State monitoring and fault diagnosis method for power station equipment

Abstract

The invention provides a state monitoring and fault diagnosis method for power station equipment, and relates to the technical field of power station monitoring. Acquiring a vibration signal; grid division is carried out based on the power station medium space distribution image, a power station medium load thermodynamic diagram is generated, and each grid of the power station medium load thermodynamic diagram is marked with a corresponding load intensity value; generating a dynamic load index based on the target equipment and a power station medium load thermodynamic diagram; obtaining a short-term degradation rate of each target device; a dynamic risk score map is generated based on the short-term degradation rate and device historical maintenance records. The problems of one-sidedness and inaccuracy of equipment health state assessment caused by the fact that the equipment working environment and potential risk factors cannot be comprehensively reflected in the prior art are effectively solved. Finally, the effects of improving the accuracy of power station equipment state monitoring, enhancing the reliability of fault prediction and optimizing the maintenance strategy are achieved.

Description

Technical Field

[0001] This application relates to the field of power plant monitoring technology, and in particular to a method for condition monitoring and fault diagnosis of power plant equipment. Background Technology

[0002] In modern power plant operation, equipment condition monitoring and fault diagnosis are crucial for ensuring the safe and reliable operation of the power system. Traditional power plant equipment monitoring methods often rely on periodic manual inspections and simple sensor data collection. This approach is not only inefficient but also struggles to achieve real-time monitoring and early fault warnings. With advancements in technology, research into power plant media (such as steam and water flow) and their impact on equipment has deepened, making it possible to improve the monitoring accuracy of power plant equipment using advanced sensing technologies and data analysis methods.

[0003] Existing monitoring systems typically focus on collecting single types of parameters, such as vibration signals or temperature changes, while neglecting the impact of the spatial distribution characteristics of the power plant medium on equipment health. This approach fails to comprehensively reflect the equipment's operating environment and its potential risk factors, leading to a one-sided and inaccurate assessment of equipment health status and reducing the effectiveness of fault diagnosis. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] To achieve the above objectives, this application proposes a method for condition monitoring and fault diagnosis of power plant equipment, comprising the following steps: Step 1: Continuously acquire real-time parameters of the power plant medium, including: spatial distribution image of the power plant medium; Step 2: Obtain the operating status data of multiple target devices affected by the power station medium: vibration signals; Step 3: Based on the spatial distribution image of the power plant medium, perform grid division to generate a heat map of the power plant medium load. Each grid of the heat map of the power plant medium load is labeled with a corresponding load intensity value. Step 4: Based on the target equipment and the power plant medium load heat map, generate a dynamic load index; Step 5: Based on the dynamic load index and vibration signal, obtain the short-term degradation rate of each target device; Step 6: Generate a dynamic risk scoring chart based on the short-term degradation rate and the equipment's historical maintenance records; Step 7: Generate diagnostic results based on the dynamic risk scoring map.

[0006] Furthermore, the power station medium is a material flow that moves continuously in space and exerts forces on external equipment.

[0007] Furthermore, the real-time parameters also include: power plant medium density and power plant medium thickness.

[0008] Furthermore, based on the spatial distribution image of the power plant medium, a grid is divided to generate a thermal map of the power plant medium load, including the following steps: Step 31: Perform grayscale processing on the spatial distribution image of the power station medium to obtain the grayscale value of each pixel; Step 32: Generate a thickness distribution map based on the grayscale value and the thickness of the power station medium; Step 33: Threshold segmentation is performed on the thickness distribution map to generate a binary mask map, wherein pixels with a mask value of 1 constitute the effective carrying area; Step 34: Multiply the binary mask image and the thickness distribution image pixel by pixel to obtain a corrected thickness distribution image composed of the effective bearing area; Step 35: Obtain the actual physical area corresponding to a single pixel; Step 36: Count the number of pixels with a mask value of 1 for each grid; obtain the effective area of ​​the grid based on the number of pixels and the actual physical area; Step 37: Based on the pre-constructed power plant medium motion velocity field, and combined with the power plant medium density, the power plant medium thickness corresponding to each pixel in the corrected thickness distribution map, and the effective area, obtain the mass flow rate passing through each grid per unit time and mark it as the load intensity value to generate a power plant medium load heat map.

[0009] Furthermore, based on the target equipment and the power plant medium load heat map, a dynamic load index is generated, including the following steps: Step 41: Obtain the equivalent load borne by each target device at the corresponding moment of the current frame; Step 42: Arrange the equivalent loads of consecutive frames in chronological order to form an equivalent load time series; Step 43: Construct a sliding window, and within the sliding window, determine the root mean square value and extreme value sequence of the equivalent load based on the equivalent load time series; Step 44: Determine the absolute difference based on each pair of adjacent extreme values ​​in the extreme value sequence; Step 45: Take the average value of all adjacent extreme value differences as the local peak-to-peak value of the sliding window; Step 46: Construct a composite volatility factor based on the equivalent load time series; Step 47: Generate a dynamic load index based on the composite fluctuation factor, local peak-to-peak value, and root mean square value.

[0010] Furthermore, based on the dynamic load index and vibration signal, the short-term degradation rate of each target device is obtained, including the following steps: Step 51: Based on the type of the target device, determine the characteristic frequency range triggered by its typical faults; Step 52: Perform bandpass filtering on the vibration signal to obtain the target vibration signal, which is composed of components within the characteristic frequency range; Step 53: Perform envelope analysis based on the target vibration signal to obtain the envelope signal; Step 54: Perform spectrum analysis based on the envelope signal to obtain the total energy within the characteristic frequency range, and label it as the characteristic frequency band energy; Step 55: Based on the characteristic frequency band energy and dynamic load index, obtain the coupling response strength at each moment and construct a coupling response strength sequence; Step 56: Perform linear least squares fitting based on the coupling response intensity sequence to obtain the trend slope that changes over time, and use the absolute value of the trend slope as the short-term degradation rate.

[0011] Furthermore, based on short-term degradation rates and historical equipment maintenance records, a dynamic risk scoring chart is generated, including the following steps: Step 61: If the short-term degradation rate is less than the first target value, the target device is determined to be in a healthy and stable stage. The first indicator is obtained based on the device's historical records and marked as the first score. Step 62: If the short-term degradation rate is greater than or equal to the first target value and less than the second target value, the target device is determined to be in the initial degradation stage. The second indicator is obtained based on the device's historical records, and the second score is obtained based on the second indicator and the short-term degradation rate. Step 63: If the short-term degradation rate is greater than or equal to the second target value, the target device is determined to be in the accelerated degradation stage. The third and fourth indicators are obtained based on the device's historical records. The third score is obtained based on the third and fourth indicators and the short-term degradation rate. Step 64: Map the first, second, and third scores of each target device to a preset color range and overlay them onto the power plant equipment spatial layout map to generate a dynamic risk scoring map.

[0012] Compared with existing technologies, this application provides a method for condition monitoring and fault diagnosis of power plant equipment, which generates a detailed thickness distribution map based on the grayscale values ​​and thickness of the medium spatial distribution image of the power plant. It fully utilizes the characteristic that grayscale values ​​are directly related to medium density and other key physical properties, ensuring accurate capture of the characteristics and distribution of the medium inside the power plant. This overcomes the limitation of traditional single-sensor data not being able to fully cover the equipment's operating environment, providing a solid foundation for subsequent analysis.

[0013] Secondly, by using mass flow rate as the load intensity value and generating a power plant medium load heat map, comprehensive monitoring of the medium flow status within the power plant is achieved. Unlike traditional methods that rely solely on single indicators such as temperature or vibration, this method can intuitively display the load intensity changes of various parts of the power plant, helping to quickly locate potential problem areas and take targeted measures. This not only improves monitoring efficiency but also significantly enhances the early warning capability for sudden failures, enabling maintenance personnel to more accurately understand the operating status of equipment and potential risks.

[0014] Furthermore, the generation of dynamic load indices and short-term degradation rates based on the power plant's dielectric load heat map, followed by the creation of a risk scoring map, is a key step in achieving accurate fault diagnosis. This allows for real-time monitoring of the power plant equipment's health status and dynamic adjustments to maintenance plans to address constantly changing operating conditions. The introduction of dynamic load indices and short-term degradation rates makes assessing the current state of power plant equipment and predicting its future performance more scientific and reasonable. The risk scoring map provides an intuitive way for maintenance personnel to quickly understand the overall risk level of the equipment, make timely decisions, and prevent major accidents.

[0015] In summary, through the aforementioned characteristics, the method of this application effectively solves the problems of one-sidedness and inaccuracy in equipment health status assessment caused by the inability of existing technologies to fully reflect the equipment's working environment and potential risk factors. Ultimately, it improves the accuracy of power plant equipment condition monitoring, enhances the reliability of fault prediction, and optimizes maintenance strategies, significantly improving the safety and economy of power plant operation while reducing economic losses and safety risks caused by equipment failures. Attached Figure Description

[0016] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A schematic flowchart illustrating a method for condition monitoring and fault diagnosis of power plant equipment provided in an embodiment of this application; Figure 2 A structural diagram of an artificial intelligence-based robotic claw control system for removing yellow leaves, provided in an embodiment of this application; Figure 3 This is a block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0018] The following describes a method for condition monitoring and fault diagnosis of power plant equipment according to an embodiment of this application, with reference to the accompanying drawings.

[0019] It should be noted that the execution subject of the condition monitoring and fault diagnosis method for power plant equipment in this application embodiment is a condition monitoring and fault diagnosis system for power plant equipment in this application embodiment. The condition monitoring and fault diagnosis system for power plant equipment can be configured in an electronic device so that the electronic device can perform condition monitoring and fault diagnosis functions.

[0020] like Figure 1 As shown, this method for condition monitoring and fault diagnosis of power plant equipment is particularly suitable for power plant equipment that uses a material flow that moves continuously in space and can exert forces on external equipment as its working medium. For example, such equipment may include, but is not limited to, steam turbines, gas turbines, and various fluid-driven power generation devices. This method can effectively improve the accuracy of condition monitoring and the efficiency of fault diagnosis for these devices. The method includes the following steps: Step 1: Continuously acquire real-time parameters of the power plant medium, including: spatial distribution image of the power plant medium; The power station medium is a material flow that moves continuously in space and exerts forces on external equipment.

[0021] By deploying high-speed industrial cameras, infrared thermal imagers, or laser scanning devices in key areas of the power plant, real-time parameters of the power plant medium are continuously collected; the real-time parameters include at least a spatial distribution image of the power plant medium, which can reflect the morphology, coverage, and local concentration changes of the medium in three-dimensional space.

[0022] Step 2: Obtain the operating status data of multiple target devices affected by the power station medium: vibration signal; the real-time parameters also include: power station medium density and power station medium thickness.

[0023] The target equipment is distributed along the medium transmission path or within the area of ​​action, including but not limited to key components such as boiler heating surface tube bundles, turbine blades, flue support structures, and water-cooled walls. In this embodiment, vibration signals are used as operating status data, which directly reflect the mechanical response characteristics of the equipment under the influence of factors such as medium impact, friction, or corrosion.

[0024] The real-time parameters also include the density and thickness of the power plant medium. Online density meters and thickness gauges installed along the medium flow path allow for continuous monitoring of changes in medium density and sediment thickness.

[0025] Step 3: Based on the spatial distribution image of the power plant medium, perform grid division to generate a heat map of the power plant medium load. Each grid of the heat map of the power plant medium load is labeled with a corresponding load intensity value.

[0026] Traditional monitoring methods typically focus only on the local responses of the equipment itself, such as vibration or temperature, while neglecting the non-uniform spatial distribution and dynamic characteristics of the external excitation source—the power plant medium. This application addresses this by refining the spatial distribution image of the power plant medium into a grid, and combining this with physical parameters such as medium density, thickness, and velocity field to calculate the mass flow rate passing through each grid region per unit time. This mass flow rate is then used as the load intensity value, generating an intuitive and quantitative thermal map of the power plant medium load. This not only realistically reproduces the spatial distribution characteristics of the forces exerted by the medium on the equipment but also achieves a physical mapping from visual images to mechanical loads, transforming the originally abstract medium flow state into quantifiable, traceable, and modelable engineering parameters. The specific implementation process is as follows: Step 31: Perform grayscale processing on the spatial distribution image of the power station medium to obtain the grayscale value of each pixel.

[0027] For the spatial distribution image of the power plant medium, the weighted average method is used to convert the color information of the RGB three channels into the grayscale information of the single channel, thereby obtaining the grayscale value of each pixel.

[0028] Step 32: Generate a thickness distribution map based on the grayscale value and the thickness of the power station medium.

[0029] There is a certain correlation between grayscale values ​​and medium thickness. Higher grayscale values ​​may correspond to thicker medium layers, because thicker medium layers absorb or scatter more light, resulting in the corresponding areas in the image appearing darker. However, this correlation is not linear, so appropriate conversion models need to be established based on the specific materials and application scenarios.

[0030] For example, in one embodiment, it is based on a water-cooled wall tube segment in a steam power plant. The material properties and operating environment of this tube segment have been thoroughly studied, and a conversion model between grayscale values ​​and medium thickness under specific conditions has been determined. Using the grayscale value of each pixel obtained in previous steps, combined with the conversion model, the grayscale value is converted into the corresponding medium thickness value.

[0031] The thickness value of each pixel is mapped to a preset color gradient, with thinner parts represented by blue, and gradually transitioning to green, yellow and then red as the thickness increases, visually demonstrating the thickness variation of the power plant medium throughout the observation area.

[0032] Step 33: Threshold segmentation is performed on the thickness distribution map to generate a binary mask map, wherein pixels with a mask value of 1 constitute the effective carrying area.

[0033] The thickness threshold is determined based on the physical properties of the power plant medium, equipment structural safety requirements, and engineering experience. This threshold is used to distinguish whether the medium has a minimum effective thickness sufficient to exert a significant mechanical effect on the target equipment. The thickness value of each pixel in the thickness distribution map is compared with the thickness threshold: if the thickness value of a pixel is greater than or equal to the thickness threshold, it is marked as 1 in the binary mask image; otherwise, it is marked as 0.

[0034] Within the effective load-bearing area, the thickness of the power plant medium reaches or exceeds the critical value that can exert a substantial load on the equipment structure, enabling it to participate in mass flow calculation and contribute to the dynamic load on the equipment. Conversely, areas with a mask value of 0 indicate that the medium is too thin, missing, or merely background noise, lacking effective mechanical transmission capability, and are therefore excluded in subsequent load intensity calculations. This binary mask map allows for precise focusing on the medium region that truly affects the equipment, avoiding interference from invalid data and laying the foundation for constructing a high-precision power plant medium load heat map.

[0035] Spatial distribution images of the medium provide intuitive visual support for understanding its specific location and layout throughout the power plant. Based on these images, a preliminary medium thickness distribution map is generated, showing the relative thickness of the medium in each area, which helps identify potential areas of abnormal load. Valid information is extracted from the preliminary thickness distribution map to create a revised thickness distribution map, thereby eliminating noise and non-critical information from the original data and accurately reflecting the impact of the medium on the equipment under actual operating conditions. Combining the spatial distribution image with the revised thickness distribution map for analysis not only provides a visual understanding of the specific layout and distribution of the medium but also accurately locates areas facing higher risks due to variations in medium thickness. This is of great significance for early detection of potential problems and the development of preventative maintenance measures.

[0036] Step 34: Multiply the binary mask image and the thickness distribution image pixel by pixel to obtain a corrected thickness distribution image composed of the effective bearing area.

[0037] Pixel values ​​at the same location in the two images are multiplied: if a pixel's value is 1 in the binary mask image, its original thickness value in the thickness distribution map is retained; if the mask value is 0, the pixel is set to 0 in the corrected image. The resulting corrected thickness distribution map retains only the thickness information of the effective carrying area, while the ineffective areas are cleared to zero.

[0038] Step 35: Obtain the actual physical area corresponding to a single pixel.

[0039] The field of view of the imaging device is pre-calibrated: that is, the size of the area covered by the power plant medium spatial distribution image captured by the device in the actual physical space. The actual physical area is calculated based on the size of the covered area and the resolution of the power plant medium spatial distribution image.

[0040] Step 36: Count the number of pixels with a mask value of 1 for each grid; obtain the effective area of ​​the grid based on the number of pixels and the actual physical area.

[0041] The corrected thickness distribution map (or its corresponding binary mask map) is divided into several spatial grids according to a preset grid division rule, with each grid covering a local region of the image. All pixels within each grid are traversed, and the number of pixels with a mask value of 1 is counted. This value represents the number of pixels belonging to the effective carrying area within that grid. Multiplying this number of pixels by the actual physical area yields the effective area of ​​that grid.

[0042] Step 37: Based on the pre-constructed power plant medium motion velocity field, and combined with the power plant medium density, the power plant medium thickness corresponding to each pixel in the corrected thickness distribution map, and the effective area, obtain the mass flow rate passing through each grid per unit time and mark it as the load intensity value to generate a power plant medium load heat map.

[0043] In specific applications, particle image velocimetry (PIV) technology is used to measure the flow of media in actual power plants to obtain velocity vectors at different locations; or computational fluid dynamics (CFD) software is used to simulate the velocity field distribution of the media based on the physical properties and boundary conditions of the media. This velocity field provides information on the velocity of the media within each spatial grid.

[0044] The power plant medium density, the thickness of the power plant medium corresponding to each pixel in the corrected thickness distribution map, and the effective area are normalized and multiplied to obtain the mass flow rate, which is then marked as the load intensity value.

[0045] Mass flow rate directly relates to the dynamic load magnitude of the power plant medium within a given area. A larger mass flow rate implies a greater energy transfer or material transport capacity, and therefore a higher load intensity. Quantifying the load situation of each grid in this way can intuitively reflect the energy or material distribution characteristics within the entire system.

[0046] Step 4: Generate a dynamic load index based on the target equipment and the power plant medium load heat map.

[0047] Step 41: Obtain the equivalent load borne by each target device at the corresponding moment of the current frame.

[0048] Based on the spatial layout and structural support relationship of the target equipment, a force response mapping model for the equipment is established. This model determines which target equipment each grid cell affects mechanically by analyzing the equipment's geometric position in space, installation method, and structural connection relationships.

[0049] A medium-device response matrix is ​​constructed, indexed by grid and device, recording the influence weight of each grid cell on each target device. The weights are determined by factors such as distance attenuation, angular projection, and structural stiffness: grids closer to the device and subject to positive impact contribute more, while those further away contribute less. For each target device, all grid cells are traversed, and the load intensity values ​​of that grid cell are weighted and accumulated according to the corresponding weight coefficients in the response matrix to obtain the equivalent load received by the device. This comprehensively reflects the total dynamic excitation exerted by the medium on the device in space, serving as an important input parameter for subsequent generation of dynamic load indices and assessment of degradation trends. This method achieves accurate mapping from spatially distributed medium loads to device-level equivalent loads, improving the physical consistency and engineering practicality of condition monitoring.

[0050] Step 42: Arrange the equivalent loads of consecutive frames in chronological order to form an equivalent load time series.

[0051] This embodiment synchronously acquires images of the power plant's medium spatial distribution at a fixed sampling frequency (e.g., 1 frame per second or 1 frame per 500 milliseconds). After processing each frame, the equivalent load values ​​obtained in chronological order are stored sequentially to form a one-dimensional sequence that evolves over time, i.e., the equivalent load time series.

[0052] Step 43: Construct a sliding window, and within the sliding window, determine the root mean square value and extreme value sequence of the equivalent load based on the equivalent load time series.

[0053] Set a sliding window of a fixed length. For example, take the data of 10 consecutive time points as a frame analysis window. The window length can be adjusted according to the device response characteristics and sampling frequency (such as 5 seconds or 10 seconds). This window starts from the starting position of the equivalent load time series and moves forward frame by frame with a step size of 1. Each time it moves, the features of the data within the current window are calculated once.

[0054] Within each sliding window, square all the equivalent load values within the window in sequence, find the arithmetic mean of each squared value, and then take the square root of this mean to obtain the root mean square value within this window. It reflects the overall energy level of the equivalent load during the window period and can effectively characterize the intensity fluctuation trend of the load received by the device.

[0055] Identify local extreme values within the same sliding window for constructing an extreme value sequence. The specific method is as follows: Traverse the equivalent load sequence within the window. The specific method is as follows: Traverse the equivalent load sequence within the window: If Li > Li-1 and Li > Li+1, Li is a local maximum value; if Li < Li-1 and Li < Li+1, Li is a local minimum value, where Li represents the i-th equivalent load.

[0056] Arrange all the identified local maximum and minimum values in chronological order to form an extreme value sequence.

[0057] Step 44, based on each pair of adjacent extreme values in the extreme value sequence, determine the absolute difference.

[0058] Step 45, take the average value of all the differences between adjacent extreme values as the local peak-to-peak value of the sliding window.

[0059] This average value comprehensively reflects the typical peak-valley span of the equivalent load fluctuation within the current sliding window. Compared with the single maximum peak-to-peak value, it can more robustly characterize the overall fluctuation intensity of the dynamic load on a short time scale and avoid evaluation biases caused by individual abnormal extreme values.

[0060] Step 46, construct a composite fluctuation factor based on the equivalent load time series.

[0061] The composite fluctuation factor is used to comprehensively reflect the short-term frequency characteristics and long-term fluctuation trends of the load received by the device to enhance the sensitivity to dynamic load changes. Perform a short-time Fourier transform on the equivalent load sequence within the current sliding window and extract its energy components within the frequency band of 0.5 - 5 Hz. This frequency band usually corresponds to the common mechanical vibration frequencies or medium flow pulsation frequencies of power station equipment. Calculate the ratio of the standard deviation to the mean of the energy within this frequency band to characterize the instability degree of the energy distribution within this frequency band, which is marked as the coefficient of variation.

[0062] The Hodrick-Prescott (HP) filter method is applied to the equivalent load sequence to separate the sequence into a trend component and a volatility component. The HP filter extracts the low-frequency trend by minimizing the smoothing penalty term; the remaining part is the higher-order volatility component. The sample entropy of the volatility component is calculated to quantify its complexity and randomness. A larger sample entropy indicates more unpredictable volatility and a potential unstable state of the system. The fundamental volatility factor is obtained by weighted summation of the sample entropy and the coefficient of variation.

[0063] If the local peak value of the current window exceeds 1.8-2 times the average of the previous three sliding windows, a significant load change is considered to have occurred. In this case, the composite volatility factor is set to 1.5 times the basic volatility factor. Otherwise, the basic volatility factor is defined as the composite volatility factor.

[0064] Step 47: Generate a dynamic load index based on the composite fluctuation factor, local peak-to-peak value, and root mean square value.

[0065] The dynamic load index is obtained by normalizing the composite fluctuation factor, local peak-to-peak value, and root mean square value, and then weighting and summing them.

[0066] In this application, the dynamic load index accurately characterizes the actual stress state of the target equipment under complex medium excitation, thereby effectively bridging the physical gap between the medium flow characteristics of the power plant and the equipment response. This index comprehensively reflects the energy level, fluctuation amplitude, and time-frequency complexity of the load, and can sensitively capture unsteady excitations caused by uneven medium distribution, sudden changes in flow velocity, or abnormal deposition, significantly improving the ability to identify early degradation causes of equipment.

[0067] Step 5: Based on the dynamic load index and vibration signal, obtain the short-term degradation rate of each target device.

[0068] Step 51: Based on the type of the target device, determine the characteristic frequency range triggered by its typical faults.

[0069] Determining the characteristic frequency range excited by typical faults of target equipment based on its type requires a pre-built equipment-fault-characteristic frequency knowledge base: collecting structural parameters and historical operating data of common target equipment in the power plant; secondly, combining technical documents provided by equipment manufacturers, industry standards, and fault maintenance records of the power plant over the years, compiling theoretical formulas or measured frequency bands of vibration characteristic frequencies corresponding to typical faults of each type of equipment.

[0070] Furthermore, by conducting fault simulation tests on similar equipment in the laboratory or on-site, collecting vibration signals and performing spectral analysis, the theoretical frequency range is verified and calibrated. Equipment type, fault mode, structural parameters, and corresponding characteristic frequency ranges are stored in a structured format in a database, forming a queryable knowledge base. In practical applications, the characteristic frequency range corresponding to typical faults of the current target equipment is automatically matched based on its type.

[0071] Step 52: Perform bandpass filtering on the vibration signal to obtain the target vibration signal, which is composed of components within the characteristic frequency range.

[0072] The characteristic frequency range of typical faults corresponding to the target equipment type is set as the passband of the bandpass filter. Digital filtering technology is used to filter the acquired vibration signal. The cutoff frequency of the filter is strictly limited to the above characteristic frequency range, effectively suppressing low-frequency mechanical noise and high-frequency random interference outside the passband. The target vibration signal obtained after filtering has its spectral energy mainly concentrated within this characteristic frequency range, representing the periodic or quasi-periodic impact response components excited by a specific fault mechanism. These components typically exhibit the following characteristics: regular impact pulses in the time domain; spectral peaks centered on the characteristic frequency in the frequency domain, often accompanied by modulation sidebands; and amplitudes that increase with fault development, indicating a clear deterioration trend.

[0073] Therefore, the target vibration signal is not a simple subset of the original signal, but a key signal component that focuses on the fault-sensitive frequency band and is rich in diagnostic information, providing high signal-to-noise ratio and high correlation input data for subsequent envelope analysis and energy extraction.

[0074] Step 53: Perform envelope analysis based on the target vibration signal to obtain the envelope signal.

[0075] Envelope analysis is used to extract low-frequency fault impact features hidden in high-frequency carrier waves. The target vibration signal is subjected to a Hilbert transform to obtain its analytic signal; the magnitude of this analytic signal is then calculated, which is the envelope signal.

[0076] When a device has a local defect, the rotating component generates a transient impact each time it passes the defect location. This impact excites high-frequency resonance in the device structure near its natural frequency (i.e., the frequency band of the target vibration signal). However, since the impact itself is low-frequency and periodic, its modulation effect causes the amplitude of the high-frequency resonance signal to vary with the impact rhythm. Through envelope analysis, the low-frequency modulation information can be demodulated from the high-frequency carrier wave to form an envelope signal.

[0077] Step 54: Perform spectral analysis based on the envelope signal to obtain the total energy within the characteristic frequency range, which is labeled as the characteristic frequency band energy.

[0078] Perform a Fast Fourier Transform on the envelope signal to transform it from the time domain to the frequency domain, obtaining the envelope spectrum. This spectrum will show significant peaks at the fault characteristic frequency and its harmonics. Focusing on the characteristic frequency range within this envelope spectrum, integrate or discretely sum the energy of all frequency components within this band (i.e., the square of the spectral amplitude) to obtain the total energy within that band, denoted as the characteristic frequency band energy.

[0079] The characteristic frequency band energy physically reflects the cumulative level of periodic impact intensity excited by a specific fault mechanism under the current operating condition of the equipment. Higher energy indicates a stronger modulation effect caused by the fault, and potentially more severe equipment degradation. Since the envelope signal effectively suppresses non-impact noise, this characteristic frequency band energy exhibits a high signal-to-noise ratio and strong fault correlation, significantly improving the accuracy of degradation trend assessment.

[0080] Step 55: Based on the characteristic frequency band energy and dynamic load index, obtain the coupling response strength at each moment and construct a coupling response strength sequence.

[0081] Since the characteristic frequency band energy originates from vibration signal processing, while the dynamic load index is calculated using a sliding window, their original time resolution and time points are often inconsistent. Therefore, time synchronization processing is required before fusion. A precise timestamp is appended to each calculated characteristic frequency band energy value and dynamic load index value. Because the dynamic load index reflects the comprehensive load state within a window period, the characteristic frequency band energy sequence is time-matched using the time point of the dynamic load index as a benchmark: for each dynamic load index sample, all characteristic frequency band energy values ​​are extracted within the time interval of its corresponding sliding window, and their average value is calculated, or the instantaneous value at the end of the window is taken as the synchronous vibration response index at that moment. Thus, the synchronous vibration response indices at all moments are integrated into a synchronous sample set.

[0082] The coupled response intensity at the current moment is obtained by normalizing the characteristic frequency band energy and the dynamic load index and multiplying them. The coupled response intensities at all moments are then integrated into a coupled response intensity sequence in chronological order. This sequence reflects the fault-related vibration response intensity of the target equipment under a unit external load excitation, effectively isolating the interference of load fluctuations on the vibration amplitude and highlighting the response anomalies caused by the equipment's own degradation.

[0083] Step 56: Perform linear least squares fitting based on the coupling response intensity sequence to obtain the trend slope that changes over time, and use the absolute value of the trend slope as the short-term degradation rate.

[0084] Using time (moment) as the independent variable and the coupling response intensity as the dependent variable, the best straight line is fitted using the linear least squares method. The absolute value of the slope is taken as the short-term degradation rate of the target equipment. As internal damage to the equipment gradually develops, even if the external load remains stable, its vibration response will continue to increase, causing the coupling response intensity to show an upward trend, with a positive slope. If the response temporarily decreases due to adjustments in operating conditions, but the overall degradation trend still exists, the absolute value can uniformly characterize the magnitude of the degradation rate, avoiding directional interference.

[0085] This short-term degradation rate has a clear time scale and can sensitively reflect recent changes in equipment health. Because it is based on a stimulus-response sequence, it effectively eliminates false degradation signals caused by load fluctuations, significantly improving the accuracy of the assessment.

[0086] By using the dynamic load index as a key input in the subsequent calculation of the coupled response intensity, the assessment of the short-term degradation rate no longer depends solely on the vibration signal itself, but is based on the matching relationship between external excitation and internal response. This significantly enhances the accuracy of fault diagnosis and effectively avoids misjudgment or omission due to ignoring the real working environment.

[0087] Step 6: Generate a dynamic risk scoring chart based on the short-term degradation rate and the equipment's historical maintenance records; Step 61: If the short-term degradation rate is less than the first target value, the target device is determined to be in a healthy and stable stage. The first indicator is obtained based on the device's historical records and marked as the first score.

[0088] A large number of short-term degradation rate samples of similar equipment under known health conditions were extracted from the power plant's historical database. Statistical analysis was performed on these samples, calculating their mean and standard deviation, and the sum of the mean and standard deviation was used as the first target value.

[0089] The primary indicator, the cumulative operating time percentage, is derived from the equipment's historical maintenance records. This percentage represents the proportion of time the target equipment has operated safely under its design conditions since the last maintenance or replacement, relative to its expected lifespan or typical maintenance cycle. The cumulative operating time percentage reflects the utilization level of the target equipment throughout its lifecycle: a lower percentage indicates greater remaining operational margin; a higher percentage suggests that even if the current condition is healthy, it is close to the maintenance window. Therefore, the cumulative operating time percentage is directly used as the primary score to reflect the long-term service status of the equipment in the dynamic risk scoring chart.

[0090] Step 62: If the short-term degradation rate is greater than or equal to the first target value and less than the second target value, the target device is determined to be in the initial degradation stage. The second indicator is obtained based on the device's historical records, and the second score is obtained based on the second indicator and the short-term degradation rate.

[0091] The initial degradation stage indicates that although the target equipment has not yet experienced a serious failure, a detectable trend of performance degradation has emerged, which requires attention and should be included in the preventive maintenance plan.

[0092] A sample set of short-term degradation rates was selected from the power plant operation and maintenance database based on the calculated rates of similar equipment during the triggering of "initial warnings" (such as the first exceedance of vibration amplitude limits, abnormal infrared temperature without shutdown, or subsequent maintenance confirming minor wear / cracks). Statistical analysis was performed on these samples, calculating the mean and standard deviation. The difference between the mean and standard deviation was used as the second target value. The second target value was 1.5-1.8 times the first target value.

[0093] The second indicator is obtained based on the equipment's historical maintenance records: the percentage of time since the last maintenance, which is the operating time since the last planned maintenance, as a percentage of the equipment's regular maintenance cycle.

[0094] The short-term degradation rate is normalized, and the difference between the normalized short-term degradation rate and the first target value is used to obtain the first deviation. The difference between the second target value and the first target value is used to obtain the target deviation. The ratio is calculated based on the first deviation and the target deviation to obtain the first ratio. The first ratio is summed with the second index to obtain the second score.

[0095] Step 63: If the short-term degradation rate is greater than or equal to the second target value, the target device is determined to be in the accelerated degradation stage. The third and fourth indicators are obtained based on the device's historical records, and a third score is obtained based on the third and fourth indicators and the short-term degradation rate.

[0096] The accelerated degradation stage indicates that the internal damage to the target equipment is rapidly expanding, posing a high risk of failure, and requires priority for maintenance or intervention.

[0097] Historical fault severity index and overhaul interval are extracted based on equipment historical records. The historical fault severity index reflects the severity of the consequences of faults occurring in the equipment during past operating cycles. This index is quantified by integrating historical maintenance work orders, downtime, fault type (e.g., cracks, fractures, leaks), and the level of impact on the system (e.g., whether it caused unplanned shutdowns or involved safety incidents). It is typically normalized to a value between 0 and 1: 0 indicates no fault has ever occurred, and 1 indicates a catastrophic fault has occurred. For example, a boiler tube bundle that experienced two minor leaks (without shutdown) in the past three years might have a historical fault severity index of 0.25; while another turbine that caused a plant-wide shutdown due to blade fracture might have an index as high as 0.85. The overhaul interval refers to the ratio of operating time since the last major overhaul to the equipment's typical overhaul cycle. For example, if a piece of equipment has a standard overhaul cycle of 24,000 hours and has currently operated for 20,000 hours, then the overhaul interval percentage is approximately 20,000 / 24,000 ≈ 83%. The higher this indicator, the closer the equipment is to the end of its design life, the lower its structural margin, and the weaker its tolerance for current degradation trends.

[0098] Starting with the second target value (corresponding to 0 points), and setting an upper limit for the project as a baseline of 1 point, any excess is truncated or saturated proportionally to obtain the degradation urgency score (0–1 points). A weighted fusion strategy is used to calculate the third score: the degradation urgency score has the highest weight (e.g., 50%), as it reflects the most urgent current state; the overhaul interval has the second highest weight (e.g., 30%), reflecting the equipment's remaining lifespan constraints; the historical failure severity index accounts for 20%, used to adjust risk expectations (equipment with many historical problems requires more vigilance). The final third score integrates the current degradation rate, remaining service margin, and historical reliability dimensions, accurately depicting the high-risk state of equipment in the accelerated degradation stage.

[0099] Step 64: Map the first, second, and third scores of each target device to a preset color range and overlay them onto the power plant equipment spatial layout map to generate a dynamic risk scoring map.

[0100] Establish standardized color mapping rules for ratings: for example, map 0–0.3 points to green (indicating low risk, healthy and stable), 0.31–0.7 points to yellow (indicating medium risk, early degradation), and 0.71–1 points to red (indicating high risk, accelerated degradation). This color range can be adjusted according to actual operation and maintenance strategies, but must ensure high visual recognition and conformity to industry practices. Select the corresponding rating for each target device and substitute this rating value into the above color mapping rules to determine its representative color. Mark this color on the location of the corresponding device in the power plant's 3D or 2D spatial layout diagram.

[0101] Step 7: Generate diagnostic results based on the dynamic risk scoring map.

[0102] The system analyzes the color identifiers of each target device in the dynamic risk scoring chart and their corresponding score values, and automatically generates structured diagnostic results by combining the device type, spatial location, and operational context.

[0103] Diagnostic results are output categorized by risk level: Green zone (score ≤ 0.3): judged as healthy and stable, the diagnosis conclusion is that the equipment is operating normally, there are no obvious signs of deterioration, it is recommended to maintain the current operating strategy and carry out routine inspections as planned; Yellow area (0.31 ≤ score ≤ 0.7): judged as initial deterioration, the diagnosis conclusion is that the equipment has shown a detectable performance degradation trend. It is recommended to increase the monitoring frequency, check whether there are any abnormalities in lubrication, alignment or media distribution, and arrange a special inspection during the next planned shutdown window. Red zone (score > 0.7): This indicates accelerated degradation. The diagnosis is that the equipment degradation rate has increased significantly, posing a high risk of failure. It is recommended to arrange a technical review immediately, and if necessary, reduce the load or arrange maintenance in advance to prevent sudden failure.

[0104] like Figure 2 As shown, this embodiment also discloses a condition monitoring and fault diagnosis system for power plant equipment, including the following modules: The data acquisition module is used to continuously collect real-time parameters of the power plant medium; The signal acquisition module is used to acquire the operating status data of multiple target devices affected by the medium of the power plant; The graph construction module is used to divide the power plant medium spatial distribution image into grids and generate a power plant medium load heat map. Each grid of the power plant medium load heat map is labeled with a corresponding load intensity value. The index calculation module is used to generate a dynamic load index based on the target equipment and the power plant medium load heat map; The rate calculation module is used to obtain the short-term degradation rate of each target device based on the dynamic load index and vibration signal. The assessment module generates a dynamic risk score map based on short-term degradation rates and equipment historical maintenance records. The diagnostic module is used to generate diagnostic results based on the dynamic risk scoring map.

[0105] To implement the above embodiments, this application also proposes an electronic device. Please see [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram of the structure of the electronic device provided in an embodiment of this application. For example... Figure 3As shown, the electronic device 500 includes: a processor 501 and a memory 502 communicatively connected to the processor 501; the memory 502 stores computer-executable instructions; the processor 501 executes the computer-executable instructions stored in the memory to implement the method provided in the foregoing embodiments.

[0106] To implement the above embodiments, this application also proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods provided in the foregoing embodiments.

[0107] To implement the above embodiments, this application also proposes a computer program product, including a computer program that, when executed by a processor, implements the methods provided in the foregoing embodiments.

[0108] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for condition monitoring and fault diagnosis of power plant equipment, characterized in that, Includes the following steps: Step 1: Continuously acquire real-time parameters of the power plant medium, including: spatial distribution image of the power plant medium; Step 2: Obtain the operating status data of multiple target devices affected by the power station medium: vibration signals; Step 3: Based on the spatial distribution image of the power plant medium, perform grid division to generate a heat map of the power plant medium load. Each grid of the heat map of the power plant medium load is labeled with a corresponding load intensity value. Step 4: Based on the target equipment and the power plant medium load heat map, generate a dynamic load index; Step 5: Based on the dynamic load index and vibration signal, obtain the short-term degradation rate of each target device; Step 6: Generate a dynamic risk scoring chart based on the short-term degradation rate and the equipment's historical maintenance records; Step 7: Generate diagnostic results based on the dynamic risk scoring map.

2. The method for condition monitoring and fault diagnosis of power plant equipment according to claim 1, characterized in that, The power station medium is a material flow that moves continuously in space and exerts forces on external equipment.

3. The method for condition monitoring and fault diagnosis of power plant equipment according to claim 2, characterized in that, The real-time parameters also include: power plant medium density and power plant medium thickness.

4. The method for condition monitoring and fault diagnosis of power plant equipment according to claim 3, characterized in that, Based on the spatial distribution image of the power plant medium, a grid is divided to generate a thermal map of the power plant medium load, including the following steps: Step 31: Perform grayscale processing on the spatial distribution image of the power station medium to obtain the grayscale value of each pixel; Step 32: Generate a thickness distribution map based on the grayscale value and the thickness of the power station medium; Step 33: Threshold segmentation is performed on the thickness distribution map to generate a binary mask map, wherein pixels with a mask value of 1 constitute the effective carrying area; Step 34: Multiply the binary mask image and the thickness distribution image pixel by pixel to obtain a corrected thickness distribution image composed of the effective bearing area; Step 35: Obtain the actual physical area corresponding to a single pixel; Step 36: Count the number of pixels with a mask value of 1 for each grid; obtain the effective area of ​​the grid based on the number of pixels and the actual physical area; Step 37: Based on the pre-constructed power plant medium motion velocity field, and combined with the power plant medium density, the power plant medium thickness corresponding to each pixel point in the corrected thickness distribution map, and the effective area, obtain the mass flow rate passing through each grid per unit time and mark it as the load intensity value to generate a power plant medium load heat map.

5. A method for condition monitoring and fault diagnosis of power plant equipment according to claim 2, characterized in that, Based on the target equipment and the power plant medium load heat map, a dynamic load index is generated, including the following steps: Step 41: Obtain the equivalent load borne by each target device at the corresponding moment of the current frame; Step 42: Arrange the equivalent loads of consecutive frames in chronological order to form an equivalent load time series; Step 43: Construct a sliding window, and within the sliding window, determine the root mean square value and extreme value sequence of the equivalent load based on the equivalent load time series; Step 44: Determine the absolute difference based on each pair of adjacent extreme values ​​in the extreme value sequence; Step 45: Take the average value of all adjacent extreme value differences as the local peak-to-peak value of the sliding window; Step 46: Construct a composite volatility factor based on the equivalent load time series; Step 47: Generate a dynamic load index based on the composite fluctuation factor, local peak-to-peak value, and root mean square value.

6. The method for condition monitoring and fault diagnosis of power plant equipment according to claim 5, characterized in that, Based on the dynamic load index and vibration signal, the short-term degradation rate of each target device is obtained, including the following steps: Step 51: Based on the type of the target device, determine the characteristic frequency range triggered by its typical faults; Step 52: Perform bandpass filtering on the vibration signal to obtain the target vibration signal, which is composed of components within the characteristic frequency range; Step 53: Perform envelope analysis based on the target vibration signal to obtain the envelope signal; Step 54: Perform spectrum analysis based on the envelope signal to obtain the total energy within the characteristic frequency range, and label it as the characteristic frequency band energy; Step 55: Based on the characteristic frequency band energy and dynamic load index, obtain the coupling response strength at each moment and construct a coupling response strength sequence; Step 56: Perform linear least squares fitting based on the coupling response intensity sequence to obtain the trend slope that changes over time, and use the absolute value of the trend slope as the short-term degradation rate.

7. A method for condition monitoring and fault diagnosis of power plant equipment according to claim 6, characterized in that, Based on short-term degradation rates and historical equipment maintenance records, a dynamic risk scoring chart is generated, including the following steps: Step 61: If the short-term degradation rate is less than the first target value, the target device is determined to be in a healthy and stable stage. The first indicator is obtained based on the device's historical records and marked as the first score. Step 62: If the short-term degradation rate is greater than or equal to the first target value and less than the second target value, the target device is determined to be in the initial degradation stage. The second indicator is obtained based on the device's historical records, and the second score is obtained based on the second indicator and the short-term degradation rate. Step 63: If the short-term degradation rate is greater than or equal to the second target value, the target device is determined to be in the accelerated degradation stage. The third and fourth indicators are obtained based on the device's historical records. The third score is obtained based on the third and fourth indicators and the short-term degradation rate. Step 64: Map the first, second, and third scores of each target device to a preset color range and overlay them onto the power plant equipment spatial layout map to generate a dynamic risk scoring map.

Images