A valve oscillation fault detection method based on oscillation index

Abstract

A valve oscillation fault detection method based on an oscillation index relates to the field of fault detection technology. The present invention is to solve the problem that existing methods cannot be applied to the early warning of faults in the middle section of a steam turbine under real peak-shaving and frequency-regulating operation scenarios. The present invention collects the actual pressure and actual opening of the turbine regulating valve before and after the stage, calculates the actual pressure ratio of the turbine regulating valve, constructs a valve opening characteristic diagram and then calculates the data points, then constructs a parallelogram of the data points to filter and archive points, and uses the archive points to fit into a valve opening characteristic curve, and then calculates the valve oscillation index. If the valve oscillation index exceeds its warning range, the turbine regulating valve has a stuck fault, otherwise the turbine regulating valve has an irregular opening or a valve loose fault.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault detection, and in particular relates to the detection of valve oscillation faults. Background Art

[0002] With the clean and efficient use of coal, intelligent upgrades to power plants have become a trend. As the primary power source for the power grid, thermal power plants boast the largest installed capacity and strong controllability, shouldering an increasing burden of peak-shaving and frequency regulation. However, most existing thermal power units are designed for base load operation and are unable to fully adapt to the operational requirements of deep peak-shaving and rapid frequency regulation. Frequent, large-scale, and rapid load fluctuations can cause alternating thermal stresses in key components, leading to thermal fatigue damage, which can ultimately cause equipment performance degradation and oscillation failures. Further improving the safe and economical operation of steam turbines is an effective means of promoting the absorption of power generation and the transition to clean energy in thermal power plants, and has therefore received widespread attention.

[0003] Due to the long-term presence of peak and frequency regulation commands during steam turbine operation, the time it takes for turbine unit failures to occur is greatly shortened. However, with traditional scheduled maintenance strategies, the level of failure in a unit may not be discovered until maintenance is performed, even after it has reached a critical level and has already caused significant losses. Therefore, sensitive identification and early warning of failures are urgently needed to effectively prevent further escalation of the impact of failures. A mathematical model of the steam turbine in a gas turbine combined cycle power plant was established, considering material and energy balance equations and a steam expansion model for the steam turbine, and used for fault detection.

[0004] Model-based research on steam turbine fault diagnosis and early warning systems has significant shortcomings in practical applications, as early fault signs are very subtle and unmodeled characteristics make them difficult to accurately identify. Existing methods have significant shortcomings in practical applications. While existing research on steam turbine equipment fault diagnosis and early warning systems has employed parameters such as relative internal efficiency and terminal error, practice and application have shown that these parameters fluctuate significantly during dynamic adjustments under peak and frequency regulation conditions, failing to accurately reflect the equipment's status in real time. These fluctuations can easily overwhelm even the smallest early signs of a fault. While other approaches have considered dynamic information about steam and gas turbines, they typically focus on rotor dynamics rather than thermodynamics. Finally, significant coupling exists between unit parameters, which existing research has failed to comprehensively consider.

[0005] For the above reasons, some existing research has certain limitations and cannot be fully applied to the early warning of faults in the middle section of the steam turbine under real peak and frequency regulation operation scenarios. Summary of the Invention

[0006] The present invention aims to solve the problem that existing methods are not applicable to early warning of faults in the middle section of a steam turbine under real peak-shaving and frequency-regulating operation scenarios, and now provides a valve oscillation fault detection method based on an oscillation index.

[0007] A valve oscillation fault detection method based on an oscillation index comprises the following steps:

[0008] Step 1: Collect the actual pressure and opening of the turbine regulating valve before and after the stage respectively, and calculate the actual pressure ratio of the turbine regulating valve;

[0009] Step 2: Establish a valve coordinate system with the actual pressure ratio and actual opening of the turbine regulating valve as the horizontal and vertical coordinates respectively. Draw scattered points in the valve coordinate system according to the actual pressure ratio and actual opening obtained in step 1 to obtain the valve opening characteristic diagram. The coordinates of the i-th scattered point are (x i ,y i ), i = 1, 2, ..., I, I is the total number of scattered points in the valve opening characteristic diagram;

[0010] Step 3: Calculate the vertical coordinate of the fitting result of each scattered point in the valve opening characteristic diagram according to the multi-fitting formula:

[0011]

[0012] Among them, Y i is the ordinate of the fitting result of the i-th scatter point, C1 is the constant term of the polynomial fitting formula, C2 and C3 are the linear term coefficient and quadratic term coefficient of the polynomial fitting formula respectively,

[0013] The horizontal coordinate x of the scattered point i The vertical coordinate Y of the fitting result of the scatter point i Construct data points (x i ,Y i ), and sort all data points along the horizontal axis;

[0014] Step 4: Use the latest archive point and the latest data point (x n ,Y n ) is used as the center line to establish a parallelogram, the side of the parallelogram intersecting the center line is parallel to the longitudinal axis and has a length of 2ΔE, ΔE is the optimal actual opening value, n = 3, 4, ..., I, the first data point (x1, y1) is the initial latest archive point;

[0015] Step 5: Determine the latest archive point and the latest data point (x n ,Y n ) are all located in the parallelogram, if so, proceed directly to step 6, otherwise the data point (x n-1 ,Y n-1) Save it to the archive set and use it as the latest archive point, then proceed to step 6;

[0016] Step 6: Determine whether n is equal to I. If so, use the curve fitted by all data points in the archive set as the valve opening characteristic curve, and then execute step 7. Otherwise, set n = n + 1, and then return to step 4.

[0017] Step 7: Calculate the valve oscillation index VOI according to the following formula:

[0018]

[0019] Among them, Δy is the difference between the vertical coordinates of the first and end points of the valve opening characteristic curve, and Δx is the difference between the horizontal coordinates of the first and end points of the valve opening characteristic curve;

[0020] Step 8: Determine whether the valve oscillation index VOI exceeds its warning range. If so, execute step 9. Otherwise, the turbine regulating valve has no oscillation fault.

[0021] Step 9: Determine whether there is a dead zone in the valve opening characteristic curve. If so, the turbine regulating valve is stuck. Otherwise, the turbine regulating valve opens irregularly or is loose.

[0022] Furthermore, in the above step 1, the actual pressure ratio P is calculated according to the following formula: r :

[0023]

[0024] Among them, P0 and P1 are the actual pressures before and after the stage of the turbine regulating valve respectively.

[0025] Furthermore, the constant term C1 of the polynomial fitting formula, the linear term coefficient C2 and the quadratic term coefficient C3 of the polynomial fitting formula in the above step 3 are obtained as follows:

[0026] Calculate the weight of each scatter point separately and establish the following two matrices:

[0027]

[0028] Among them, w i is the weight of the i-th scatter point,

[0029] The elements in the matrix resulting from dividing the matrix A and the matrix B are respectively used as the constant term C1 of the polynomial fitting formula, the linear term coefficient C2 and the quadratic term coefficient C3 of the polynomial fitting formula.

[0030] Furthermore, the weight of each of the above scatter points is obtained by the following method:

[0031] The grid search method is used to select the optimal actual pressure ratio value in the valve opening characteristic diagram, and all scattered points in the valve opening characteristic diagram are divided into multiple intervals with the optimal actual pressure ratio value as the unit interval.

[0032] Calculate the weight of each scatter point according to the following formula:

[0033] w i =(1-f i 3 ) 3 ,

[0034] Among them, f i is the distance between the i-th scattered point and the midpoint of its interval.

[0035] Furthermore, the above-mentioned optimal actual pressure ratio value is 0.03.

[0036] Furthermore, a grid search method is used to select the optimal actual opening value in the valve opening characteristic diagram.

[0037] Furthermore, the above-mentioned optimal actual opening value is 0.03.

[0038] Aiming at the demand for frequent peak and frequency regulation of steam turbines, the present invention analyzes the fault principle and actual impact, utilizes actual unit power data, and establishes a valve oscillation fault detection method based on oscillation index to achieve sensitive identification and early warning of valve oscillation faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a schematic diagram of load oscillation fault data;

[0040] Figure 2 Schematic diagram of valve structure, where (a) represents valve group configuration and (b) represents a single valve body;

[0041] Figure 3 This is the time domain diagram of the pressure change of the fault case unit;

[0042] Figure 4 This is the valve opening curve of the case unit

[0043] Figure 5 The final result of the valve opening curve of the case unit

[0044] Figure 6 Analysis results of the case unit oscillation failure

[0045] Figure 7 The valve opening curve results of the case unit

[0046] Figure 8 Analysis results of the case unit oscillation failure

[0047] Figure 9 The flowchart of a valve oscillation fault detection method based on oscillation index is shown. DETAILED DESCRIPTION

[0048] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not 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. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other in the absence of conflict.

[0049] The main steam inlet flow rate of a steam turbine is controlled by adjusting the opening of the regulating valve in the steam inlet section. As the most important control device for adjusting the load of the steam turbine unit, the operating performance of the regulating valve directly affects the regulating performance of the entire unit. However, long-term operation can cause the mechanical mechanism of the steam turbine regulating valve to be susceptible to scaling, aging, valve stem damage, valve head wear, valve core deviation, and other problems. Frequent adjustment of the high-pressure regulating valve group will cause increased wear of the valve components, leading to valve component oscillation failures, significantly changing the regulating performance of the unit, and adversely affecting the stability of the power grid. In addition, load oscillation and load mutation failures are the two most typical load loss control failures of valve components. With the increase in the demand for peak and frequency regulation of the power grid, these two failures will have a serious impact on turbine load control and the safe and stable operation of the power grid.

[0050] When a valve oscillation fault occurs during peak and frequency regulation of a thermal power unit, general performance data cannot reflect the performance change in real time. The purpose of this embodiment is to provide a valve oscillation fault detection method based on an oscillation index, which can use real-time parameters for online detection, thereby accurately detecting the oscillation fault, reminding operators to check and deal with the abnormal system as soon as possible, and improving the system's operational safety performance. The details are as follows:

[0051] A valve oscillation fault detection method based on an oscillation index comprises the following steps:

[0052] Step 1: Collect the actual pressure P0, P1 and actual opening of the turbine regulating valve before and after the stage, and calculate the actual pressure ratio P of the turbine regulating valve according to the following formula: r :

[0053]

[0054] Step 2: Establish a valve coordinate system with the actual pressure ratio and actual opening of the turbine regulating valve as the horizontal and vertical coordinates respectively. Draw scattered points in the valve coordinate system based on the actual pressure ratio and actual opening obtained in step 1 to obtain the valve opening characteristic diagram. The coordinates of the i-th scattered point are (x i ,y i ), i=1,2,...,I, I is the total number of scattered points in the valve opening characteristic diagram.

[0055] Step 3: Use a grid search method to select the optimal actual pressure ratio value from the valve opening characteristic diagram. In this embodiment, the optimal actual pressure ratio value is 0.03. Divide all scattered points in the valve opening characteristic diagram into multiple intervals, using the optimal actual pressure ratio value as the unit interval. For each interval, first determine the center point of the interval. Then, calculate the weight of each scattered point using the cubic weight function:

[0056] w i =(1-f i 3 ) 3 ,

[0057] Among them, w i is the weight of the i-th scatter point, f i is the distance between the i-th scattered point and the midpoint of its interval.

[0058] After obtaining the weight of each scatter point, the following two matrices are established:

[0059]

[0060] The result matrix of dividing matrix A and matrix B is a matrix containing three elements. These three elements are used as the constant term C1, the linear term coefficient C2, and the quadratic term coefficient C3 of the polynomial fitting formula. Substitute the three parameters into the polynomial fitting formula and calculate the vertical coordinate of the fitting result for each scattered point in the valve opening characteristic diagram:

[0061]

[0062] Among them, Y i is the ordinate of the fitting result of the i-th scatter point.

[0063] The horizontal coordinate x of the scattered point i The vertical coordinate Y of the fitting result of the scatter point i Construct data points (x i ,Y i ), all data points are fitted into a curve, which is the initial valve opening curve. At this time, the data points on the valve opening curve are fluctuating. In order to make the curve smoother, the following operations are required.

[0064] Step 4: Sort all data points along the horizontal axis. Use the grid search method to select the optimal actual opening value ΔE in the valve opening characteristic diagram. In this embodiment, the optimal actual opening value is 0.03.

[0065] The latest archive point and the latest data point (x n ,Y n ) is used as the center line to establish a parallelogram, the side of the parallelogram intersecting the center line is parallel to the longitudinal axis and has a length of 2ΔE, n=3,4,...,I, and the first data point (x1,Y1) is the initial latest archive point.

[0066] Step 5: Determine the latest archive point and the latest data point (x n ,Y n ) are all located in the parallelogram, if so, proceed directly to step 6, otherwise the data point (x n-1 ,Y n-1 ) is saved to the archive set and used as the latest archive point, then proceed to step 6.

[0067] Step 6: Determine whether n is equal to I. If so, use the curve fitted by all data points in the archive set as the valve opening characteristic curve, and then execute step 7. Otherwise, set n=n+1, and then return to step 4.

[0068] After the above steps 4 to 6 are cycled, all fluctuating data points can be eliminated, and the curve fitted by all the data points finally saved in the archive set is a smooth valve opening characteristic curve.

[0069] Step 7: Calculate the valve oscillation index VOI according to the following formula:

[0070]

[0071] Among them, Δy is the difference between the vertical coordinates of the first and end points of the valve opening characteristic curve, and Δx is the difference between the horizontal coordinates of the first and end points of the valve opening characteristic curve.

[0072] Step 8: Determine whether the valve oscillation index VOI exceeds its warning range. If so, execute step 9. Otherwise, the turbine regulating valve has no oscillation fault.

[0073] Step 9: Determine whether there is a dead zone phenomenon in the valve opening characteristic curve, that is, the opening characteristics when the pressure ratio increases and decreases present different distributions. If so, the turbine regulating valve has a stuck fault; otherwise, the turbine regulating valve has an irregular opening or a loose valve fault.

[0074] In actual application, based on the analysis of a large amount of operating data from this steam turbine, a valve oscillation index warning threshold of -0.75 to 2.75 is more reasonable for valves #1 and #2, which open first. For valves #3 and #4, which open later, a valve oscillation index warning threshold of -0.75 to 4.75 is more reasonable.

[0075] In order to test the beneficial effects of the valve oscillation fault detection method based on the oscillation index proposed in this embodiment, an experimental verification is conducted on operating history data including actual unit data in normal state and different fault types for detection and verification.

[0076] First, the actual operating data of a steam turbine unit in a thermal power plant was selected for analysis. The unit is a N600-16.7 / 538 / 538 steam turbine manufactured by Shanghai Steam Turbine Co., Ltd. The steam distribution method adopts nozzle regulation, and 4 high-pressure regulating steam valves are set inside. The test data is obtained by splicing the two segments of data collected. The data length is 2000s. The specific parameter change trends are as follows Figure 3 The flow characteristics of the case unit are analyzed and the actual valve opening characteristic curve is obtained by fitting algorithm. Figure 4 shown. Figure 5 The results of the analysis of the specific valve opening characteristics of the case unit are presented. Direct observation of the opening pattern reveals that the GV1 valve opening is relatively slow within the 50%-65% pressure ratio range. Furthermore, within the 65%-70% range of the integrated flow command, the flow characteristics exhibit different distributions during load increases and decreases, creating a dead zone. This suggests that the cause of the valve oscillation failure is a stuck valve. The valve integrated oscillation index (VOI) within the 46%-49% and 73%-79% command ranges is very small, even close to 0. Within the 65%-73% command range, significant jitter is observed, with the absolute value significantly exceeding 2.75. The valve oscillation index (VOI) significantly exceeds the warning threshold, indicating that the actual valve position differs during opening and closing. This analysis indicates that the GV1 has an oscillation fault and has been identified.

[0077] Based on the above analysis of the valve opening characteristics of Unit 1, combined with the valve oscillation fault early warning method proposed in this embodiment, the fault analysis results show that the valve opening pattern design within the 49%-50% and 65%-73% pressure ratio ranges is unreasonable or a hardware fault exists. Due to the presence of dead zones in the valve opening curve, the initial diagnosis of valve oscillation is that the valve is stuck. Hardware faults in these areas should be inspected and repaired to mitigate the negative impact of the oscillation fault.

[0078] In order to further verify the reliability of the method, the data of normal non-oscillation case units were analyzed and verified. By calculating the valve opening characteristics of the normal unit, it was found that the valve oscillation index VOI of the valve flow characteristics of the full operating conditions was within the warning threshold range. It can be judged that the unit did not have an oscillation fault in the full operating range.

[0079] According to direct observation, since the valve oscillation index VOI is within a reasonable range, there is no valve oscillation fault in the case unit.

[0080] It can be seen from the above experiments that the method proposed in this embodiment can achieve better identification accuracy than the traditional method. Since this embodiment is based on the analysis of the fault mechanism and external manifestations, it has good transferability and is applicable to different types of units.

[0081] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the invention. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in conjunction with other described embodiments.

Claims

1. A valve oscillation fault detection method based on oscillation index, characterized in that: The following steps are involved: Step 1: Collect the actual pressure and opening of the turbine regulating valve before and after the stage respectively, and calculate the actual pressure ratio of the turbine regulating valve; Step 2: Establish a valve coordinate system with the actual pressure ratio and actual opening of the turbine regulating valve as the horizontal and vertical coordinates respectively. Draw scattered points in the valve coordinate system according to the actual pressure ratio and actual opening obtained in step 1 to obtain the valve opening characteristic diagram. The coordinates of the i-th scattered point are (x i ,y i ), i = 1, 2, ..., I, I is the total number of scattered points in the valve opening characteristic diagram; Step 3: Calculate the vertical coordinate of the fitting result of each scattered point in the valve opening characteristic diagram according to the multi-fitting formula: Among them, Y i is the ordinate of the fitting result of the i-th scatter point, C1 is the constant term of the polynomial fitting formula, C2 and C3 are the linear term coefficient and quadratic term coefficient of the polynomial fitting formula respectively, The horizontal coordinate x of the scattered point i The vertical coordinate Y of the fitting result of the scatter point i Construct data points (x i ,Y i ), and sort all data points along the horizontal axis; Step 4: Use the latest archive point and the latest data point (x n ,Y n ) is used as the center line to establish a parallelogram, the side of the parallelogram intersecting the center line is parallel to the longitudinal axis and has a length of 2ΔE, ΔE is the optimal actual opening value, n = 3, 4, ..., I, the first data point (x1, y1) is the initial latest archive point; Step 5: Determine the latest archive point and the latest data point (x n ,Y n ) are all located in the parallelogram, if so, proceed directly to step 6, otherwise the data point (x n-1 ,Y n-1 ) Save it to the archive set and use it as the latest archive point, then proceed to step 6; Step 6: Determine whether n is equal to I. If so, use the curve fitted by all data points in the archive set as the valve opening characteristic curve, and then execute step 7. Otherwise, set n = n + 1, and then return to step 4. Step 7: Calculate the valve oscillation index VOI according to the following formula: Among them, Δy is the difference between the vertical coordinates of the first and end points of the valve opening characteristic curve, and Δx is the difference between the horizontal coordinates of the first and end points of the valve opening characteristic curve; Step 8: Determine whether the valve oscillation index VOI exceeds its warning range. If so, execute step 9. Otherwise, the turbine regulating valve has no oscillation fault. Step 9: Determine whether there is a dead zone in the valve opening characteristic curve. If so, the turbine regulating valve is stuck. Otherwise, the turbine regulating valve opens irregularly or is loose.

2. A valve oscillation fault detection method based on oscillation index according to claim 1, characterized in that: In step 1, the actual pressure ratio P is calculated according to the following formula: r : Among them, P0 and P1 are the actual pressures before and after the stage of the turbine regulating valve respectively.

3. The valve oscillation fault detection method based on oscillation index according to claim 1, characterized in that: The method for obtaining the constant term C1 of the polynomial fitting formula, the linear term coefficient C2 and the quadratic term coefficient C3 of the polynomial fitting formula in step 3 is as follows: Calculate the weight of each scatter point separately and establish the following two matrices: Among them, w i is the weight of the i-th scatter point, The elements in the matrix resulting from dividing the matrix A and the matrix B are respectively used as the constant term C1 of the polynomial fitting formula, the linear term coefficient C2 and the quadratic term coefficient C3 of the polynomial fitting formula.

4. The valve oscillation fault detection method based on oscillation index according to claim 3 is characterized in that: The weight of each scatter point is obtained by the following method: The grid search method is used to select the optimal actual pressure ratio value in the valve opening characteristic diagram, and all scattered points in the valve opening characteristic diagram are divided into multiple intervals with the optimal actual pressure ratio value as the unit interval. Calculate the weight of each scatter point according to the following formula: w i =(1-f i 3 ) 3 , Among them, f i is the distance between the i-th scattered point and the midpoint of its interval.

5. The valve oscillation fault detection method based on oscillation index according to claim 4 is characterized in that: The optimal actual pressure ratio value is 0.

03.

6. The valve oscillation fault detection method based on oscillation index according to claim 1, characterized in that: The grid search method is used to select the optimal actual opening value in the valve opening characteristic diagram.

7. The valve oscillation fault detection method based on oscillation index according to claim 6, characterized in that: The optimal actual opening value is 0.03.

Images