Hydroelectric generating set vibration area fine division method, system, equipment and medium

Abstract

The invention relates to the technical field of digital data processing and analysis, in particular to a hydroelectric generating set vibration area fine division method, system, equipment and medium, based on multi-source vibration and pressure pulsation monitoring data, working condition area division is achieved by screening key characteristic quantities representing set vibration characteristics and utilizing a three-dimensional Gaussian mixture model, and the working condition area of a hydroelectric generating set is divided. Multi-scale data cleaning is carried out by combining a density-based noisy point space clustering algorithm, abnormal data is synchronously eliminated, and data set integration is optimized; establishing a Gaussian process regression model through double optimization, fusing cleaned multi-dimensional related data to realize dynamic refined fitting of a vibration area, and combining three-dimensional data visualization to accurately divide the boundary of the vibration area; through dynamic analysis of multi-source data relevance, automatic starting and stopping of the hydroelectric generating set and accurate load regulation and control are supported to achieve rapid vibration interval crossing, and the safety stability and the operation efficiency of the set are improved in combination with an anti-vibration economic dispatching strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of digital data processing and analysis, and particularly to a method, system, device and medium for refined division of vibration zones of hydro-generator units. Background Art

[0002] At present, although stability tests have been carried out for the planned head range during the operation of hydro-generator units, and the operation stability of the unit output at specific heads has been verified, there are still deficiencies in the refined analysis of head and load, and it is difficult to cover the operation requirements of the unit under all working conditions. With the large-scale grid connection of renewable energy, the randomness and high-frequency switching characteristics of the load of hydro-generator units are significantly enhanced, and they face greater challenges especially when participating in deep peak shaving and frequency modulation. The existing stability test results cannot comprehensively reflect the efficiency and stability characteristics of the unit under all working conditions. In addition, there are limited opportunities for real-machine tests for the refined division of vibration zones, and the vibration zone ranges provided by manufacturers are too broad, resulting in prominent problems in actual operation: First, the accuracy of vibration zone division is insufficient, causing the unit to frequently cross the vibration zone during startup, shutdown and load adjustment, aggravating the wear of mechanical components, shortening the service life and causing economic losses; Second, the broad vibration zone division restricts the optimal load dispatching of power plants, resulting in a decrease in unit efficiency, and even being forced to operate within the vibration zone, threatening safety and economy; Third, the frequent participation of giant units in peak shaving and frequency modulation accelerates the condition switching, triggering the dynamic evolution of vibration zones, and the original division is difficult to meet the economic optimization requirements; Fourth, there are characteristic differences among different units due to manufacturing and installation differences, but the existing vibration zone division standards are single, weakening the collaborative dispatching effect of units; Fifth, due to environmental interferences such as hydraulics and electromagnetics, there are many outliers in the multi-source monitoring data of the unit, making it difficult to accurately characterize the true operating state. These problems jointly restrict the realization of the goals of safe, stable, efficient and economic operation of hydro-generator units. Summary of the Invention

[0003] In view of the problems existing in the above-mentioned prior art, the present invention is proposed.

[0004] Therefore, the technical problem to be solved by the present invention is: how to select characteristic quantities representing the vibration of the unit, eliminate outliers in the historical operation data set of the unit, and establish a vibration fitting model of the associated working conditions of the unit based on the normal set of the historical operation data of the unit to achieve refined division of the vibration zones of hydro-generator units.

[0005] To solve the above technical problems, the present invention provides the following technical solutions. A method for refined division of the vibration area of a hydropower unit includes: obtaining a vibration characteristic quantity data set of the hydropower unit, analyzing the correlation, and selecting the vibration characteristic quantities of the unit; using a three-dimensional Gaussian mixture model optimized by the expectation-maximization algorithm to divide the historical operation data set of the unit's head-power-vibration into multiple operating condition areas; using a density-based spatial clustering of applications with noise (DBSCAN) algorithm to clean the operating condition data set, removing abnormal data sets, merging normal data sets, and using the DBSCAN algorithm to divide them into a dense area and a sparse area of the unit's operating conditions; using double optimization of maximum likelihood estimation and cross-validation to establish a Gaussian process regression model for fitting the operating condition-vibration probability in the dense and sparse areas of the unit's operating conditions; according to the established Gaussian process regression model for fitting the operating condition-vibration probability in the dense and sparse areas of the unit's operating conditions, drawing three-dimensional and two-dimensional heat maps of the operating condition-vibration in the dense and sparse areas of the unit's operating conditions based on the test set data, and refinedly dividing the vibration area of the unit according to the two-dimensional heat map of the operating condition-vibration in the dense and sparse areas of the unit's operating conditions.

[0006] As a preferred scheme of the method for refined division of the vibration area of a hydropower unit described in the present invention, wherein: the characteristic quantity data set includes the historical operation condition data set of the hydropower unit and the characteristic quantity data set characterizing the vibration of the unit.

[0007] As a preferred scheme of the method for refined division of the vibration area of a hydropower unit described in the present invention, wherein: the selection of the vibration characteristic quantities of the unit includes selecting the horizontal vibration of the water guide bearing of the hydropower unit, the horizontal vibration of the top cover and the corresponding vertical vibration, and the runner pressure as characteristic quantities; calculating the Pearson correlation coefficients of the unit's power, head, opening, horizontal vibration of the water guide bearing, horizontal vibration of the top cover and the corresponding vertical vibration, and runner pressure; and obtaining the target characteristic quantity as the characterization index of the vibration characteristics of the unit.

[0008] As a preferred scheme of the method for refined division of the vibration area of a hydropower unit described in the present invention, wherein: the division of the historical operation data set of the unit's head-power-vibration into multiple operating condition areas includes inputting the operating condition parameters and vibration characteristic quantities into the three-dimensional Gaussian mixture model, and using the expectation-maximization algorithm for iterative optimization. Iteratively optimize the estimation of the mean vector, covariance matrix, and weight parameters of the Gaussian distribution, fit the three-dimensional space distribution of the data set, and divide the historical operation data set of the unit's head-power-vibration into multiple operating condition areas according to the true distribution of the data set.

[0009] As a preferred solution of a method for refined division of vibration areas of a hydropower unit according to the present invention, the method includes: the density-based spatial clustering algorithm for noise points includes determining the neighborhood radius and minimum sample number of the density-based spatial clustering algorithm for noise points of the cleaned data set in each working condition area, cleaning each working condition area, and marking and removing the abnormal points in each working condition area; merging the normal data sets after cleaning the working condition areas, and setting the appropriate neighborhood radius and minimum sample number of the density-based spatial clustering algorithm according to the operating characteristics of the unit, and dividing the normal data set of the unit operation history into the dense area and the sparse area of the unit operation working conditions.

[0010] As a preferred solution of a method for refined division of vibration areas of a hydropower unit according to the present invention, the method includes: the method of establishing a working condition-vibration probability fitting Gaussian process regression model for the dense and sparse areas of the unit operation working conditions by using maximum likelihood estimation and cross-validation double optimization includes standardizing the data in the dense area and the sparse area of the unit operation working conditions; dividing the training set and the test set of the data in the dense area and the sparse area of the unit operation working conditions; defining a cross-validation function in combination with the mean square error; optimizing the hyperparameters by using maximum likelihood estimation and evaluating the performance in combination with cross-validation; and training the final working condition-vibration probability fitting Gaussian process regression model for the dense area and the sparse area of the unit operation working conditions by using the optimal parameters obtained by double optimization.

[0011] As a preferred solution of a method for refined division of vibration areas of a hydropower unit according to the present invention, the method includes: the method of refined division of the vibration area of the hydropower unit according to the two-dimensional heat map of the working condition-vibration in the dense and sparse areas of the unit operation working conditions includes inputting the test set data in the dense area and the sparse area of the unit operation working conditions into the established working condition-vibration probability fitting Gaussian process regression model, performing cubic interpolation on the vibration output of the test set, drawing the three-dimensional and two-dimensional heat maps of the working condition-vibration in the dense area and the sparse area of the unit operation working conditions according to the interpolation results, obtaining the two-dimensional heat map of the working condition-vibration in the dense area and the sparse area of the unit operation working conditions according to the three-dimensional heat map of the working condition-vibration in the dense area and the sparse area of the unit operation working conditions, and refinedly dividing the vibration area of the hydropower unit according to the two-dimensional heat map of the working condition-vibration in the dense area and the sparse area of the unit operation working conditions.

[0012] Another object of the present invention is to provide a refined division system for the vibration area of a hydropower unit. Through the full-process collaborative architecture integrating the vibration characteristic quantity selection module, abnormal data cleaning module, unit vibration fitting model module and vibration area refined division module, combined with the multi-condition partitioning of the three-dimensional Gaussian mixture model based on the expectation maximization algorithm and the data cleaning technology of density-based spatial clustering of noise, a dynamic screening strategy for the vibration characterization characteristic quantities of the hydropower unit is realized; a Gaussian process regression model driven by double optimization of hyperparameters is adopted to improve the vibration probability prediction accuracy for the dense and sparse condition areas of the unit; through the calibrated cubic interpolation and the visual coupling analysis of three-dimensional and two-dimensional heat maps, a multi-dimensional space mapping mechanism with self-adaptive vibration area thresholds is established, effectively solving the problems of high sensitivity of traditional methods to non-uniform condition data and fuzzy vibration boundary characterization, and providing an intelligent hierarchical control system with strong robustness for the safe operation of hydropower units.

[0013] To solve the above technical problems, the present invention provides the following technical solutions: A refined division system for the vibration area of a hydropower unit, comprising: a vibration characteristic quantity selection module, an abnormal data cleaning module, a unit vibration fitting model module and a vibration area refined division module; The vibration characteristic quantity selection module obtains the vibration characteristic quantity set of the hydropower unit under associated conditions, calculates the Pearson correlation coefficient of each characteristic quantity, and selects the vibration characteristic quantity of the unit; The abnormal data cleaning module uses a three-dimensional Gaussian mixture model optimized by the maximum expectation algorithm to divide the historical operation dataset of the unit's head-power-vibration into multiple condition areas; uses a density-based spatial clustering algorithm for noise to clean the three-dimensional data of head-power-vibration in each condition area, and eliminates the abnormal data sets in each condition area; merges the normal data sets in each condition area, and uses a density-based spatial clustering algorithm for noise to divide the normal data set into a dense area and a sparse area of the unit's operation conditions; The unit vibration fitting model module uses double optimization of maximum likelihood estimation and cross-validation to establish a condition-vibration probability fitting Gaussian process regression model for the dense and sparse areas of the unit's operation conditions; The vibration area refined division module, according to the established condition-vibration probability fitting Gaussian process regression model for the dense and sparse areas of the unit's operation conditions, draws three-dimensional and two-dimensional heat maps of the conditions-vibration in the dense and sparse areas of the unit's operation conditions based on the test set data, and refinedly divides the vibration area of the unit according to the two-dimensional heat map of the conditions-vibration in the dense and sparse areas of the unit's operation conditions.

[0014] A computer device includes a memory and a processor. The memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the above-mentioned method for refined division of the vibration area of a hydropower unit are implemented.

[0015] A computer-readable storage medium, on which a computer program is stored, characterized in that when the computer program is executed by a processor, the steps of a method for refined division of vibration regions of a hydropower unit as described above are implemented.

[0016] Advantages of the present invention: By proposing a method and system for refined division of vibration regions of a hydropower unit, the present invention aims to regularly and dynamically explore the correlation of multi-source vibration and pressure pulsation monitoring data of a hydropower unit, and at the same time perform multi-scale cleaning on the multi-source vibration and pressure pulsation monitoring data. Based on the cleaned multi-source correlated vibration and pressure pulsation data, dynamic refined fitting division of the vibration regions of the hydropower unit under associated operating conditions is realized, guiding the automatic start-up and shutdown of the hydropower unit and the accurate and rapid crossing of the vibration region during load increase and decrease, as well as vibration avoidance and economic optimal dispatching operation, improving the safety and stability and efficiency of the unit, and providing technical support for the efficient, safe and stable operation of giant hydropower units. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of a method for refined division of vibration regions of a hydropower unit provided by an embodiment of the present invention.

[0019] Figure 2 It is a Pearson correlation coefficient diagram of each characteristic quantity of a method for refined division of vibration regions of a hydropower unit provided by an embodiment of the present invention.

[0020] Figure 3 It is a three-dimensional spatial distribution diagram of a historical operation data set of a method for refined division of vibration regions of a hydropower unit provided by an embodiment of the present invention.

[0021] Figure 4 It is a multi-condition zone diagram of a method for refined division of vibration regions of a hydropower unit provided by an embodiment of the present invention.

[0022] Figure 5 It is an abnormal point distribution diagram of a method for refined division of vibration regions of a hydropower unit provided by an embodiment of the present invention.

[0023] Figure 6 It is a distribution diagram of a dense area and a sparse area of the unit operation conditions of a method for refined division of vibration regions of a hydropower unit provided by an embodiment of the present invention.

[0024] Figure 7The three-dimensional thermal map of vibration under the operating conditions in the dense area of the operating conditions of a hydropower unit, which is provided by an embodiment of the present invention for a method for refined division of vibration areas of a hydropower unit.

[0025] Figure 8 The two-dimensional thermal map of vibration under the operating conditions in the dense area of the operating conditions of a hydropower unit, which is provided by an embodiment of the present invention for a method for refined division of vibration areas of a hydropower unit.

[0026] Figure 9 The three-dimensional thermal map of vibration under the operating conditions in the sparse area of the operating conditions of a hydropower unit, which is provided by an embodiment of the present invention for a method for refined division of vibration areas of a hydropower unit.

[0027] Figure 10 The two-dimensional thermal map of vibration under the operating conditions in the sparse area of the operating conditions of a hydropower unit, which is provided by an embodiment of the present invention for a method for refined division of vibration areas of a hydropower unit. Detailed implementation manners

[0028] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the detailed implementation manners of the present invention with reference to the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0029] Example 1, referring to Figures 1 - 10 , which is an embodiment of the present invention. This embodiment provides a method for refined division of vibration areas of a hydropower unit, including: S1: Obtain the vibration characteristic quantity data set of the hydropower unit, analyze the correlation, and select the vibration characteristic quantity of the unit.

[0030] It should be noted that, as shown in step S1 in Figure 1 , obtain the historical operating condition data set of the hydropower unit and the characteristic quantity data set representing the vibration of the unit from the computer monitoring system of the hydropower station, analyze the correlation of each characteristic quantity in the historical operating data set of the unit, and select the characteristic quantity that can effectively represent the vibration characteristics of the unit.

[0031] Furthermore, select the horizontal vibration of the water guide bearing of the hydropower unit in the direction, the horizontal vibration of the top cover in the direction and the corresponding vertical Z-direction vibration in its direction, and the runner pressure as the characteristic quantity; calculate the unit power, water head, opening, horizontal vibration of the water guide bearing in the direction, the horizontal vibration of the top cover in the direction and the corresponding vertical direction in its Pearson correlation coefficient of directional vibration and runner pressure; select the characteristic quantity with the smallest sum of absolute values of the differences between this index and other indexes as the characteristic quantity representing the vibration characteristics of the unit.

[0032] Specifically, select the horizontal directional vibration of the water guide bearing of a certain hydropower unit, the horizontal direction of the top cover and its corresponding vertical directional vibration ; obtain the unit power , water head , opening ; calculate the Pearson correlation coefficients of each characteristic quantity from the historical operation data of the horizontal directional vibration of the water guide bearing, the horizontal direction of the top cover and its Figure 2 corresponding vertical directional vibration, and runner pressure, as shown in ; select the characteristic quantity with the smallest sum of absolute values of the differences between each index and other indexes as the characteristic quantity representing the vibration characteristics of the unit. It can be calculated that the characteristic quantity is: the horizontal Figure 3 directional vibration of the top cover.

[0033] S2: Use a three-dimensional Gaussian mixture model optimized by the expectation-maximization algorithm to divide the historical operation data set of unit water head-power-vibration into multiple operating condition zones.

[0034] It should be noted that, as shown in step S2 in Figure 1 , use a three-dimensional Gaussian mixture model optimized by the expectation-maximization algorithm to divide the historical operation data set of unit water head-power-vibration into multiple operating condition zones.

[0035] Furthermore, input the operating condition parameters power, water head, and vibration characteristic quantity into the three-dimensional Gaussian mixture model, and at the same time use the expectation-maximization algorithm to iteratively optimize to estimate the mean vector, covariance matrix, and weight parameters of each Gaussian distribution, fit the true three-dimensional space distribution of the data set, and then divide the historical operation data set of unit water head-power-vibration into multiple operating condition zones according to the true distribution of the data set.

[0036] Specifically, input the operating condition parameters power, water head, and vibration characteristic quantity into the three-dimensional Gaussian mixture model; in three-dimensional space, the probability density function of a single Gaussian distribution is expressed as: Among them, is the probability density function of a single Gaussian distribution, is a data point in three-dimensional space, is the mean vector of the Gaussian distribution and is also a three-dimensional vector, is the covariance matrix of the Gaussian distribution, which is a matrix, is the determinant of the covariance matrix, is the square of the Mahalanobis distance, which is used to measure the distance between the data point and the mean vector ; represents the exponential function with the natural constant as the base, is the inverse matrix of the covariance matrix of the Gaussian distribution; The Gaussian mixture model assumes that the data is generated by mixing multiple Gaussian distributions, and its probability density function is expressed as: , Among them, is the number of Gaussian distributions, also known as the number of mixture components, is the mixing weight of the th Gaussian component, satisfying and , is the probability density function of the th Gaussian distribution, with parameters being the mean vector and the covariance matrix , , and respectively represent the sets of mean vectors, covariance matrices, and mixing weights of all Gaussian components; The expectation-maximization algorithm is used to iteratively optimize and estimate the mean vector, covariance matrix, and weight parameters of each Gaussian distribution to fit the true distribution of the three-dimensional space of the dataset; in the algorithm, the goal is to maximize the log-likelihood function of the observed data is expressed as: , Among them, represents the observed data, represents the latent variable, represents the model parameters, represents a whole and is expressed as the log-likelihood function; , Among them, represents the parameter estimate value at the th iteration, represents the conditional expectation of the latent variable given the current parameter estimates and the observed data, is expressed as maximizing as a whole function function, represents the number of iterations; At step, update the parameter estimates by maximizing function , which is expressed as: , wherein, represents a mathematical function for returning the complex argument, represents the maximum value of the model parameters; According to the true distribution of the dataset, the historical operation dataset of unit head-power-vibration is divided into multiple operating condition areas as Figure 4 shown.

[0037] S3: Use the density-based spatial clustering of applications with noise (DBSCAN) algorithm to clean the operating condition dataset, remove the abnormal dataset, merge the normal datasets, and use the DBSCAN algorithm to divide them into the dense area and the sparse area of unit operating conditions.

[0038] It should be noted that, as shown in step S3 of Figure 1 , use the DBSCAN algorithm to clean the three-dimensional data of head-power-vibration in each operating condition area, and remove the abnormal dataset in each operating condition area; merge the normal datasets in each operating condition area, and use the algorithm to divide them into the dense area and the sparse area of unit operating conditions.

[0039] Furthermore, determine the neighborhood radius and the minimum number of samples of the cleaning dataset algorithm for each operating condition area, clean each operating condition area, and mark and remove the abnormal points in each operating condition area; merge the normal datasets after cleaning in each operating condition area, and according to the operating characteristics of the unit, set the appropriate neighborhood radius and the minimum number of samples of the algorithm, and divide the historical normal dataset of unit operation into the dense area and the sparse area of unit operating conditions.

[0040] Specifically, determine the neighborhood radius and the minimum number of samples of the cleaning dataset algorithm for each operating condition area, clean each operating condition area one by one, and mark and remove the abnormal points in each operating condition area, as shown in Figure 5 , wherein, the algorithm includes: initializing the clustering clusters, determining the clustering density threshold and the minimum number of points ; read any point in the dataset, and judge whether this point is a core point according to and , if it is not a core point but is within the neighborhood of the core point, then record it as a border point, otherwise record it as a noise point; for the core point, determine the set of all points that are density-reachable as a clustering cluster ; Read other unvisited core points in the dataset, determine the set of points reachable by their density, and obtain new clustering clusters; repeat the above steps until all points in the dataset are judged; the clustering density thresholds and minimum number of points for each working condition area are shown in Table 1; Table 1 Clustering density thresholds and minimum number of points for each working condition area , Merge the cleaned normal datasets for each working condition area, and according to the operating characteristics of the unit, set the appropriate neighborhood radius and minimum number of samples for the algorithm, as shown in Table 2, and divide the historical normal dataset of unit operation into dense and sparse areas of unit operation conditions, as Figure 6 shown; Table 2 Clustering density thresholds and minimum number of points for dense and sparse areas of unit operation conditions , S4: Adopt double optimization of maximum likelihood estimation and cross-validation to establish a Gaussian process regression model for fitting the probability of working condition-vibration in the dense and sparse areas of unit operation conditions.

[0041] It should be noted that, as shown in step S4 of Figure 1 , adopt double optimization of maximum likelihood estimation and cross-validation to establish a Gaussian process regression model for fitting the probability of working condition-vibration in the dense and sparse areas of unit operation conditions.

[0042] Furthermore, standardize the data in the dense and sparse areas of unit operation conditions; divide the training set and test set of the data in the dense and sparse areas of unit operation conditions; define a cross-validation function in combination with the mean square error; optimize the hyperparameters using maximum likelihood estimation and evaluate the performance in combination with cross-validation; use the optimal parameters obtained by double optimization to train the final Gaussian process regression model for fitting the probability of working condition-vibration in the dense and sparse areas of unit operation conditions.

[0043] Specifically, Gaussian process regression is a non-parametric Bayesian regression method, assuming that the observed data is generated by a Gaussian process, that is, the output function follows a Gaussian process with a mean of and a covariance of ; the model is expressed as: , where is the noise term, assumed to be Gaussian noise; Training stage: Given the training dataset , where is the input variable, is the corresponding output variable; Calculate the covariance matrix of the training dataset, where the elements of are determined by the covariance function Calculated; Prediction stage: For a new input point , predict its corresponding output value ; According to the properties of the Gaussian process, the prediction distribution is a Gaussian distribution, and its mean and variance are expressed as: , where is the covariance matrix of the new input point and the training data set, is the covariance matrix of the training data set; The specific optimization steps include: standardizing the data in the dense and sparse areas of the unit operating conditions; randomly sampling and dividing the data in the dense and sparse areas of the unit operating conditions into a training set and a test set according to a ratio of 1:4; selecting the radial basis kernel function as the Gaussian process regression kernel function, randomly dividing the divided training set into 4 parts, training the Gaussian process regression model, and calculating the mean square error of the output of each training set. Automatically optimize the hyperparameters of the Gaussian process regression model using grid search, and cross-validate and evaluate the generalization performance of the model; combine the hyperparameters of the Gaussian process regression model optimized by automatic search and use maximum likelihood estimation to optimize the hyperparameters, and train the final Gaussian process regression model for fitting the probability of the unit operating conditions in the dense and sparse areas and the vibration probability.

[0044] S5: According to the established Gaussian process regression model for fitting the probability of the unit operating conditions in the dense and sparse areas and the vibration probability, draw the three-dimensional and two-dimensional heat maps of the unit operating conditions in the dense and sparse areas and the vibration probability based on the test set data, and refine the division of the unit vibration area according to the two-dimensional heat map of the unit operating conditions in the dense and sparse areas and the vibration probability.

[0045] It should be noted that, as shown in step S5 in Figure 1 , according to the established Gaussian process regression model for fitting the probability of the unit operating conditions in the dense and sparse areas and the vibration probability, draw the three-dimensional and two-dimensional heat maps of the unit operating conditions in the dense and sparse areas and the vibration probability based on the test set data, and refine the division of the unit vibration area according to the two-dimensional heat map of the unit operating conditions in the dense and sparse areas and the vibration probability.

[0046] Furthermore, input the test set data in the dense and sparse areas of the unit operating conditions into the established Gaussian process regression model for fitting the probability of the unit operating conditions and the vibration probability, perform cubic interpolation on the vibration output of the test set, draw the three-dimensional and two-dimensional heat maps of the unit operating conditions in the dense and sparse areas and the vibration probability according to the interpolation results, obtain the two-dimensional heat map of the unit operating conditions in the dense and sparse areas and the vibration probability based on the three-dimensional heat map of the unit operating conditions in the dense and sparse areas and the vibration probability, and finally refine the division of the unit vibration area according to the two-dimensional heat map of the unit operating conditions in the dense and sparse areas and the vibration probability.

[0047] Specifically, the test set data of the unit operating condition intensive area is input into the established operating condition intensive area working condition-vibration probability fitting Gaussian process regression model, and the vibration output of the test set is interpolated three times. According to the interpolation results, the three-dimensional and two-dimensional thermal diagrams of the unit operating condition intensive area working condition-vibration are drawn. The results are as follows: Figure 7 and Figure 8 As shown; At the same time, the test set data of the sparse area of the unit operating condition is input into the established sparse area of the operating condition-vibration probability fitting Gaussian process regression model, and the vibration output of the test set is interpolated three times. According to the interpolation results, the three-dimensional and two-dimensional thermal diagrams of the sparse area of the unit operating condition-vibration are drawn. The results are as follows: Figure 9 and Figure 10 As shown; According to the unit operating condition intensive area and sparse area condition-vibration two-dimensional thermal map, the unit operating condition-vibration degree partition map is obtained, such as Figure 8 and Figure 10 As shown in the figure, the refined division result of the vibration zone of the hydropower unit can be obtained.

[0048] The above is a schematic scheme of a method for fine division of vibration zones of a hydropower unit in this embodiment. It should be noted that the technical scheme of the system for fine division of vibration zones of a hydropower unit and the technical scheme of the above-mentioned method for fine division of vibration zones of a hydropower unit belong to the same concept. For details not described in detail in the technical scheme of the system for fine division of vibration zones of a hydropower unit in this embodiment, please refer to the description of the technical scheme of the above-mentioned method for fine division of vibration zones of a hydropower unit.

[0049] Embodiment 2 is an embodiment of the present invention, which provides a system for fine division of vibration zones of hydropower units, characterized by comprising: a vibration feature quantity selection module, an abnormal data cleaning module, a unit vibration fitting model module and a vibration zone fine division module; The vibration characteristic quantity selection module obtains the vibration characteristic quantity set of the hydropower unit under the associated working conditions, calculates the Pearson correlation coefficient of each characteristic quantity, and selects the vibration characteristic quantity of the unit; The abnormal data cleaning module uses a three-dimensional Gaussian mixture model optimized by the maximum expectation algorithm to divide the unit head-power-vibration historical operation data set into multiple operating conditions; uses a density-based noise point spatial clustering algorithm to clean the head-power-vibration three-dimensional data of each operating condition area, and removes the abnormal data set of each operating condition area; merges the normal data set of each operating condition area, and uses a density-based noise point spatial clustering algorithm to divide the normal data set into a unit operating condition intensive area and a sparse area; The unit vibration fitting model module adopts double optimization of maximum likelihood estimation and cross-validation to establish a Gaussian process regression model for fitting the probability of operating conditions-vibration in the dense and sparse areas of the unit operating conditions. The refined vibration area division module, based on the established Gaussian process regression model for fitting the probability of operating conditions-vibration in the dense and sparse areas of the unit operating conditions, draws three-dimensional and two-dimensional heat maps of operating conditions-vibration in the dense and sparse areas of the unit operating conditions based on the test set data, and refinedly divides the unit vibration area according to the two-dimensional heat map of operating conditions-vibration in the dense and sparse areas of the unit operating conditions.

[0050] This embodiment also provides a computing device applicable to the case of a method for refined division of the vibration area of a hydropower unit, including: A memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement a method for refined division of the vibration area of a hydropower unit as proposed in the above embodiment.

[0051] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements a method for refined division of the vibration area of a hydropower unit as proposed in the above embodiment.

[0052] The storage medium proposed in this embodiment and the method for refined division of the vibration area of a hydropower unit proposed in the above embodiment belong to the same inventive concept. Technical details not described in detail in this embodiment can be referred to in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0053] Embodiment 3 is the third embodiment of the present invention. What is different from the previous two embodiments is: If the said function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0054] Logic and / or steps described otherwise herein, for example, can be considered a defined sequence of executable instructions for implementing logical functions and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device. As used in this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0055] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0056] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A refined division method for the vibration area of a hydropower unit, characterized in that: Including: Obtain the vibration characteristic quantity dataset of the hydropower unit, analyze the correlation, and select the unit vibration characteristic quantities; Use a three-dimensional Gaussian mixture model optimized by the expectation-maximization algorithm to divide the historical operation dataset of the unit's water head-power-vibration into multiple operating condition zones; Use the density-based spatial clustering of applications with noise (DBSCAN) algorithm to clean the operating condition dataset, remove the abnormal dataset, merge the normal datasets, and use the DBSCAN algorithm to divide them into the dense and sparse zones of the unit's operating conditions; Adopt double optimization of maximum likelihood estimation and cross-validation to establish the operating condition-vibration probability fitting Gaussian process regression model for the dense and sparse zones of the unit's operating conditions; According to the established operating condition-vibration probability fitting Gaussian process regression model for the dense and sparse zones of the unit's operating conditions, draw the three-dimensional and two-dimensional heat maps of the operating condition-vibration for the dense and sparse zones of the unit's operating conditions based on the test set data, and refine the division of the unit's vibration zones according to the two-dimensional heat map of the operating condition-vibration for the dense and sparse zones of the unit's operating conditions.

2. The refined division method for the vibration area of a hydropower unit according to claim 1, wherein: The characteristic quantity dataset includes the historical operating condition dataset of the hydropower unit and the characteristic quantity dataset representing the unit's vibration.

3. The refined division method for vibration zones of a hydropower unit according to claim 2, characterized in that: The selection of the unit vibration characteristic quantities includes selecting the horizontal vibration of the water guide bearing of the hydropower unit, the horizontal vibration of the top cover and the corresponding vertical vibration, and the runner pressure as the characteristic quantities; calculating the Pearson correlation coefficients of the unit's power, water head, opening, horizontal vibration of the water guide bearing, horizontal vibration of the top cover and the corresponding vertical vibration, and runner pressure; and obtaining the target characteristic quantities as the characterization indexes of the unit's vibration characteristics.

4. The refined division method for vibration areas of a hydropower unit according to claim 3, characterized in that: The division of the historical operation dataset of the unit's water head-power-vibration into multiple operating condition zones includes inputting the operating condition parameters and vibration characteristic quantities into the three-dimensional Gaussian mixture model, using the expectation-maximization algorithm, iteratively optimizing and estimating the mean vector, covariance matrix, and weight parameters of the Gaussian distribution, fitting the three-dimensional space distribution of the dataset, and dividing the historical operation dataset of the unit's water head-power-vibration into multiple operating condition zones according to the true distribution of the dataset.

5. The refined division method for vibration areas of a hydropower unit according to claim 4, characterized in that: The use of the DBSCAN algorithm includes determining the neighborhood radius and minimum sample number of the DBSCAN algorithm for the cleaned dataset of each operating condition zone, cleaning each operating condition zone, and marking and removing the abnormal points in each operating condition zone; merging the normal datasets after cleaning the operating condition zones, and setting the appropriate neighborhood radius and minimum sample number of the DBSCAN algorithm according to the operating characteristics of the unit, and dividing the historical normal dataset of the unit's operation into the dense and sparse zones of the unit's operating conditions.

6. The refined division method of the vibration area of a hydropower unit according to claim 5, characterized in that: The adoption of double optimization of maximum likelihood estimation and cross-validation to establish the operating condition-vibration probability fitting Gaussian process regression model for the dense and sparse zones of the unit's operating conditions includes standardizing the data of the dense and sparse zones of the unit's operating conditions; dividing the training set and test set of the data of the dense and sparse zones of the unit's operating conditions; combining the mean square error to define the cross-validation function; using maximum likelihood estimation to optimize the hyperparameters and combining cross-validation to evaluate the performance; and using the optimal parameters obtained by double optimization to train the final operating condition-vibration probability fitting Gaussian process regression model for the dense and sparse zones of the unit's operating conditions.

7. The refined division method for the vibration area of a hydropower unit according to claim 6, wherein: The refined division of the vibration area of the unit according to the dense and sparse area condition-vibration two-dimensional heat maps of the unit operation conditions includes inputting the test set data of the dense area and sparse area of the unit operation conditions into the established condition-vibration probability fitting Gaussian process regression model, performing cubic interpolation on the vibration output of the test set, drawing the three-dimensional and two-dimensional heat maps of the condition-vibration of the dense area and sparse area of the unit operation conditions according to the interpolation results, obtaining the two-dimensional heat maps of the condition-vibration of the dense area and sparse area of the unit operation conditions from the three-dimensional heat maps of the condition-vibration of the dense area and sparse area of the unit operation conditions, and refining the division of the vibration area of the unit according to the two-dimensional heat maps of the condition-vibration of the dense area and sparse area of the unit operation conditions.

8. A refined division system for the vibration zones of a hydropower unit, which applies the refined division method for the vibration zones of a hydropower unit as described in any one of claims 1 to 7, characterized in that: It includes: a vibration characteristic quantity selection module, an abnormal data cleaning module, a unit vibration fitting model module, and a vibration area refinement division module; The vibration characteristic quantity selection module obtains the vibration characteristic quantity set of the hydropower unit related to the working conditions, calculates the Pearson correlation coefficient of each characteristic quantity, and selects the vibration characteristic quantity of the unit; The abnormal data cleaning module uses a three-dimensional Gaussian mixture model optimized by the expectation-maximization algorithm to divide the historical operation data set of the unit head-power-vibration into multiple working condition areas; uses a density-based spatial clustering algorithm with noise to clean the three-dimensional data of the head-power-vibration in each working condition area, and eliminates the abnormal data sets in each working condition area; merges the normal data sets in each working condition area, and uses a density-based spatial clustering algorithm with noise to divide the normal data set into the dense area and sparse area of the unit operation conditions; The unit vibration fitting model module uses double optimization of maximum likelihood estimation and cross-validation to establish a condition-vibration probability fitting Gaussian process regression model for the dense and sparse areas of the unit operation conditions; The vibration area refinement division module, according to the established condition-vibration probability fitting Gaussian process regression model for the dense and sparse areas of the unit operation conditions, draws the three-dimensional and two-dimensional heat maps of the condition-vibration of the dense and sparse areas of the unit operation conditions based on the test set data, and refines the division of the vibration area of the unit according to the two-dimensional heat maps of the condition-vibration of the dense and sparse areas of the unit operation conditions.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it realizes the steps of a method for refined division of the vibration area of a hydropower unit described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it realizes the steps of a method for refined division of the vibration area of a hydropower unit described in any one of claims 1 to 7.

Images