A method for detecting and locating faults of steam turbine valves

Through the method based on flow characteristics and opening characteristics identification, the steam turbine valve fault detection is used to use historical and real-time data, and the problem of failure in the existing technology cannot be accurately judged in real time, and the accurate monitoring and early warning of high-pressure regulating valve faults is achieved, thereby improving the The sensitivity and accuracy of the detection are achieved.

CN115773158BActive Publication Date: 2024-11-19GUODIAN DAWUKOU THERMAL POWER CO LTD +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211584702.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-12-09
Publication Date
2024-11-19
Estimated Expiration
2042-12-09

AI Technical Summary

Technical Problem

The prior art cannot accurately judge and early warning of valve fault types through real-time data when the turbine is frequently peak-to-frequency regulation, and the traditional methods have low operability for valve fault detection.

Method used

Using a method based on flow characteristics and opening characteristics identification, fault characteristic parameters are obtained by collecting historical data, fault detection and positioning are carried out in combination with real-time data, and curve fitting and adaptive compression are used to eliminate interference information. Realize accurate judgment of fault characteristic parameters.

Benefits of technology

Accurate and accurate judgment and early warning of the fault of the high-pressure regulating valve of the turbine are achieved, which improves the sensitivity and accuracy of fault detection, and avoids the specialized test dependence of traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115773158B_ABST
    Figure CN115773158B_ABST
Patent Text Reader

Abstract

A method for detecting and locating a turbine valve fault belongs to the technical field of turbine fault detection. The specific scheme includes the following steps: Step 1: Collect historical data related to regulating valves under normal operating conditions of the unit; Step 2: According to the defined fault parameter calculation method, obtain the valve flow characteristic curve and each valve opening characteristic curve based on the collected historical data, and determine the normal value and variation range of the fault characteristic parameter; Step 3: Calculate the fault characteristic parameter based on the real-time data actually collected; Step 4: Compare the fault characteristic parameters of the real-time data with the fault characteristic parameters under normal conditions to determine whether the valve has a fault, what kind of fault has occurred, and the location of the fault. Compared with the traditional method, the present invention can realize real-time and accurate monitoring, eliminating the disadvantage that the traditional method can only perform special opening and closing tests on the valve.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of steam turbine fault detection, and in particular relates to a steam turbine valve fault detection and positioning method. Background Art

[0002] Steam turbine regulating valves work in a high temperature and high pressure environment, and the thermal expansion of the valves has high requirements for valve clearance. However, the frequent adjustment of the working conditions under peak and frequency regulation conditions leads to frequent operation of the regulating valve group and increased wear, which will inevitably lead to problems such as valve component material wear, coating shedding, valve stem packing peeling, valve core deviation, etc., which will lead to abnormal valve operation or contamination of EH oil quality, and eventually cause valve hardware failure. In the long run, it is easy to develop into a fault and eventually affect the safe and stable operation of the steam turbine unit. Therefore, it is of great practical significance to conduct fault monitoring and early warning research on steam turbine regulating valves.

[0003] Previous valve failure early warning methods were all based on special test data, which also required high practical operability of the methods. At the same time, facing the need to ensure the continuous operation of the unit in the future deep peak and frequency regulation scenarios, the original methods are no longer fully applicable, and it is impossible to use real-time data for accurate performance evaluation and early warning of performance degradation failures.

[0004] For the regulating valve group, the information of the measuring points that can be collected is relatively comprehensive, including the steam status before and after the valve, valve opening instructions and feedback, etc. The numerous parameters contain relatively rich information. However, the frequent changes in working conditions caused by peak and frequency regulation lead to certain interference in the measured parameters. How to extract parameters that can characterize the working status of the valve from the numerous parameters containing interference, and perform fault detection and location based on the characterization parameters is the main difficulty in the research. Summary of the invention

[0005] In order to solve the problem that the methods of valve fault detection and positioning using special tests in the prior art cannot be fully applicable to the frequent peak and frequency regulation operation of thermal power units, resulting in the inability to accurately judge the valve fault type in real time and issue early warnings for the faults, the present invention provides a turbine valve fault detection and positioning method based on flow characteristic and opening characteristic identification.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for detecting and locating a steam turbine valve fault comprises the following steps:

[0008] Step 1: Collect historical data related to the regulating valve under normal operating conditions of the unit;

[0009] Step 2: According to the defined fault characteristic parameter calculation method, the valve flow characteristic curve and each valve opening characteristic curve are obtained based on the collected historical data, and the normal value and variation range of the fault characteristic parameters are determined; the fault characteristic parameters include flow characteristic linearity, flow characteristic discreteness and valve opening characteristic discreteness; the slope of the tangent line of the flow characteristic curve at a certain point is used as the flow characteristic linearity index; the mean square error is used to evaluate the flow characteristic discreteness compared with the discreteness of the flow characteristic curve by the actual data; the mean square error is used to evaluate the valve opening characteristic discreteness compared with the discreteness of the valve opening characteristic curve by the actual data;

[0010] Step 3: Based on the real-time data actually collected, the fault characteristic parameters are calculated according to the calculation method of step 2;

[0011] Step 4: Compare the fault characteristic parameters of the real-time data with the fault characteristic parameters under normal working conditions to determine whether the valve has a fault, what kind of fault has occurred, and the location of the fault.

[0012] Furthermore, in step one, the relevant historical data collected include main steam pressure, regulating stage pressure, valve opening feedback value and comprehensive flow command.

[0013] Furthermore, in step one, the historical data collected needs to satisfy the unit load to cover all operating conditions.

[0014] Furthermore, in step 2, based on the collected historical data, scatter plots of the comprehensive flow command and the actual flow through the valve and scatter plots of the comprehensive flow command and the feedback values ​​of the opening of each valve are drawn respectively; the relationship between the comprehensive flow command and the actual flow through the valve is fitted using a local linear regression method, and the curve obtained by fitting is adaptively compressed by a comprehensive use of a revolving door algorithm to obtain the valve flow characteristic curve; the relationship between the comprehensive flow command and the feedback values ​​of the opening of each valve is fitted using a local linear regression method, and the curve obtained by fitting is adaptively compressed by a comprehensive use of a revolving door algorithm to obtain the opening characteristic curve of each valve.

[0015] Furthermore, in step 4, the method for determining whether the fault characteristic parameters of real-time data deviate is as follows: for the standard for determining whether the linearity deviates: if the linearity of the flow characteristic is between 0.75-1.25, it is considered that the linearity of the flow characteristic is in a normal state; if it is less than 0.75 or greater than 1.25, it is considered that a fault has occurred; for the standard for determining whether the discreteness deviates: the standard deviation σ of the error between the actual data and the fitting curve is obtained through step 3, and the valve fault detection threshold is determined to be [-3σ, 3σ], and then the Pauta criterion is used to perform valve fault detection. If the deviation between the fault characteristic parameters calculated according to the actual data and the calculated value of the historical data exceeds the above detection threshold, it is determined that the fault characteristic parameters have deviated, otherwise, it is determined that no deviation has occurred; when detecting whether the flow characteristic discreteness deviates, the fitting curve is the flow characteristic curve, and when detecting whether the valve opening characteristic discreteness deviates, the fitting curve is the valve opening characteristic curve.

[0016] Furthermore, the criteria for judging the fault are: a. If the linearity of the flow characteristic is normal and the discreteness of the flow characteristic has not deviated, the valve is in a non-faulty state; b. If the linearity of the flow characteristic is abnormal and the discreteness of the flow characteristic has not deviated, the valve is in a faulty state with unreasonable opening rules; c. If the linearity of the flow characteristic is abnormal and the discreteness of the flow characteristic has also deviated, the valve state is judged in combination with the valve opening rule: if the discreteness of the valve opening characteristic has not deviated, the valve is in a state where the valve head is detached or loose, and if the discreteness of the valve opening characteristic has also deviated, it is judged that a valve jamming fault has occurred.

[0017] Furthermore, the full operating condition represents the minimum load to the maximum load at which the unit can operate.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] By using the real-time data available from the existing units, it is possible to accurately determine whether the high-pressure regulating valve of the steam turbine has failed. At the same time, the feature compression algorithm is used to effectively eliminate the interference information and sensor measurement noise in the relevant measuring points of the steam turbine valve, thereby achieving a more sensitive and accurate early warning of the steam turbine regulating valve failure. Compared with the traditional method, it can achieve real-time and accurate monitoring, eliminating the disadvantage that the traditional method can only perform special opening and closing tests on the valve. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of a method for detecting and locating a steam turbine valve fault based on flow characteristic and opening characteristic identification;

[0021] Figure 2(a) is the flow characteristic curve of the unit in Example 1, and (b) is a partial enlarged view of (a);

[0022] Figure 3 (a) is the LOESS+SDT result diagram of the flow characteristic curve of the case unit in Example 1, and (b) is a partial enlarged diagram of (a);

[0023] Figure 4 (a) is a diagram showing the linearity analysis result of the flow characteristic curve of the case unit in Example 1, and (b) is a diagram showing the discreteness analysis result of the flow characteristic curve of the case unit in Example 1;

[0024] Figure 5 This is a diagram showing the results of the valve discreteness analysis of the case unit in Example 1. DETAILED DESCRIPTION

[0025] The technical solution of the present invention will be clearly and completely described below in conjunction with the drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] Compared with the original method, the main difference of the present invention is that the previous method is mainly based on the monitoring of defined thermal parameters, and the limitations of the previous parameter definitions make the current method not fully applicable.

[0027] To solve the above problems, the present invention uses the valve-related data acquired in real time to accurately monitor, detect and locate the high-pressure regulation of the steam turbine under the condition that the steam turbine unit is frequently peak-shaving and frequency-regulating at the current stage, the unit is in dynamic adjustment at all times and a special fault detection test cannot be performed arbitrarily. The present invention is described below in conjunction with a specific embodiment. Specific implementation method one:

[0029] A method for detecting and locating a steam turbine valve fault based on flow characteristics and opening characteristics identification comprises the following steps:

[0030] Step 1: Collect data from the steam turbine DCS or SIS system when the high-pressure regulating valve of the unit is in normal working condition (the equipment status is in the optimal state). The data collection interval is 1s, and the relevant data of the steam turbine unit is collected. The specific collection points are shown in Table 1; the collected data needs to meet the unit load to meet the full working condition (the minimum load to the maximum load that the unit can operate) coverage;

[0031] Table 1

[0032]

[0033] Step 2: According to the defined fault characteristic parameter calculation method, the valve flow characteristic curve and each valve opening characteristic curve are obtained based on the collected historical data, and the normal value and variation range of the fault characteristic parameter are determined;

[0034] Preferably, based on the obtained data, a scatter plot is drawn between the comprehensive flow command and the actual flow through the valve, and the relationship between the comprehensive flow command and the actual flow through the valve is fitted using a local linear regression method (LOWESS), and the curve obtained by fitting is adaptively compressed by a revolving door algorithm to obtain a valve flow characteristic curve; a scatter plot is drawn between the comprehensive flow command and the feedback value of the opening of each valve, and the relationship between the comprehensive flow command and the feedback value of the opening of each valve is fitted using a local linear regression method, and the curve obtained by fitting is adaptively compressed by a revolving door algorithm (SDT) to obtain each valve opening characteristic curve.

[0035] The fault characteristic parameters include three indicators: flow characteristic linearity, flow characteristic discreteness and valve opening characteristic discreteness. The calculation methods of the three indicators are described below:

[0036] Flow characteristic linearity calculation method: When the medium pressure cylinder regulating valve does not participate in the regulation, the flow characteristic of the high pressure regulating valve refers to the total valve position command FDEM and the actual steam flow G flowing through the valve. r The actual steam flow rate G r It is determined by the comprehensive valve position command and the main steam pressure. In order to eliminate the influence of the operating condition change, the pressure ratio is used to characterize the steam flow in the calculation, and the change of the measured pressure ratio is used to determine the flow change through the valve, thereby eliminating the influence of the main steam pressure on the calculation result. r The calculation method is as follows:

[0037]

[0038] Among them, P1, P 1-rate Respectively represent the actual pressure and rated pressure after the adjustment stage, P0, P 0-rate They represent the actual pressure and rated pressure before the regulating valve respectively.

[0039] The slope of the tangent line of the extracted flow characteristic curve at a certain point is used as the linearity index.

[0040] Calculation method of flow characteristic dispersion and valve opening characteristic dispersion: Use mean square error (MSE) to evaluate the dispersion of actual data compared with the flow characteristic curve fitting result. Use mean square error to evaluate the dispersion of valve opening characteristic dispersion index compared with the actual data compared with the valve opening characteristic curve fitting result. The calculation method of valve opening characteristic dispersion index is as follows. Where m represents the size of the data set, y represents the size of the valve opening characteristic curve. i represents the model output, Represents the expected output.

[0041]

[0042] Step 3: Based on the real-time data actually collected, the fault characteristic parameters are calculated according to the calculation method of step 2;

[0043] Step 4: Compare the fault characteristic parameters of the real-time data with the fault characteristic parameters under normal working conditions to determine whether the valve has a fault, what kind of fault has occurred, and the location of the fault.

[0044] Furthermore, the method for determining whether the fault characteristic parameters of the real-time data deviate is as follows:

[0045] The standard for judging whether the linearity deviates is: if the flow characteristic linearity is between 0.75-1.25, it is considered that the flow characteristic linearity is in a normal state; if it is less than 0.75 or greater than 1.25, it is considered that a fault has occurred.

[0046] For the standard of judging whether the discreteness has deviated: the standard deviation σ of the error between the actual data and the fitting curve is obtained through step three, and the valve fault detection threshold is determined to be [-3σ, 3σ], and then the valve fault detection is performed using the Pauta criterion. If the deviation between the fault characteristic parameter calculated based on the actual data and the calculated value of the historical data exceeds the above detection threshold, it is judged that the fault characteristic parameter has deviated, otherwise, it is judged that no deviation has occurred; the fault characteristic parameter is the flow characteristic discreteness or the valve opening characteristic discreteness, when detecting whether the flow characteristic discreteness has deviated, the fitting curve is the flow characteristic curve, and when detecting whether the valve opening characteristic discreteness has deviated, the fitting curve is the valve opening characteristic curve.

[0047] Furthermore, the type of valve failure is determined in the following manner:

[0048] 1. If the linearity of the flow characteristic is normal and the discreteness of the flow characteristic has not deviated, the valve is in a non-fault state;

[0049] 2. If the linearity of the flow characteristic is abnormal, but the discreteness of the flow characteristic has not deviated, the valve is in a fault state with unreasonable opening regularity;

[0050] 3. If the linearity of the flow characteristic is abnormal and the discreteness of the flow characteristic also deviates, the valve status is judged in combination with the valve opening rule: if the discreteness of the valve opening characteristic does not deviate, the valve is in a state of valve head detachment or looseness, and if the discreteness of the valve opening characteristic also deviates, it is judged that the valve is stuck.

[0051] The present invention may also have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

[0052] Embodiment 1:

[0053] The data of a unit were selected for analysis. The unit was an N600-16.7 / 538 / 538 steam turbine manufactured by Shanghai Steam Turbine Co., Ltd. The steam distribution method adopted nozzle regulation and was equipped with 4 high-pressure regulating steam valves.

[0054] Firstly, the flow characteristics of the case unit are analyzed, such as Figure 2 Through direct observation of the results, it is found that the flow characteristics are obviously slow in the 62%-69% and 87%-90% comprehensive flow instruction intervals; and in the 84%-90% comprehensive flow instruction interval, the flow characteristics when the load increases and decreases show different distributions.

[0055] The flow characteristic curve is obtained by fitting and compressing the flow characteristics through LOESS+SDT, such as Figure 3 As shown, the linearity and discreteness are analyzed based on this curve, and the results are as follows Figure 4 As shown in the figure, it can be seen that the linearity in the 55%-69% instruction range is very small or even close to 0, and there is obvious jitter in the 79%-90% instruction range and the absolute value obviously exceeds 1.25; at the same time, according to the discreteness analysis results, it is found that there is an obvious increase in discreteness in the 85%-90% range.

[0056] Figure 5The results of the analysis of the opening characteristics of the valves in the case unit are displayed. From the direct observation of the opening pattern, it can be seen that GV2 has an obvious dead band phenomenon in the 85%-90% instruction range, that is, the actual position of the valve is different during the opening and closing processes; the same phenomenon is also observed in the discreteness analysis of the valve opening, that is, the discreteness analysis results of the valves except GV2 are basically all below 0.5, and the discreteness of GV2 in the 85%-90% instruction range has increased significantly, which also proves that GV2 has a hardware jam failure in this range.

[0057] Based on the above analysis of the valve flow characteristics and valve opening characteristics of the case unit, combined with the valve fault warning method proposed in this paper, the following conclusions are drawn: (1) The valve opening pattern design within the 62%-69% comprehensive flow instruction interval is unreasonable and there is no obvious hardware failure. The valve opening pattern in this area should be redesigned; (2) Within the 85%-90% comprehensive flow instruction interval, GV2 has a hardware failure of valve jamming, and its hardware should be repaired to solve its impact on the flow characteristics.

[0058] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for detecting and locating a steam turbine valve fault, characterized in that , including the following steps: Step 1: Collect historical data related to regulating valves under normal operating conditions of the unit. The collected historical data must meet the requirements of the unit load to cover all operating conditions; Step 2: According to the defined fault characteristic parameter calculation method, based on the collected historical data, draw the scatter plot of the comprehensive flow command and the actual flow through the valve and the scatter plot of the comprehensive flow command and the feedback value of each valve opening respectively; use the local linear regression method to fit the relationship between the comprehensive flow command and the actual flow through the valve, and use the revolving door algorithm to adaptively compress the curve obtained by fitting to obtain the valve flow characteristic curve; use the local linear regression method to fit the relationship between the comprehensive flow command and the feedback value of each valve opening, and use the revolving door algorithm to adaptively compress the curve obtained by fitting to obtain the opening characteristic curve of each valve, and determine the normal value of the fault characteristic parameter and its variation range; the fault characteristic parameters include flow characteristic linearity, flow characteristic discreteness and valve opening characteristic discreteness; the slope of the tangent of the flow characteristic curve at a certain point is used as the flow characteristic linearity index; the mean square error is used to evaluate the flow characteristic discreteness by comparing the discreteness of the actual data with the flow characteristic curve; the mean square error is used to evaluate the valve opening characteristic discreteness by comparing the discreteness of the actual data with the valve opening characteristic curve; Step 3: Based on the real-time data actually collected, the fault characteristic parameters are calculated according to the calculation method of step 2; Step 4: Compare the fault characteristic parameters of the real-time data with the fault characteristic parameters under normal working conditions to determine whether the valve has a fault, what kind of fault has occurred, and the location of the fault.

2. A method for detecting and locating a steam turbine valve fault according to claim 1, characterized in that: In step 1, the relevant historical data collected include main steam pressure, regulating stage pressure, valve opening feedback value and comprehensive flow command.

3. A method for detecting and locating a steam turbine valve fault according to claim 1, characterized in that: In step 4, the method for judging whether the fault characteristic parameters of the real-time data deviate is as follows: for the standard for judging whether the linearity deviates: if the linearity of the flow characteristic is between 0.75-1.25, it is considered that the linearity of the flow characteristic is in a normal state; if it is less than 0.75 or greater than 1.25, it is considered that a fault occurs; for the standard for judging whether the discreteness deviates: the standard deviation σ of the error between the actual data and the fitting curve is obtained through step 3, and the valve fault detection threshold is determined to be [-3σ, 3σ], and then the Pauta criterion is used to perform valve fault detection. If the deviation between the fault characteristic parameter calculated according to the actual data and the calculated value of the historical data exceeds the above detection threshold, it is judged that the fault characteristic parameter deviates, otherwise, it is judged that no deviation occurs; when detecting whether the flow characteristic discreteness deviates, the fitting curve is the flow characteristic curve, and when detecting whether the valve opening characteristic discreteness deviates, the fitting curve is the valve opening characteristic curve.

4. A method for detecting and locating a steam turbine valve fault according to claim 3, characterized in that: The criteria for judging the fault are: a. If the linearity of the flow characteristic is normal and the discreteness of the flow characteristic has not deviated, the valve is in a non-fault state; b. If the linearity of the flow characteristic is abnormal and the discreteness of the flow characteristic has not deviated, the valve is in a fault state with unreasonable opening regularity; c. If the linearity of the flow characteristic is abnormal and the discreteness of the flow characteristic also deviates, the valve status is judged in combination with the valve opening rule: if the discreteness of the valve opening characteristic does not deviate, the valve is in a state of valve head detachment or looseness, and if the discreteness of the valve opening characteristic also deviates, it is judged that the valve is stuck.

5. A method for detecting and locating a steam turbine valve fault according to claim 1, characterized in that: The full operating condition represents the minimum load to the maximum load at which the unit can operate.