of this norm can be saved in the tangent space through logarithmic mapping. In particular, for the ellipsoid, the geodesic distance is equivalent to the arc length of the ellipsoid surface, that is, when we consider the ellipsoid ε n (center is 0), then the cutting trajectory of point x is the set of large ellipsoids corresponding to x and its antipodal point x = -x. The logarithmic mapping is obtained by rolling the ellipsoid into its tangent space, so that a large ellipsoid passing through x becomes a straight line. Then define the domain In tangent space, it can be represented by a low-dimensional ellipsoid, with the maximum semi-axis length equal to the arc length A. x x , which ensures that the data originally located on the ellipsoid surface is still within a boundary after being projected to the tangent space by logarithmic mapping. Then, through the nonlinear mapping of the manifold and the Euclidean tangent space given in this application, the measurement data of the steam turbine valve can be projected to the points on the manifold tangent space for further principal component analysis, thereby obtaining a representation method for low-dimensional data on the tangent space for estimating the size of the nonlinear viscosity of the valve. On the basis of obtaining the corresponding tangent space by using the ellipsoid logarithmic mapping, considering that the traditional principal component analysis method obtains the principal component coordinates with the help of the covariance matrix, this patent also adopts steps similar to the principal component analysis PCA (principle component analysis) in the Euclidean space and the singular value decomposition SVD (singular value decomposition) for each subset Ω k (k=1,2,...,κ) are analyzed and will not be elaborated here.

[0195] An embodiment of the present application also provides a steam turbine valve nonlinear state quantification system based on manifold theory, and the steam turbine valve nonlinear state quantification system based on manifold theory can implement the steam turbine valve nonlinear state quantification method based on manifold theory described in the present application, but the implementation device of the steam turbine valve nonlinear state quantification method based on manifold theory described in the present application includes but is not limited to the structure of the steam turbine valve nonlinear state quantification system based on manifold theory listed in the present embodiment. All structural deformations and replacements of the prior art made according to the principles of the present application are included in the protection scope of the present application.

[0196] See also Fig.16 In one embodiment, this embodiment provides a steam turbine valve nonlinear state quantification system 160 based on manifold theory, the system comprising:

[0197] A construction module 161, configured to construct a target fitting manifold, wherein the target fitting manifold includes an ellipsoid;

[0198] An obtaining module 162 is used to obtain the exponential and logarithmic mapping of the target fitting manifold, and then obtain the distribution of the mapping points in the tangent space;

[0199] An analysis module 163, configured to perform principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and calculate target data based on the dimension reduction point data;

[0200] The calculation module 164 is used to calculate the quantitative data corresponding to the steam turbine valve based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and the jump value.

[0201] Since the specific implementation of this embodiment corresponds to the aforementioned method embodiment, the same details will not be repeated here, and those skilled in the art should also understand that Fig.16 The division of the modules in the embodiment is merely a division of logical functions, and in actual implementation, all or part of them can be integrated into one or more physical entities, and these modules can be implemented entirely in the form of software called by processing elements, or entirely in the form of hardware, or some modules can be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware.

[0202] This embodiment provides an electronic device. Specifically, the electronic device includes at least: a memory and a processor connected via a bus, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to execute all or part of the steps in the aforementioned method embodiment.

[0203] In summary, the present invention has the following beneficial effects: firstly, the complexity of the existing valve viscosity quantification method in estimating the dead zone value plus the hysteresis zone value through iteration is reduced, and the quantitative index proposed in the present invention can directly estimate the size value of the dead zone plus the hysteresis zone through a mapping relationship; secondly, through a series of preliminary preparations such as preprocessing and selection of real-time valve measurement data, the adverse effects of other factors such as high-order integral action, slowly changing mean drift, external environment or other loop disturbances, and high-frequency noise on valve nonlinear quantification are effectively eliminated; in addition, the present invention well reveals the nonlinear relationship and underlying structure of the measurement data in the presence of valve nonlinearity through a series of operations such as manifold construction, calculation of exponential and logarithmic mapping, and calculation of tangent space principal components; and the present invention comprehensively considers all observed variables of complex loops such as cascade control in turbine systems, and solves the limitation that the existing valve viscosity quantification method is not suitable for cascade control loops.

[0204] In the several embodiments provided in the present application, it should be understood that the disclosed system, device or method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules / units is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules or units, which can be electrical, mechanical or other forms.

[0205] The modules / units described as separate components may or may not be physically separated, and the components displayed as modules / units may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules / units may be selected according to actual needs to achieve the purpose of the embodiments of the present application. For example, the functional modules / units in the various embodiments of the present application may be integrated into one processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into one module / unit.

[0206] Those of ordinary skill in the art should further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0207] The present application embodiment also provides a computer-readable storage medium. A person of ordinary skill in the art can understand that all or part of the steps in the method for implementing the above embodiment can be completed by a program to instruct the processor, and the program can be stored in a computer-readable storage medium, and the storage medium is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state hard disk, a magnetic tape, a floppy disk, an optical disc, and any combination thereof. The above storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a digital video disc (DVD)), or a semiconductor medium (e.g., a solid-state hard disk (SSD)), etc.

[0208] The present application embodiment may also provide a computer program product, the computer program product including one or more computer instructions. When the computer instructions are loaded and executed on a computing device, the process or function described in the embodiment of the present application is generated in whole or in part. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer or data center.

[0209] When the computer program product is executed by a computer, the computer executes the method described in the above method embodiment. The computer program product may be a software installation package, and when the above method is required, the computer program product may be downloaded and executed on a computer.

[0210] The descriptions of the processes or structures corresponding to the above-mentioned figures have different emphases. For parts that are not described in detail in a certain process or structure, please refer to the relevant descriptions of other processes or structures.

[0211] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A quantitative method for nonlinear state of steam turbine valve based on manifold theory, It is characterized in that include: Construct target fitting manifold; Obtaining the exponential and logarithmic mapping of the target fitting manifold, and then obtaining the distribution of the mapping points in the tangent space; Performing principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and calculating target data based on the dimension reduction point data; Quantitative data corresponding to the steam turbine valve is calculated based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and a jump value.

2. The nonlinear state quantitative method of steam turbine valve based on manifold theory according to claim 1, It is characterized in that The method further includes collecting measurement data of the steam turbine valve, wherein the measurement data includes valve controller output data and variable data of a valve-controlled process.

3. The nonlinear state quantitative method of steam turbine valve based on manifold theory according to claim 2, It is characterized in that The method further includes performing nonlinear detection based on the valve data, specifically comprising: Sampling based on the measurement data to obtain a set of oscillating discrete signal data; Performing high-order statistical analysis based on the discrete signal data to obtain target data, wherein the target data includes non-Gaussian exponents and nonlinear exponents; Based on the non-Gaussian index and the nonlinear index, it is determined whether the discrete signal data has nonlinear data, wherein: When the non-Gaussian index is less than or equal to a first threshold and the nonlinear index is less than or equal to a second threshold, the discrete signal data does not contain nonlinear data; otherwise, the discrete signal data contains nonlinear data.

4. The nonlinear state quantitative method of a steam turbine valve based on manifold theory according to claim 2 or claim 3, It is characterized in that The constructing target fitting manifold specifically includes: The target fitting manifold is constructed based on the measurement data using a preset calculation formula, and the calculation formula is as follows: Among them, x (k) represents the measurement data, m represents the dimension, a ij represents the first-order parameter, b j represents the second-order term parameter, c represents the constant term parameter, i and j are counting parameters, k=1,···,N; Solve for the first-order parameter a ij , the second-order parameter b j And the constant term parameter c obtains the target fitting manifold, wherein the target fitting manifold is an ellipsoid.

5. The nonlinear state quantitative method of steam turbine valve based on manifold theory according to claim 4, It is characterized in that The step of obtaining the exponential and logarithmic mapping of the target fitting manifold and then obtaining the distribution of mapping points in the tangent space specifically includes: Solving the exponential and logarithmic mapping of the target fitting manifold based on the center of curvature, the arc length on the ellipsoid surface, and the center of the curvature sphere; Based on the logarithmic mapping of the ellipsoid, the corresponding mapping point distribution in the target tangent space is solved, and the target tangent space includes the Euclidean tangent space.

6. The nonlinear state quantitative method of steam turbine valve based on manifold theory according to claim 5, It is characterized in that The step of performing principal component analysis on the mapping points based on the tangent space to obtain dimension reduction point data, and calculating target data based on the dimension reduction point data, specifically includes: Performing dimensionality reduction processing based on the mapping points to obtain the reduced-dimensionality point data, wherein the reduced-dimensionality point data is a two-dimensional principal component space point; The target data is obtained by calculating the maximum sum of squares of the quantitative indicators based on the dimension reduction point data.

7. The nonlinear state quantitative method of steam turbine valve based on manifold theory according to claim 6, It is characterized in that The quantitative data corresponding to the steam turbine valve is calculated based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and the jump value, specifically including: The maximum value of the controller output is obtained based on the target data And the maximum value of the valve controlled process output Based on the target data, the maximum output value of the i-th controller of the i-th closed loop is obtained. and the maximum value of the controlled process output of the ith valve in the ith closed loop based on as well as The quantitative data are obtained by calculation.

8. A nonlinear state quantitative system for steam turbine valves based on manifold theory. It is characterized in that include: A construction module, used for constructing a target fitting manifold, wherein the target fitting manifold includes an ellipsoid; An obtaining module, used to obtain the exponential and logarithmic mapping of the target fitting manifold, and then obtain the distribution of the mapping points in the tangent space; An analysis module, configured to perform principal component analysis based on the mapping points of the tangent space to obtain dimension reduction point data, and calculate target data based on the dimension reduction point data; A calculation module is used to calculate the quantitative data corresponding to the steam turbine valve based on the target data, and the quantitative data includes the sum of the dead zone and the hysteresis zone corresponding to the steam turbine valve and the jump value.

9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the nonlinear state quantitative method of a steam turbine valve based on manifold theory as described in any one of claims 1 to 7 is implemented.

10. An electronic device, It is characterized in that The electronic device comprises: a processor and a memory; wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the quantitative method for nonlinear state of a turbine valve based on manifold theory as described in any one of claims 1 to 7.