Online evaluation and abnormity early warning method and system for through-flow efficiency of steam turbine

Abstract

The invention relates to the field of steam turbine through-flow efficiency evaluation and early warning, and provides a steam turbine through-flow efficiency online evaluation and abnormity early warning method and system, and the method comprises the steps: deploying through-flow monitoring points, collecting the real-time working parameters of the through-flow part of a steam turbine unit in real time, and carrying out the normalization processing of the real-time working parameters, and obtaining a normalization set; calculating a real-time output coefficient of an associated component of the through-flow monitoring point according to the normalized set, capturing a gradual change abnormal parameter in the normalized set, and integrating the gradual change abnormal parameter and the real-time output coefficient into a real-time output set; the node through-flow value of the through-flow monitoring point is evaluated based on a pre-constructed online evaluation model by taking the real-time output set as a prior probability and combining the real-time working parameters; and combining the node through-flow value and the three-dimensional model of the through-flow part of the turboset to evaluate the through-flow abnormity degree of the through-flow part of the turboset, and when the through-flow abnormity degree exceeds a preset abnormal threshold value, triggering a through-flow abnormity early warning instruction. The method solves the problem that an existing method only depends on limited parameters to evaluate the through-flow efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of steam turbine flow efficiency evaluation and early warning, and in particular to a method and system for online evaluation and abnormal early warning of steam turbine flow efficiency. Background Art

[0002] As core equipment in thermal power plants, nuclear power plants, and various industrial power systems, the efficiency of a steam turbine's flow passage is directly related to the energy efficiency and economical operation of the entire unit. During the long-term operation of a steam turbine, the performance of the flow passage is affected by a variety of factors, such as deviations from design steam parameters, wear, scaling, and deformation of flow passage components, damage or breakage of blades, and deformation of nozzles and baffles. These issues can reduce the efficiency of the turbine's flow passage, thereby affecting key performance indicators such as the unit's output power and heat rate, increasing fuel consumption and operating costs, and in severe cases, may even cause equipment failure, threatening the unit's safe operation.

[0003] Traditional methods for evaluating steam turbine flow efficiency rely primarily on offline performance testing and periodic inspections. Offline performance testing typically requires shutting down the turbine under specific operating conditions, which is not only time-consuming and labor-intensive but also fails to accurately reflect dynamic performance changes during actual operation. While regular inspections can reveal obvious component damage and failures, they struggle to detect subtle changes in flow efficiency and early signs of abnormalities, often preventing effective preventive measures before failures occur.

[0004] Chinese patent CN106908249B discloses a method for diagnosing abnormal flow stage efficiency in a steam turbine's high-pressure cylinder. The method involves the following steps: first, testing the effect of changes in inlet steam temperature on the flow stage efficiency of the high-pressure cylinder. The linearization trend of the steam expansion process within the turbine stage is used to determine whether there is steam leakage within the high-pressure cylinder. Next, the electric doors of each extraction stage of the high-pressure cylinder are closed, and the pressures and temperatures of the inlet, exhaust, and extraction stages are measured. The extraction pressure and post-stage flow ratio are calculated under varying operating conditions. The difference between the extraction pressure and the corresponding post-stage pressure is compared to ultimately determine the cause of the abnormal flow stage efficiency. However, this diagnostic method relies solely on limited parameters such as inlet steam temperature and extraction pressure, and fails to integrate a three-dimensional model of the flow section with multi-source dynamic data (such as vibration and wear). This results in incomplete fault feature extraction, making it difficult to adapt to dynamic changes in unit load (such as peak-shaving scenarios) and resulting in a high rate of missed detections. Summary of the Invention

[0005] The present invention aims to solve at least one of the problems existing in the prior art and provides a method and system for online evaluation of flow efficiency of a steam turbine and abnormality early warning.

[0006] One aspect of the present invention provides a method for online evaluation of steam turbine flow efficiency and abnormality early warning, the method comprising:

[0007] Deploying flow monitoring points based on a three-dimensional model of the flow section of the steam turbine unit, collecting real-time operating parameters of the flow section of the steam turbine unit through the flow monitoring points, and normalizing the real-time operating parameters to obtain a normalized set;

[0008] Loading the normalized set, calculating the real-time output coefficients of the associated components of the flow monitoring point according to the normalized set, capturing the gradually changing abnormal parameters in the normalized set based on a CUSUM algorithm, and integrating the gradually changing abnormal parameters and the real-time output coefficients into a real-time output set;

[0009] Acquire the real-time output set, use the real-time output set as a priori probability, combine with the real-time operating parameters in the normalized set, evaluate the node flow value of the flow monitoring point based on a pre-built online evaluation model, and feed back the node flow value;

[0010] loading the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, comprehensively evaluating the flow abnormality of the flow portion of the steam turbine unit by combining the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, and determining whether the flow abnormality exceeds a preset abnormality threshold;

[0011] If the flow abnormality exceeds the preset abnormality threshold, a flow abnormality warning instruction is triggered.

[0012] Optionally, the normalizing the real-time working parameters to obtain a normalized set includes:

[0013] Loading the real-time operating parameters and processing missing values ​​and abnormal values ​​of the real-time operating parameters, wherein the real-time operating parameters include thermodynamic parameters, geometric parameters, flow characteristic parameters, and monitoring point coordinate parameters;

[0014] Performing spectrum analysis on the real-time operating parameters, calculating data dispersion of the real-time operating parameters, and determining smoothness of the real-time operating parameters using the data dispersion;

[0015] Determining whether the data discreteness of the real-time operating parameter exceeds a preset discrete threshold;

[0016] If the data discreteness exceeds the preset discrete threshold, performing adaptive filtering processing on the real-time working parameters based on an adaptive filtering algorithm;

[0017] If the data discreteness does not exceed the preset discrete threshold, a high-order wavelet noise reduction function is used to perform noise reduction processing on the real-time working parameters;

[0018] The real-time operating parameters after denoising by an adaptive filtering algorithm and a high-order wavelet denoising function are integrated, and the real-time operating parameters after denoising are mapped to a corresponding uniform normal distribution using a quantile normalization method to obtain the normalized set.

[0019] Optionally, the performing adaptive filtering processing on the real-time working parameter based on an adaptive filtering algorithm includes:

[0020] Loading the real-time operating parameters, and performing digital low-pass filtering on the real-time operating parameters based on a digital low-pass filter;

[0021] Using a cubic spline interpolation method to interpolate and reconstruct the filtered real-time operating parameters to obtain parameter reconstruction data, obtaining parameter estimation values ​​corresponding to the filtering threshold, combining with an optimal combination weighted average method to obtain parameter replacement values, and integrating the parameter replacement values ​​into the real-time operating parameters to obtain a parameter integration set;

[0022] Loading the parameter integration set, determining filter coefficients of the digital low-pass filter based on the data discreteness of the parameter integration set in combination with a Gaussian weighting algorithm, and dynamically adjusting the digital low-pass filter using the filter coefficients;

[0023] The filter coefficient is calculated by the following formula:

[0024]

[0025] Among them, L but , L0 represent the filter coefficient and initial coefficient respectively, f s 、f c Respectively represent the parameter frequency and parameter frequency range, λ1 and λ2 represent the first weighting coefficient and the second weighting coefficient based on the optimal combination weighted average method, Δf is the transition bandwidth of the digital low-pass filter, LN is the filter order of the digital low-pass filter, D L represents the data dispersion and SNR represents the filter signal-to-noise ratio and A t 、 σ(A t ) represent the working parameters, parameter mean, and parameter covariance of the digital low-pass filter at the current moment t, respectively, and f max 、f min are the maximum sampling frequency and the minimum sampling frequency respectively, and z is the number of noise signals in the parameter;

[0026] Taking the parameter integration set as input, the digital low-pass filter is used to perform secondary dynamic filtering on the parameter integration set, and the real-time working parameters after secondary filtering are output.

[0027] Optionally, calculating the real-time output coefficient of the associated components of the flow monitoring point according to the normalized set includes:

[0028] Loading the normalized set, and calculating the comprehensive flow efficiency of the flow monitoring point based on the flow balance formula of the flow monitoring point;

[0029] Loading the throughflow comprehensive efficiency of the throughflow monitoring point, traversing the normalized set, calculating the loss reduction coefficient of the throughflow monitoring point in the normalized set, correcting the throughflow comprehensive efficiency based on the loss reduction coefficient, and outputting the corrected throughflow comprehensive efficiency;

[0030] Traversing the normalized set, and determining the associated comprehensive efficiency of the flow monitoring point based on the number of associated components of the flow monitoring point and the associated component loss reduction coefficient;

[0031] The corrected throughflow comprehensive efficiency and the associated comprehensive efficiency are obtained, and the real-time output coefficient of the associated components of the throughflow monitoring point is determined based on the throughflow comprehensive efficiency and the associated comprehensive efficiency in combination with an intra-stage loss decomposition algorithm.

[0032] Optionally, the calculating the comprehensive flow efficiency of the flow monitoring point based on the flow balance formula of the flow monitoring point includes:

[0033] The overall flow efficiency is calculated according to the following formula:

[0034]

[0035] Among them, η all represents the overall flow efficiency, P is the number of associated components at the flow monitoring point, G P , Q net Respectively represent the air flow rate and low calorific value of the flow monitoring point, D gr 、D zr Respectively represent the superheated gas flow rate and reheated gas flow rate at the flow monitoring point, h gr 、h fw Represents superheated steam enthalpy and feed steam enthalpy, h" zr 、h' zr They respectively represent the reheat steam outlet enthalpy and the reheat steam inlet enthalpy of the flow monitoring point.

[0036] Optionally, the objective function of the online evaluation model is expressed as:

[0037]

[0038] Where ΔP ltIt represents the output value of the objective function of the online evaluation model and is also the node flow value. w 、ω j are the power output value and the flow gain factor of the flow monitoring point respectively, P is the number of associated components of the flow monitoring point and also the number of monitoring points of the flow monitoring point, ΔW j 、T j They represent the power change and inertia response time respectively, J(x) is the mean value of the gradual abnormal parameter, x is the gradual abnormal parameter, and κ represents the real-time output coefficient.

[0039] Optionally, the loading of the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, and comprehensively evaluating the flow abnormality of the flow portion of the steam turbine unit in combination with the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, includes:

[0040] Loading the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, and determining a real-time flow abnormality point based on the node flow value, the number of monitoring points in the three-dimensional model of the flow portion of the steam turbine unit, and the real-time output coefficient;

[0041] Loading the real-time abnormal flow point and the node flow value corresponding to the real-time abnormal flow point, and determining the point abnormality degree of the real-time abnormal flow point based on a loss term quantification algorithm;

[0042] Obtaining at least one group of point abnormality degrees of the real-time flow abnormality points, performing weighted summation of the point abnormality degrees of the real-time flow abnormality points based on the flow abnormality point weights, and obtaining the flow abnormality degree of the flow abnormality portion of the steam turbine unit.

[0043] Another aspect of the present invention provides a system for online evaluation and abnormality warning of steam turbine flow efficiency, the system comprising:

[0044] a data acquisition module, configured to deploy flow monitoring points based on a three-dimensional model of the flow passage of the steam turbine unit, collect real-time operating parameters of the flow passage of the steam turbine unit through the flow monitoring points, and normalize the real-time operating parameters to obtain a normalized set;

[0045] a real-time output calculation module, configured to load the normalized set, calculate the real-time output coefficients of the associated components of the flow monitoring point according to the normalized set, capture the gradually changing abnormal parameters in the normalized set based on a CUSUM algorithm, and integrate the gradually changing abnormal parameters and the real-time output coefficients into a real-time output set;

[0046] a node flow calculation module, configured to obtain the real-time output set, use the real-time output set as a priori probability, combine the real-time operating parameters in the normalized set, evaluate the node flow value of the flow monitoring point based on a pre-built online evaluation model, and provide feedback on the node flow value;

[0047] An evaluation and early warning module is used to load the node flow value and the three-dimensional model of the flow part of the steam turbine unit, comprehensively evaluate the flow abnormality of the flow part of the steam turbine unit in combination with the node flow value and the three-dimensional model of the flow part of the steam turbine unit, and determine whether the flow abnormality exceeds a preset abnormality threshold; if the flow abnormality exceeds the preset abnormality threshold, trigger a flow abnormality early warning instruction.

[0048] Optionally, the data acquisition module includes:

[0049] A monitoring point deployment unit, used to deploy flow monitoring points based on a three-dimensional model of the flow section of the steam turbine unit;

[0050] A monitoring point acquisition unit, configured to acquire real-time operating parameters of the flow-through portion of the steam turbine unit through the flow-through monitoring point;

[0051] The normalization unit is used to load the real-time working parameters and perform normalization processing on the real-time working parameters to obtain the normalized set.

[0052] Optionally, the real-time output calculation module includes:

[0053] an output coefficient determination unit, configured to load the normalized set and calculate the real-time output coefficient of the associated components of the flow monitoring point according to the normalized set;

[0054] a gradual anomaly identification unit, configured to capture gradual anomaly parameters in the normalized set based on a CUSUM algorithm;

[0055] The parameter integration unit is used to integrate the gradually changing abnormal parameter and the real-time output coefficient into the real-time output set.

[0056] Compared with the existing technology, the present invention solves the problem that the existing method only relies on limited parameters such as inlet steam temperature and extraction steam pressure, and does not integrate the three-dimensional model of the flow part and multi-source dynamic data, and has the following beneficial effects:

[0057] 1. When evaluating the flow abnormality of the flow section of the steam turbine unit, not only parameters such as pressure and temperature are taken into consideration, but also key indicators such as real-time output coefficient and node flow value are introduced. This more comprehensively reflects the operating status of the flow section of the steam turbine unit, enhances the accuracy and reliability of the evaluation, can adapt to changes in different operating conditions of the steam turbine, and timely capture gradual abnormalities in flow efficiency, providing operators with more accurate flow efficiency evaluation and abnormality warnings.

[0058] 2. When normalizing the real-time working parameters, the data discreteness of the real-time working parameters is calculated, and the data discreteness is used to determine the smoothness of the real-time working parameters. Based on the data discreteness, personalized noise reduction filtering is performed on the real-time working parameters. Among them, the adaptive filtering algorithm can dynamically adjust the filter parameters when processing high-discrete data, better track signal changes, and remove complex and variable noise. The high-order wavelet denoising function can retain signal details while reducing noise when dealing with low-discrete data, thanks to its excellent time-frequency localization characteristics. By selecting an appropriate noise reduction algorithm based on the data discreteness, the targeted signal processing can be enhanced, the noise reduction effect can be improved, and the parameters can be ultimately mapped to a uniform distribution (rather than a normal distribution), thereby solving the distortion problem of the traditional standard score (Z-score) for skewed data (such as the leakage of the exponential distribution).

[0059] 3. When adaptively filtering real-time operating parameters based on an adaptive filtering algorithm, cubic spline interpolation is used to fill missing or outliers, ensuring time series continuity and data integrity. Dynamic adjustment of the digital low-pass filter coefficients based on a Gaussian weighting algorithm enables the filter to adapt to non-stationary noise environments, thereby reducing redundant calculations in fixed-threshold filters. Subsequent secondary filtering further eliminates residual noise and significantly improves the signal-to-noise ratio.

[0060] 4. By applying the intra-stage loss decomposition algorithm, various losses in the flow process can be accurately allocated to each associated component, thereby more accurately determining each associated component's contribution to the overall flow efficiency, namely the real-time output coefficient, improving the accuracy and credibility of the evaluation results. By using the CUSUM algorithm to capture gradually changing abnormal parameters in the normalized set and integrating these gradually changing abnormal parameters with the real-time output coefficient into a real-time output set, it is helpful to promptly identify abnormal change trends at flow monitoring points, providing a timely and comprehensive data foundation for subsequent abnormal warning and diagnosis, and enhancing monitoring and early warning capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0062] Figure 1 A flow chart of a method for online evaluation of flow efficiency and abnormality warning of a steam turbine provided by one embodiment of the present invention;

[0063] Figure 2 A flowchart of normalization processing of real-time working parameters provided by another embodiment of the present invention;

[0064] Figure 3 A flowchart of performing adaptive filtering processing on real-time working parameters based on an adaptive filtering algorithm provided in another embodiment of the present invention;

[0065] Figure 4 A flowchart of calculating the real-time output coefficient of the associated components of the flow monitoring point according to the normalized set provided in another embodiment of the present invention;

[0066] Figure 5 A flowchart for comprehensively evaluating the flow anomaly degree of the steam turbine unit flow section by combining the node flow values ​​and the three-dimensional model of the steam turbine unit flow section provided by another embodiment of the present invention;

[0067] Figure 6 A schematic structural diagram of a steam turbine flow efficiency online evaluation and abnormality warning system provided in another embodiment of the present invention. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, each embodiment of the present invention will be described in detail below with reference to the accompanying drawings. However, it will be understood by those skilled in the art that in each embodiment of the present invention, many technical details are provided to enable the reader to better understand the present invention. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present invention can be implemented. The division of the following embodiments is for convenience of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined and referenced with each other under the premise that there is no contradiction.

[0069] The existing method for diagnosing abnormal flow efficiency of steam turbines only relies on limited parameters such as inlet steam temperature and extraction steam pressure, and does not integrate the three-dimensional model of the flow part with multi-source dynamic data (such as vibration and wear). This leads to incomplete fault feature extraction, difficulty in adapting to dynamic changes in unit load (such as peak load regulation), and a high missed diagnosis rate. To this end, one embodiment of the present invention provides a method for online evaluation and abnormal warning of flow efficiency of steam turbines, the process of which is as follows: Figure 1 As shown, it includes steps S10 to S60. Figure 1 Steps S10 to S60 will be described in detail.

[0070] Step S10: deploying flow monitoring points based on the three-dimensional model of the flow section of the steam turbine unit, collecting real-time operating parameters of the flow section of the steam turbine unit through the flow monitoring points, and normalizing the real-time operating parameters to obtain a normalized set.

[0071] Step S20: load the normalized set, calculate the real-time output coefficients of the associated components of the flow monitoring point based on the normalized set, capture the gradually changing abnormal parameters in the normalized set based on the CUSUM algorithm, and integrate the gradually changing abnormal parameters and the real-time output coefficients into the real-time output set.

[0072] Step S30: obtaining a real-time output set, taking the real-time output set as a priori probability, combining the real-time working parameters in the normalized set, evaluating the node flow value of the flow monitoring point based on a pre-built online evaluation model, and feeding back the node flow value.

[0073] Step S40 , loading the node flow values ​​and the three-dimensional model of the flow portion of the steam turbine unit, and comprehensively evaluating the flow abnormality of the flow portion of the steam turbine unit by combining the node flow values ​​and the three-dimensional model of the flow portion of the steam turbine unit.

[0074] Step S50: Determine whether the flow abnormality exceeds a preset abnormality threshold, wherein the preset abnormality threshold can be set to any value between 0.3 and 0.45.

[0075] Step S60: If the flow abnormality exceeds the preset abnormality threshold, a flow abnormality warning instruction is triggered. The flow abnormality warning instruction may include the specific value of the flow abnormality and its corresponding abnormality type, warning level, and response measures.

[0076] like Figure 1 As shown, the method for online evaluation of flow efficiency and abnormality warning of a steam turbine provided in this embodiment further includes step S70.

[0077] Step S70: If the flow abnormality does not exceed the preset abnormality threshold, the flow abnormality is fed back.

[0078] The method for online evaluation and abnormal warning of flow efficiency of a steam turbine provided by an embodiment of the present invention solves the problem that the existing method only relies on limited parameters such as inlet steam temperature and extraction steam pressure, and does not integrate the three-dimensional model of the flow part and multi-source dynamic data, compared with the existing technology. When evaluating the flow abnormality of the flow part of the steam turbine unit, not only parameters such as pressure and temperature are considered, but also key indicators such as real-time output coefficient and node flow value are introduced, which more comprehensively reflects the operating status of the flow part of the steam turbine unit, enhances the accuracy and reliability of the evaluation, can adapt to changes in different operating conditions of the steam turbine, and timely capture the gradual abnormality of the flow efficiency, providing operators with more accurate flow efficiency evaluation and abnormal warning.

[0079] For example, in step S10, the real-time working parameters are normalized to obtain a normalized set, including steps S101 to S106. Figure 2 , and steps S101 to S106 are described in detail.

[0080] Step S101: Load real-time operating parameters and process missing values ​​and abnormal values ​​of the real-time operating parameters, wherein the real-time operating parameters include thermodynamic parameters, geometric parameters, flow characteristic parameters, and monitoring point coordinate parameters.

[0081] Step S102 performs spectral analysis on the real-time operating parameters to calculate the data dispersion of the real-time operating parameters. The data dispersion is then used to determine the smoothness of the real-time operating parameters. Calculating data dispersion based on spectral analysis allows for quantitative assessment of data fluctuation and smoothness, providing a basis for selecting appropriate processing methods. When performing spectral analysis on the real-time operating parameters, metrics such as the power spectral density of the real-time operating parameters can be calculated, and the data dispersion can be determined based on these metrics.

[0082] Step S103: determining whether the data discreteness of the real-time working parameter exceeds a preset discrete threshold.

[0083] In step S104, if the data discreteness exceeds a preset discreteness threshold, adaptive filtering is performed on the real-time operating parameters using an adaptive filtering algorithm. When the data discreteness exceeds the preset discreteness threshold, the adaptive filtering algorithm is used to adaptively filter the real-time operating parameters, adjusting the step size in real time to suppress time-varying noise and improve the signal-to-noise ratio.

[0084] In step S105, if the data discreteness does not exceed the preset discreteness threshold, a high-order wavelet denoising function is used to denoise the real-time operating parameters. When the data discreteness does not exceed the preset discreteness threshold, denoising the real-time operating parameters using the high-order wavelet denoising function combined with a soft threshold can effectively preserve detailed features. The high-order wavelet denoising function may be a 6th-order wavelet denoising function.

[0085] Step S106 , integrating the real-time operating parameters after denoising by the adaptive filtering algorithm and the high-order wavelet denoising function, and mapping the real-time operating parameters after denoising to the corresponding uniform normal distribution using the quantile normalization method to obtain a normalized set.

[0086] By mapping the real-time operating parameters after noise reduction to a uniform normal distribution, we can eliminate dimensional differences between different parameters, making the data comparable and facilitating subsequent analysis and modeling. Quantile normalization can be used to transform the data into a form with a uniform distribution, ensuring data consistency.

[0087] When normalizing the real-time working parameters, this embodiment calculates the data discreteness of the real-time working parameters, uses the data discreteness to determine the smoothness of the real-time working parameters, and performs personalized noise reduction filtering on the real-time working parameters based on the data discreteness. Among them, the adaptive filtering algorithm can dynamically adjust the filter parameters when processing high-discrete data, better track the changes in the signal, and remove complex and changeable noise. The high-order wavelet denoising function can, when dealing with low-discrete data, retain signal details while reducing noise by virtue of its excellent time-frequency localization characteristics. By selecting a suitable noise reduction algorithm based on the data discreteness, the pertinence of signal processing can be enhanced, the noise reduction effect can be improved, and the parameters can be finally mapped to a uniform distribution (rather than a normal distribution), thereby solving the distortion problem of Z-score for skewed data (such as the leakage of the exponential distribution).

[0088] For example, in step S104, adaptive filtering is performed on the real-time working parameters based on the adaptive filtering algorithm, including steps S201 to S204. Figure 3 , and steps S201 to S204 are described in detail.

[0089] Step S201 loads real-time operating parameters and performs digital low-pass filtering on the parameters using a digital low-pass filter. Digital low-pass filters can effectively remove high-frequency noise from the real-time operating parameters, retain low-frequency signals, and smooth data curves, providing cleaner data for subsequent processing. Step S201 loads the real-time operating parameters and designs a suitable digital low-pass filter to filter the real-time operating parameters to remove high-frequency noise components from the real-time operating parameters.

[0090] Step S202 uses cubic spline interpolation to reconstruct the filtered real-time operating parameters, obtaining parameter reconstruction data. Parameter estimates corresponding to the filtering threshold are then obtained, and the optimal combination weighted average method is used to obtain parameter replacement values. These replacement values ​​are then integrated into the real-time operating parameters to obtain an integrated parameter set. Using cubic spline interpolation to reconstruct the filtered real-time operating parameters complements missing data and improves data integrity.

[0091] Step S203: Load the integrated parameter set, determine the filter coefficients of the digital low-pass filter based on the data discreteness of the integrated parameter set in conjunction with a Gaussian weighting algorithm, and dynamically adjust the digital low-pass filter using the filter coefficients. By dynamically determining the filter coefficients based on the data discreteness of the integrated parameter set and the Gaussian weighting algorithm, the digital low-pass filter can adaptively adjust its parameters, better adapting to data changes and improving filtering effectiveness.

[0092] The filter coefficients are calculated using the following formula:

[0093]

[0094] Among them, L but , L0 represent the filter coefficient and initial coefficient respectively, f s 、f c Respectively represent the parameter frequency and parameter frequency range, λ1 and λ2 represent the first weighting coefficient and the second weighting coefficient based on the optimal combination weighted average method, Δf is the transition bandwidth of the digital low-pass filter, LN is the filter order of the digital low-pass filter, D L represents the data dispersion and SNR represents the filter signal-to-noise ratio and A t 、 σ(A t ) represent the working parameters, parameter mean, and parameter covariance of the digital low-pass filter at the current moment t, respectively, and f max 、f min are the maximum sampling frequency and the minimum sampling frequency respectively, and z is the number of noise signals in the parameter.

[0095] In step S204, the integrated parameter set is input and subjected to secondary dynamic filtering using a digital low-pass filter, outputting the filtered real-time operating parameters. This secondary dynamic filtering further removes noise, improves data smoothness and stability, and enhances data quality and reliability.

[0096] This implementation ensures time series continuity and data integrity by using cubic spline interpolation to fill missing or outliers when adaptively filtering real-time operating parameters based on an adaptive filtering algorithm. Dynamic adjustment of the digital low-pass filter coefficients based on a Gaussian weighting algorithm enables the filter to adapt to non-stationary noise environments, thereby reducing redundant calculations in fixed-threshold filters. Subsequent secondary filtering further eliminates residual noise and significantly improves the signal-to-noise ratio.

[0097] For example, in step S20, the real-time output coefficient of the associated components of the flow monitoring point is calculated based on the normalized set, including steps S301 to S304. Figure 4 , and steps S301 to S304 are described in detail.

[0098] Step S301: Load a normalized set and calculate the overall flow efficiency of the flow monitoring point based on the flow balance formula for that flow monitoring point. Loading the normalized set and calculating the overall flow efficiency based on the flow balance formula allows for a quantitative assessment of the overall performance of the flow monitoring point, providing basic data support for subsequent efficiency correction and output coefficient calculation.

[0099] Exemplarily, step S301, calculating the throughflow comprehensive efficiency of the throughflow monitoring point based on the throughflow balance formula of the throughflow monitoring point, includes: calculating the throughflow comprehensive efficiency according to the following formula:

[0100]

[0101] Among them, η all represents the overall flow efficiency, P is the number of associated components at the flow monitoring point, G P , Q net They represent the air flow rate and low-level heating value at the flow monitoring point, respectively. gr 、D zr They represent the superheated gas flow rate and reheated gas flow rate at the flow monitoring point, respectively. gr 、h fw Represents superheated steam enthalpy and feed steam enthalpy, h" zr 、h' zr They represent the reheat steam outlet enthalpy and reheat steam inlet enthalpy at the flow monitoring point respectively.

[0102] Step S302 loads the comprehensive flow efficiency of the flow monitoring points, traverses the normalized set, calculates the loss reduction coefficients for the flow monitoring points in the normalized set, corrects the comprehensive flow efficiency based on the loss reduction coefficients, and outputs the corrected comprehensive flow efficiency. By traversing the normalized set to calculate the loss reduction coefficients and correcting the comprehensive flow efficiency accordingly, the actual operating efficiency of the flow monitoring points can be more accurately reflected, the influence of loss factors can be eliminated, and the evaluation accuracy can be improved.

[0103] Step S303 , traversing the normalized set, and determining the associated comprehensive efficiency of the flow monitoring point based on the number of associated components of the flow monitoring point and the loss reduction coefficient of the associated components.

[0104] Step S304 obtains the corrected overall flow efficiency and the associated overall efficiency. Based on these efficiency and the associated overall efficiency, combined with the intra-stage loss decomposition algorithm, the real-time output coefficients of the components associated with the flow monitoring point are determined. By combining the corrected overall flow efficiency and the associated overall efficiency and applying the intra-stage loss decomposition algorithm to determine the real-time output coefficients, the real-time output coefficients of the components associated with the flow monitoring point can be accurately determined, enabling accurate quantification of the output contributions of the associated components and providing strong support for operational status assessment and optimization.

[0105] This implementation utilizes an intra-stage loss decomposition algorithm to precisely allocate various losses during the flow process to each associated component, thereby more accurately determining each associated component's contribution to overall flow efficiency, i.e., the real-time output coefficient. This improves the accuracy and credibility of the assessment results. By utilizing the CUSUM algorithm to capture gradually changing abnormal parameters within the normalized set and integrating these gradually changing abnormal parameters with the real-time output coefficient into a real-time output set, it facilitates the timely identification of abnormal trends in flow monitoring points, providing a timely and comprehensive data foundation for subsequent abnormality warning and diagnosis, and enhancing monitoring and early warning capabilities.

[0106] For example, the objective function of the online evaluation model is expressed as:

[0107]

[0108] Where ΔP lt It represents the output value of the objective function of the online evaluation model and is also the node flow value. w 、ω j are the power output value and the current gain factor of the current monitoring point respectively. P is the number of associated components of the current monitoring point and also the number of monitoring points of the current monitoring point. ΔW j 、T j Respectively represent the power change and inertia response time, is the mean of the gradual abnormal parameter, x is the gradual abnormal parameter, and κ represents the real-time output coefficient.

[0109] For example, step S40 is to load the node flow value and the three-dimensional model of the flow part of the steam turbine unit, and comprehensively evaluate the flow abnormality of the flow part of the steam turbine unit by combining the node flow value and the three-dimensional model of the flow part of the steam turbine unit, including steps S401 to S403. Figure 5 , and steps S401 to S403 are described in detail.

[0110] Step S401: Loading node flow values ​​and a three-dimensional model of the flow portion of the steam turbine unit, and determining a real-time flow anomaly point based on the node flow values, the number of monitoring points in the three-dimensional model of the flow portion of the steam turbine unit, and the real-time output coefficient. Specifically, determining a real-time flow anomaly point can include loading the node flow values ​​and the three-dimensional model of the flow portion of the steam turbine unit, and determining the real-time flow anomaly point based on the number of monitoring points and the real-time output coefficient.

[0111] Step S402 loads the real-time outlier points and their corresponding node outlier values, and determines the point-by-point outlier degree for each of the real-time outlier points based on a loss-term quantification algorithm. Determining the point-by-point outlier degree based on the loss-term quantification algorithm allows for a quantitative assessment of the degree of anomaly at each outlier point, providing a detailed quantitative basis for subsequent calculation of the comprehensive outlier degree. To determine the point-by-point outlier degree, the real-time outlier points and their corresponding node outlier values ​​are loaded, and the loss-term quantification algorithm is applied to calculate the point-by-point outlier degree for each outlier point.

[0112] Step S403 obtains the point-by-point anomaly degrees of at least one set of real-time flow anomaly points, and performs a weighted summation of the point-by-point anomaly degrees of the real-time flow anomaly points based on the flow anomaly point weights to obtain the flow anomaly degree of the flow anomaly portion of the steam turbine unit. By obtaining the point-by-point anomaly degrees of multiple flow anomaly points and performing a weighted summation based on the flow anomaly point weights, the importance and contribution of each anomaly point can be comprehensively considered to determine a comprehensive and accurate flow anomaly degree of the flow anomaly portion of the steam turbine unit, providing operators with intuitive and reliable anomaly assessment results.

[0113] Another embodiment of the present invention provides a system for online evaluation and abnormal warning of steam turbine flow efficiency, such as Figure 6 As shown, it includes a data acquisition module 100, a real-time output calculation module 200, a node flow calculation module 300, and an evaluation and early warning module 400.

[0114] The data acquisition module 100 is used to deploy flow monitoring points based on the three-dimensional model of the flow part of the steam turbine unit, collect real-time operating parameters of the flow part of the steam turbine unit through the flow monitoring points, normalize the real-time operating parameters, and obtain a normalized set.

[0115] The real-time output calculation module 200 is used to load the normalized set, calculate the real-time output coefficients of the associated components of the flow monitoring point based on the normalized set, and capture the gradually changing abnormal parameters in the normalized set based on the CUSUM algorithm, and integrate the gradually changing abnormal parameters and the real-time output coefficients into the real-time output set.

[0116] The node flow calculation module 300 is used to obtain the real-time output set, use the real-time output set as the prior probability, combine the real-time working parameters in the normalized set, evaluate the node flow value of the flow monitoring point based on the pre-built online evaluation model, and feedback the node flow value.

[0117] The evaluation and early warning module 400 is used to load the node flow value and the three-dimensional model of the flow part of the turbine unit, and comprehensively evaluate the flow abnormality of the flow part of the turbine unit in combination with the node flow value and the three-dimensional model of the flow part of the turbine unit, and determine whether the flow abnormality exceeds the preset abnormality threshold; if the flow abnormality exceeds the preset abnormality threshold, the flow abnormality early warning instruction is triggered.

[0118] For example, Figure 6 As shown, the data collection module 100 includes a monitoring point deployment unit 110 , a monitoring point collection unit 120 , and a normalization unit 130 .

[0119] The monitoring point deployment unit 110 is used to deploy flow monitoring points based on the three-dimensional model of the flow portion of the steam turbine unit.

[0120] The monitoring point acquisition unit 120 is used to collect real-time operating parameters of the flow-through part of the steam turbine unit through the flow-through monitoring points.

[0121] The normalization unit 130 is used to load the real-time operating parameters and normalize the real-time operating parameters to obtain a normalized set.

[0122] For example, Figure 6 As shown, the real-time output calculation module 200 includes an output coefficient determination unit 210 , a gradual abnormality identification unit 220 , and a parameter integration unit 230 .

[0123] The output coefficient determination unit 210 is used to load the normalized set and calculate the real-time output coefficients of the associated components of the flow monitoring point according to the normalized set.

[0124] The gradual change anomaly identification unit 220 is used to capture the gradual change anomaly parameters in the normalized set based on the CUSUM algorithm.

[0125] The parameter integration unit 230 is used to integrate the gradually changing abnormal parameters and the real-time output coefficient into a real-time output set.

[0126] The turbine flow efficiency online evaluation and abnormal warning system provided in the embodiment of the present invention can be used to implement the turbine flow efficiency online evaluation and abnormal warning method described in the above embodiment of the present invention. The specific implementation method of the turbine flow efficiency online evaluation and abnormal warning system can be found in the turbine flow efficiency online evaluation and abnormal warning method provided in the embodiment of the present invention, and will not be repeated here.

[0127] Compared with the existing technology, the online evaluation and abnormal warning system for turbine flow efficiency provided by the embodiment of the present invention solves the problem that the existing method only relies on limited parameters such as inlet steam temperature and extraction pressure, and does not integrate the three-dimensional model of the flow part and multi-source dynamic data. When evaluating the flow abnormality of the flow part of the turbine unit, not only parameters such as pressure and temperature are considered, but also key indicators such as real-time output coefficient and node flow value are introduced, which more comprehensively reflects the operating status of the flow part of the turbine unit, enhances the accuracy and reliability of the evaluation, can adapt to changes in different operating conditions of the turbine, and timely capture the gradual abnormality of the flow efficiency, providing operators with more accurate flow efficiency evaluation and abnormal warning.

[0128] Another embodiment of the present invention provides a computer-readable storage medium storing computer program instructions, which are executable by a processor. When executed, the computer program instructions implement the method for online evaluation and abnormality warning of steam turbine flow efficiency provided in any of the above embodiments.

[0129] Another embodiment of the present invention also provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, it implements the online evaluation and abnormal warning method of turbine flow efficiency provided by any of the above embodiments.

[0130] Among them, the memory, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as the program instructions / modules corresponding to the method for online evaluation of steam turbine flow efficiency and abnormality warning in the above-mentioned embodiment of the present application. The memory may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function. The data storage area can store data used and created by the method for online evaluation of steam turbine flow efficiency and abnormality warning, etc. In addition, the memory may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory may optionally include a memory remotely located relative to the processor, and these remote memories may be connected to the local module via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network and a combination thereof.

[0131] Finally, it should be noted that the computer-readable storage medium (e.g., memory) can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. By way of example and not limitation, non-volatile memory can include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can include random access memory (RAM), which can act as an external cache memory. By way of example and not limitation, RAM can be obtained in a variety of forms, such as synchronous RAM (DRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and interface dynamic random access memory (DRRAM). The storage devices of the disclosed aspects are intended to include, but are not limited to, these and other suitable types of memory.

[0132] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present invention, and that in actual applications, various changes may be made in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for online evaluation and abnormal warning of steam turbine flow efficiency, characterized in that: The method comprises: Deploying flow monitoring points based on a three-dimensional model of the flow section of the steam turbine unit, collecting real-time operating parameters of the flow section of the steam turbine unit through the flow monitoring points, and normalizing the real-time operating parameters to obtain a normalized set; Loading the normalized set, calculating the real-time output coefficients of the associated components of the flow monitoring point according to the normalized set, capturing the gradually changing abnormal parameters in the normalized set based on a CUSUM algorithm, and integrating the gradually changing abnormal parameters and the real-time output coefficients into a real-time output set; Acquire the real-time output set, use the real-time output set as a priori probability, combine with the real-time operating parameters in the normalized set, evaluate the node flow value of the flow monitoring point based on a pre-built online evaluation model, and feed back the node flow value; loading the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, comprehensively evaluating the flow abnormality of the flow portion of the steam turbine unit by combining the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, and determining whether the flow abnormality exceeds a preset abnormality threshold; If the flow abnormality exceeds the preset abnormality threshold, a flow abnormality warning instruction is triggered.

2. The method according to claim 1, characterized in that The normalization processing of the real-time working parameters to obtain a normalized set includes: Loading the real-time operating parameters and processing missing values ​​and abnormal values ​​of the real-time operating parameters, wherein the real-time operating parameters include thermodynamic parameters, geometric parameters, flow characteristic parameters, and monitoring point coordinate parameters; Performing spectrum analysis on the real-time operating parameters, calculating data dispersion of the real-time operating parameters, and determining smoothness of the real-time operating parameters using the data dispersion; Determining whether the data discreteness of the real-time operating parameter exceeds a preset discrete threshold; If the data discreteness exceeds the preset discrete threshold, performing adaptive filtering processing on the real-time working parameters based on an adaptive filtering algorithm; If the data discreteness does not exceed the preset discrete threshold, a high-order wavelet noise reduction function is used to perform noise reduction processing on the real-time working parameters; The real-time operating parameters after denoising by an adaptive filtering algorithm and a high-order wavelet denoising function are integrated, and the real-time operating parameters after denoising are mapped to a corresponding uniform normal distribution using a quantile normalization method to obtain the normalized set.

3. The method according to claim 2, characterized in that The adaptive filtering process of the real-time working parameters based on the adaptive filtering algorithm includes: Loading the real-time operating parameters, and performing digital low-pass filtering on the real-time operating parameters based on a digital low-pass filter; Using a cubic spline interpolation method to interpolate and reconstruct the filtered real-time operating parameters to obtain parameter reconstruction data, obtaining parameter estimation values ​​corresponding to the filtering threshold, combining with an optimal combination weighted average method to obtain parameter replacement values, and integrating the parameter replacement values ​​into the real-time operating parameters to obtain a parameter integration set; Loading the parameter integration set, determining filter coefficients of the digital low-pass filter based on the data discreteness of the parameter integration set in combination with a Gaussian weighting algorithm, and dynamically adjusting the digital low-pass filter using the filter coefficients; The filter coefficient is calculated by the following formula: Among them, Lbut and L0 represent the filter coefficient and initial coefficient respectively, f s 、f c Respectively represent the parameter frequency and parameter frequency range, λ1 and λ2 represent the first weighting coefficient and the second weighting coefficient based on the optimal combination weighted average method, Δf is the transition bandwidth of the digital low-pass filter, LN is the filter order of the digital low-pass filter, D L represents the data dispersion and SNR represents the filter signal-to-noise ratio and A t 、 σ(A t ) represent the working parameters, parameter mean, and parameter covariance of the digital low-pass filter at the current moment t, fmax and fmin are the maximum sampling frequency and the minimum sampling frequency, respectively, and z is the number of noise signals in the parameter; Taking the parameter integration set as input, the digital low-pass filter is used to perform secondary dynamic filtering on the parameter integration set, and the real-time working parameters after secondary filtering are output.

4. The method according to claim 1, wherein Calculating the real-time output coefficient of the associated components of the flow monitoring point according to the normalized set includes: Loading the normalized set, and calculating the comprehensive flow efficiency of the flow monitoring point based on the flow balance formula of the flow monitoring point; Loading the throughflow comprehensive efficiency of the throughflow monitoring point, traversing the normalized set, calculating the loss reduction coefficient of the throughflow monitoring point in the normalized set, correcting the throughflow comprehensive efficiency based on the loss reduction coefficient, and outputting the corrected throughflow comprehensive efficiency; Traversing the normalized set, and determining the associated comprehensive efficiency of the flow monitoring point based on the number of associated components of the flow monitoring point and the associated component loss reduction coefficient; The corrected throughflow comprehensive efficiency and the associated comprehensive efficiency are obtained, and the real-time output coefficient of the associated components of the throughflow monitoring point is determined based on the throughflow comprehensive efficiency and the associated comprehensive efficiency in combination with an intra-stage loss decomposition algorithm.

5. The method according to claim 4, characterized in that The calculating the comprehensive flow efficiency of the flow monitoring point based on the flow balance formula of the flow monitoring point includes: The overall flow efficiency is calculated according to the following formula: Wherein, ηall represents the overall flow efficiency, P is the number of associated components of the flow monitoring point, GP and Qnet represent the airflow rate and low calorific value of the flow monitoring point, respectively, Dgr and Dzr represent the superheated gas flow rate and reheated gas flow rate of the flow monitoring point, respectively, hgr and hfw represent the superheated steam enthalpy and feed steam enthalpy, respectively, and hz"r 、 hz'r represent the reheat steam outlet enthalpy and the reheat steam inlet enthalpy of the flow monitoring point respectively.

6. The method according to claim 5, characterized in that The objective function of the online evaluation model is expressed as: Wherein, ΔPlt represents the output value of the objective function of the online evaluation model, and is also the node flow value, H w 、ω j are the power output value and the flow gain factor of the flow monitoring point, respectively. P is the number of associated components of the flow monitoring point and also the number of monitoring points of the flow monitoring point. ΔWj and Tj represent the power change and the inertia response time, respectively. is the mean of the gradual abnormal parameter, x is the gradual abnormal parameter, and κ represents the real-time output coefficient.

7. The method according to claim 5, characterized in that The step of loading the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, and comprehensively evaluating the flow abnormality of the flow portion of the steam turbine unit by combining the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, includes: Loading the node flow value and the three-dimensional model of the flow portion of the steam turbine unit, and determining a real-time flow abnormality point based on the node flow value, the number of monitoring points in the three-dimensional model of the flow portion of the steam turbine unit, and the real-time output coefficient; Loading the real-time abnormal flow point and the node flow value corresponding to the real-time abnormal flow point, and determining the point abnormality degree of the real-time abnormal flow point based on a loss term quantification algorithm; Obtaining at least one group of point abnormality degrees of the real-time flow abnormality points, performing weighted summation of the point abnormality degrees of the real-time flow abnormality points based on the flow abnormality point weights, and obtaining the flow abnormality degree of the flow abnormality portion of the steam turbine unit.

8. A steam turbine flow efficiency online evaluation and abnormality warning system, characterized in that: The system comprises: a data acquisition module, configured to deploy flow monitoring points based on a three-dimensional model of the flow passage of the steam turbine unit, collect real-time operating parameters of the flow passage of the steam turbine unit through the flow monitoring points, and normalize the real-time operating parameters to obtain a normalized set; a real-time output calculation module, configured to load the normalized set, calculate the real-time output coefficients of the associated components of the flow monitoring point according to the normalized set, capture the gradually changing abnormal parameters in the normalized set based on a CUSUM algorithm, and integrate the gradually changing abnormal parameters and the real-time output coefficients into a real-time output set; a node flow calculation module, configured to obtain the real-time output set, use the real-time output set as a priori probability, combine the real-time operating parameters in the normalized set, evaluate the node flow value of the flow monitoring point based on a pre-built online evaluation model, and provide feedback on the node flow value; An evaluation and early warning module is used to load the node flow value and the three-dimensional model of the flow part of the steam turbine unit, comprehensively evaluate the flow abnormality of the flow part of the steam turbine unit in combination with the node flow value and the three-dimensional model of the flow part of the steam turbine unit, and determine whether the flow abnormality exceeds a preset abnormality threshold; if the flow abnormality exceeds the preset abnormality threshold, trigger a flow abnormality early warning instruction.

9. The system according to claim 8, characterized in that The data acquisition module includes: A monitoring point deployment unit, used to deploy flow monitoring points based on a three-dimensional model of the flow section of the steam turbine unit; A monitoring point acquisition unit, configured to acquire real-time operating parameters of the flow-through portion of the steam turbine unit through the flow-through monitoring point; The normalization unit is used to load the real-time working parameters and perform normalization processing on the real-time working parameters to obtain the normalized set.

10. The system according to claim 8, wherein: The real-time output calculation module includes: an output coefficient determination unit, configured to load the normalized set and calculate the real-time output coefficient of the associated components of the flow monitoring point according to the normalized set; a gradual anomaly identification unit, configured to capture gradual anomaly parameters in the normalized set based on a CUSUM algorithm; The parameter integration unit is used to integrate the gradually changing abnormal parameter and the real-time output coefficient into the real-time output set.

Images