Sediment mineral type similarity analysis method

Abstract

The invention discloses a sediment mineral type similarity analysis method which comprises the following steps: collecting mineral samples along different positions of a river, and preprocessing the samples, including cleaning, drying, crushing and the like; analyzing key features from a mineral sample by using XRD (X-Ray Diffraction), and processing data to calculate feature parameters; determining an optimal clustering number K by adopting an elbow method; the sample features are grouped through a K-means clustering algorithm; the clustering quality is evaluated by adopting a contour coefficient, the similarity of minerals in the river is explored, and the correlation between different features is analyzed; the invention provides a new method for solving the problem of wear of flow passage components of a hydraulic turbine set caused by silt by analyzing the characteristics of silt minerals at different positions of a river to reflect the similarity of the types of the overflow silt minerals of the hydraulic turbine set.

Description

Technical Field

[0001] The invention belongs to the technical field of hydropower generation, and in particular relates to a method for similarity analysis of sediment mineral types. Background Art

[0002] As the world pays increasing attention to reducing greenhouse gas emissions and addressing climate change, hydropower, as a clean energy source, is of great significance for achieving the transformation of energy structure and promoting sustainable development. It helps to reduce dependence on fossil fuels, reduce air pollution and greenhouse gas emissions, and provide stable and economical energy solutions. With the continuous construction of large hydropower stations, large hydropower units are continuously manufactured and put into operation. Once these large units fail, it will not only seriously affect the entire power grid, but also cause huge economic losses. Therefore, it is very important to maintain the normal operation of the turbine units.

[0003] A major problem that affects the stable operation of large units is the wear of turbine flow parts. Specifically, under the influence of high head, high specific speed and low sand content, the parts of the turbine unit, such as the runner chamber, blades, runner, anti-wear ring, etc., will have defects such as pitting, ripples, fish scale pits, pinholes, pitting, etc. after long-term operation, which will eventually threaten the safety of the entire unit. After the turbine is worn, it is easy to lose balance of the runner, metal loss, and increased vibration, resulting in increased maintenance and operation costs of the unit; for hydropower stations on rivers with high sand content, sediment wear is often the decisive factor affecting the degree of wear and maintenance of turbine units.

[0004] At present, the research on the types of minerals flowing through turbine units mainly uses X-ray or SEM / EDS technology to analyze the mineral morphology and composition; Ma Lin et al. used X-ray diffraction experiments to determine the main mineral components in the flowing sediment when preliminarily monitoring and analyzing the amount of sediment passing through the Three Gorges Hydropower Station; Zhang Yunfeng et al. used D / MAX-3A automatic X-ray diffractometer to study the mineral deposition characteristics of the Qidongzui tidal flat in the Yangtze River estuary; Ni Liang et al. used an X-ray diffractometer for analysis, and by comparing the XRD spectrum with the standard card, determined that the main components of the samples included quartz, potassium feldspar, albite, common pyroxene, muscovite, kaolinite and amphibole; but the above methods only measure the types of sediment minerals at the sample points, and cannot reflect the relationship between the mineral types in the upstream and downstream of the river, as well as the characteristics of mineral samples at different locations in the river channel upstream of the reservoir. Summary of the invention

[0005] The present invention provides a method for analyzing the similarity of sediment mineral types, which solves the problem of wear of flow-through components of a turbine set caused by sediment. The method reflects the similarity of sediment mineral types flowing through the turbine set by analyzing the sediment mineral characteristics at different positions of a river, solves the problem that the traditional sediment mineral detection method is single and does not reflect the relationship between sediment minerals at different positions, verifies the similarity of river sediment mineral types, and explores the characteristics of sediment mineral samples at different positions of a river.

[0006] In order to solve the above problems, the present invention provides a method for analyzing the similarity of sediment mineral species, comprising the following steps:

[0007] Step 1: Collect sediment mineral samples along the same river from the reservoir tail, reservoir area, river bank, riverbed, sedimentation area, and near the turbine unit;

[0008] Step 2: Perform necessary pretreatment on the collected sediment mineral samples, including cleaning, drying and crushing;

[0009] Step 3: Analyze key features from mineral samples using XRD, perform descriptive statistical analysis on the extracted features, and evaluate the central tendency and dispersion of the data;

[0010] Step 4: Use the elbow method to determine the optimal number of clusters K;

[0011] Step 5: Use K-means clustering algorithm to group sample features to identify mineral sample sets with similar features;

[0012] Step 6: Use the silhouette coefficient to evaluate the clustering quality, explore the similarities of minerals in the river, and analyze the correlation between different features.

[0013] Furthermore, collecting sediment mineral samples specifically includes: conducting systematic multi-point sampling in the river, and collecting sediment mineral samples at the reservoir tail, reservoir area, river bank, riverbed, sedimentation area, and near the turbine unit along the same river.

[0014] Furthermore, the pretreatment of sediment mineral samples specifically includes: taking an appropriate amount of sediment sample, adding 5 ml of hydrogen peroxide to remove organic matter after the reaction is complete (no bubbles, no lumps), adding distilled water and stirring to form a 1000 mL suspension; according to the Stokes sedimentation law, extracting the suspension below 2 um, repeating several times until enough particles are obtained; then using a centrifuge to separate the particles, and drying them at 60°C for use.

[0015] Furthermore, XRD is used to analyze key features from mineral samples, and descriptive statistical analysis is performed on the extracted features. The specific steps include: obtaining data such as the type and content of minerals contained in the sample through X-ray diffraction (XRD) full spectrum analysis, preprocessing the data, filling in missing data, eliminating abnormal data, and performing descriptive statistical analysis on the extracted features to obtain characteristic data such as the mineral content, median and standard deviation.

[0016] Furthermore, the specific steps of using the elbow method to determine the optimal clustering number K include: calculating the sum of square errors (SSE) under different clustering numbers by formula (1) to determine the optimal clustering number K.

[0017]

[0018] (1) Where: C i is the i-th cluster; P is C i Sample points in m i C i The centroid of ; SSE is the clustering error of all samples.

[0019] Furthermore, the K-means clustering algorithm is used to divide the sample features into K clusters so that the sum of the distances from each object to the center of its cluster is minimized. The specific steps include:

[0020] S101: Randomly select K data points as initial cluster centers;

[0021] S102: using formula (2) to calculate the Euclidean distance as the distance metric to calculate the distance between each data point and all cluster centers, and classifying it into the cluster with the closest distance;

[0022]

[0023] (2) Where: x is the data object; C i is the i-th cluster center; m is the dimension of the data object; X j , C ij are x and C respectively i The j-th attribute value of ;

[0024] S103: For each cluster, calculate the average of all its data points and use it as the new center;

[0025] S104: Repeat S102 and S103 until the cluster center does not change or the maximum number of iterations is reached;

[0026] Furthermore, the profile function is constructed by calculating the inter-group separation of the mineral data sample points and the intra-group compactness of the data sample points. Finally, the overall average value is calculated to obtain the final index value. The specific steps of using the profile coefficient to evaluate the clustering quality include:

[0027] S201: For n i data set, assuming sample n i is clustered into cluster G, for the silhouette coefficient s i The definition of should comply with the rule of formula (3);

[0028]

[0029] (3) Where: x i For sample n i , and the average distance of other samples in the same cluster G; for cluster J, i.e., the cluster in the sample that is not cluster G, N(i, J) is the sample s i The average distance from all samples in J, then

[0030] S202: The silhouette coefficient s of the data set is calculated using formula (4). k definition;

[0031]

[0032] (4) Where n is the number of samples in the data set; k is the number of clustering times; s k is the mean silhouette coefficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flow chart of a method for analyzing similarity of sediment mineral types provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0035] See also Figure 1 A method for similarity analysis of sediment mineral types according to an embodiment of the present invention is shown, comprising the following steps:

[0036] Step 1: For the same river, select the tail of the reservoir, the reservoir area, the river bank, the riverbed, the sedimentary area, and the vicinity of the turbine unit to collect sediment mineral samples. For different rivers, the processing and collection methods are the same, and samples are taken at the tail of the reservoir and the reservoir area respectively;

[0037] Step 2: Perform necessary pretreatment on the collected sediment mineral samples, including cleaning, drying and crushing;

[0038] Step 3: Analyze key features from mineral samples using XRD, perform descriptive statistical analysis on the extracted features, and evaluate the central tendency and dispersion of the data;

[0039] Step 4: Use the elbow method to determine the optimal number of clusters K;

[0040] Step 5: Use K-means clustering algorithm to group sample features to identify mineral sample sets with similar features;

[0041] Step 6: Use the silhouette coefficient to evaluate the clustering quality, explore the similarities of minerals in the river, and analyze the correlation between different features.

[0042] The collection of mineral samples specifically includes: conducting systematic multi-point sampling in the river, and collecting sediment mineral samples at the reservoir tail, reservoir area, river bank, riverbed, sedimentary area, and near the turbine unit along the same river.

[0043] The pretreatment of sediment mineral samples specifically includes: taking an appropriate amount of sediment sample, adding 5 ml of hydrogen peroxide to remove organic matter after the reaction is complete (no bubbles, no lumps), adding distilled water and stirring to form a 1000 mL suspension; according to the Stokes sedimentation law, extracting the suspension below 2 um, repeating several times until enough particles are obtained; then using a centrifuge to separate the particles, and drying them at 60°C for use.

[0044] XRD is used to analyze key features from mineral samples, and descriptive statistical analysis is performed on the extracted features. The specific steps include: obtaining data such as the type and content of minerals contained in the sample through X-ray diffraction (XRD) full spectrum analysis, preprocessing the data, filling in missing data, eliminating abnormal data, and performing descriptive statistical analysis on the extracted features to obtain characteristic data such as the mineral content, median and standard deviation.

[0045] In this example, the elbow method is used to determine the optimal cluster number K. The specific steps include: calculating the sum of square errors (SSE) under different cluster numbers by formula (1) to determine the optimal cluster number K.

[0046]

[0047] (1) Where: C i is the i-th cluster; P is C i Sample points in m i C i The centroid of ; SSE is the clustering error of all samples.

[0048] The K-means clustering algorithm is used to divide the sample features into K clusters so that the sum of the distances from each object to the center of its cluster is minimized. The specific steps include:

[0049] S101: Randomly select K data points as initial cluster centers;

[0050] S102: using formula (2) to calculate the Euclidean distance as the distance metric to calculate the distance between each data point and all cluster centers, and classifying it into the cluster with the closest distance;

[0051]

[0052] (2) Where: x is the data object; C i is the i-th cluster center; m is the dimension of the data object; X j , C ij are x and C respectively i The j-th attribute value of ;

[0053] S103: For each cluster, calculate the average of all its data points and use it as the new center;

[0054] S104: Repeat S102 and S103 until the cluster center no longer changes or the maximum number of iterations is reached.

[0055] Furthermore, the profile function is constructed by calculating the inter-group separation of the mineral data sample points and the intra-group compactness of the data sample points. Finally, the overall average value is calculated to obtain the final index value. The specific steps of using the profile coefficient to evaluate the clustering quality include:

[0056] S201: For n i data set, assuming sample n i is clustered into cluster G, for the silhouette coefficient s i The definition of should comply with the rule of formula (3);

[0057]

[0058] (3) Where: x i For sample n i , and the average distance of other samples in the same cluster G; for cluster J, i.e., the cluster in the sample that is not cluster G, N(i, J) is the sample s i The average distance from all samples in J, then

[0059] S202: The silhouette coefficient s of the data set is calculated using formula (4). k definition;

[0060]

[0061] (4) Where n is the number of samples in the data set; k is the number of clustering times; s k is the mean silhouette coefficient.

[0062] The present invention aims to solve the wear problem of flow-through parts of a turbine set caused by sediment. The similarity of the types of minerals in the flow-through sediment of a turbine set is reflected by analyzing the mineral characteristics of sediment at different positions of a river. A clustering algorithm is used to analyze the mineral characteristics of river sediment. Samples are taken at different positions along the river, and the samples are pre-processed. Key features are analyzed from the mineral samples using XRD, and the data are processed to calculate feature parameters. The elbow method is used to determine the optimal clustering number K, and the K-means clustering algorithm is used to group the sample features. Finally, the silhouette coefficient is used to evaluate the clustering quality. The similarity between samples is evaluated through the clustering results. If most samples are clustered in fewer clusters and the similarity of the samples in the clusters is high, this indicates that the minerals in the river have a high similarity.

[0063] For the current method of similarity analysis of the types of sediment minerals flowing through turbine units, the present invention solves the problem that the traditional sediment mineral detection method is single and does not reflect the relationship between sediment minerals in different positions. The machine learning method is used to classify the mineral data points according to their similarities to verify the similarity of river sediment mineral types. The algorithm has strong interpretability and can explore the mineral sample characteristics of different positions of river sediment minerals, providing a method for solving the problem of wear of flow components of turbine units caused by sediment.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for similarity analysis of sediment mineral species, characterized in that: The following steps are involved: Step 1: Collect sediment mineral samples along the same river from the reservoir tail, reservoir area, river bank, riverbed, sedimentation area, and near the turbine unit; Step 2: Perform necessary pretreatment on the collected sediment mineral samples, including cleaning, drying and crushing; Step 3: Analyze key features from mineral samples using XRD, perform descriptive statistical analysis on the extracted features, and evaluate the central tendency and dispersion of the data; Step 4: Use the elbow method to determine the optimal number of clusters K; Step 5: Use K-means clustering algorithm to group sample features to identify mineral sample sets with similar features; Step 6: Use the silhouette coefficient to evaluate the clustering quality, explore the similarities of minerals in the river, and analyze the correlation between different features.

2. A sediment mineral type similarity analysis method according to claim 1, characterized in that: The collection of mineral samples specifically includes: conducting systematic multi-point sampling in the river, and collecting sediment mineral samples at the reservoir tail, reservoir area, river bank, riverbed, sedimentary area, and near the turbine unit along the same river.

3. The method for similarity analysis of sediment mineral species according to claim 1, characterized in that: The pretreatment of sediment mineral samples specifically includes: taking an appropriate amount of sediment sample, adding 5 ml of hydrogen peroxide to remove organic matter after the reaction is complete (no bubbles, no lumps), adding distilled water and stirring to form a 1000 mL suspension; according to the Stokes sedimentation law, extracting the suspension below 2 um, repeating several times until enough particles are obtained; then using a centrifuge to separate the particles, and drying them at 60°C for use.

4. The method for similarity analysis of sediment mineral species according to claim 1, characterized in that: XRD is used to analyze key features from mineral samples, and descriptive statistical analysis is performed on the extracted features. The specific steps include: obtaining data such as the type and content of minerals contained in the sample through X-ray diffraction (XRD) full spectrum analysis, preprocessing the data, filling in missing data, eliminating abnormal data, and performing descriptive statistical analysis on the extracted features to obtain characteristic data such as the mineral content, median and standard deviation.

5. The method for similarity analysis of sediment mineral species according to claim 1, characterized in that: The specific steps of using the elbow method to determine the optimal clustering number K include: calculating the sum of squared errors (SSE) under different clustering numbers by formula (1) to determine the optimal clustering number K; (1) Where: C i is the i-th cluster; p is C i Sample points in m i C i The centroid of ; SSE is the clustering error of all samples.

6. The method for similarity analysis of sediment mineral species according to claim 1, characterized in that: The specific steps of grouping sample features using the K-means clustering algorithm include: S101: Randomly select K data points as initial cluster centers; S102: using formula (2) to calculate the Euclidean distance as the distance metric to calculate the distance between each data point and all cluster centers, and classifying it into the cluster with the closest distance; (2) Where: x is the data object; C i is the i-th cluster center; m is the dimension of the data object; X j , C ij are x and C respectively i The j-th attribute value of ; S103: For each cluster, calculate the average of all its data points and use it as the new center; S104: Repeat S102 and S103 until the cluster center no longer changes or the maximum number of iterations is reached.

7. The method for similarity analysis of sediment mineral species according to claim 1, characterized in that: The silhouette coefficient index constructs a silhouette function by calculating the inter-group separation of mineral data sample points and the intra-group compactness of data sample points, and finally calculates the overall average value to obtain the final index value. The specific steps of using the silhouette coefficient to evaluate the clustering quality include: S201: For n i data set, assuming sample n i is clustered into cluster G, for the silhouette coefficient s i The definition should conform to the rule of formula (3); (3) Where: x i For sample n i , and the average distance of other samples in the same cluster G; for cluster J, i.e., the cluster in the sample that is not cluster G, N(i, J) is the sample s i The average distance from all samples in J, then S202: The silhouette coefficient s of the data set is calculated using formula (4). k definition; (4) Where n is the number of samples in the data set; k is the number of clustering times; s k is the mean silhouette coefficient.

Images