A method for analyzing the service status of dams based on strongly similar sets.
The method addresses the limitations of single-point dam monitoring by using strongly similar sets for real-time, comprehensive dam service status analysis, integrating multi-point data through similarity mining and cloud models to enhance monitoring accuracy and safety.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- HOHAI UNIV
- Filing Date
- 2025-08-13
- Publication Date
- 2026-05-19
AI Technical Summary
Existing dam service status analysis methods, particularly those using single-measurement points, fail to adequately utilize similarity information between measurement points, leading to incomplete and potentially misleading assessments due to changes in dam conditions and external factors, resulting in inaccurate monitoring and risk of serious accidents.
A method utilizing strongly similar sets for dam service status analysis, involving data mining, time-series tracking, and hierarchical fusion to integrate multi-point data, ensuring real-time monitoring and comprehensive analysis through similarity mining and cloud model integration.
Ensures real-time, accurate dam service status monitoring by tracking changes in similarity relationships and integrating multi-point data, providing comprehensive analysis from local to whole-dam levels, identifying abnormal parts for intensive monitoring.
Smart Images
Figure 2026082646000001_ABST
Abstract
Description
[Technical Field]
[0001] The present invention relates to the technical field of dam service state analysis (operational behavior or operational characteristics), and more specifically to a method for analyzing the service state of a dam based on a strongly similar set. [Background technology]
[0002] During the operation of a dam, its service status is constantly changing due to the effects of complex and diverse environmental and load factors, as well as the invasive impact of various sudden disasters. Therefore, it is necessary to grasp the dam's service status in real time through various safety monitoring means, promptly detect any abnormal changes, and take corresponding measures to prevent serious accidents or failures at the dam, thereby ensuring the long-term safe operation of the project. Currently, in the practical application of projects, the majority of dam service status analysis methods adopted are single-measurement point methods. However, monitoring information from a single measurement point cannot fully reflect the service status changes of the entire project, and the change disciplines between monitoring effect quantities (monitoring indicator quantities) at different measurement points do not match, making it easy for contradictory analysis results to appear. Therefore, it is necessary to obtain service status analysis results for the entire project by utilizing monitoring information from multiple measurement points from multiple perspectives and integrating them. However, the majority of existing multi-measurement point service status analysis methods only utilize the measurement value information of the measurement point itself, and do not sufficiently utilize the similarity information between measurement values at the measurement points, which can lead to the loss of important information and misjudgment of the state.
[0003] When analyzing the service status of a dam in real time using strongly similar sets (strongly similar clusters), it is necessary to use the mining results of strongly similar sets obtained during previous monitoring periods. However, during the dam service process, the state of change is altered due to constant changes in the dam's own conditions and external working conditions. This causes changes in the similarity relationship between the measured values of the monitoring effect amounts at each measurement point, and the mining results of the original strongly similar set may become invalid at the present time. If service status analysis is performed using mining results of a strongly similar set that have already become invalid, it is likely to lead to misjudgments of the service status. [Overview of the project]
[0004] Objective of the Invention: The objective of the present invention is to provide a method for analyzing the service status of dams based on strongly similar sets, in order to overcome the shortcomings of the prior art. Through mining of strongly similar sets and multi-layer fusion analysis, it is possible to comprehensively reflect changes in the service status of dams and to provide a new analytical means for dam safety monitoring.
[0005] Technical solution: The method for analyzing the operational characteristics of dams based on strongly similar sets according to the present invention includes the following steps. Step S1 involves dividing the dam into multiple sections and acquiring monitoring data from multiple measurement points in each section. Step S2 involves performing strong similarity mining (extraction and discovery processing) on monitoring data from multiple measurement points within the same area to obtain a strong similarity set encompassing all measurement points, employing a tracking analysis method for the composition (structure) of the strong similarity set in relation to changes in the time-series data of the measured values (measurement time series), obtaining the determination result of the composition of the strong similarity set at the current point in time under changes in the time-series data of the measured values, and constructing a hierarchical analysis system (a hierarchical analysis framework or analysis system based on a hierarchical structure) of the dam service state based on multiple strong similarity sets, Step S3 involves using a dam service status analysis method for the range of a single strongly similar set to obtain the analysis results of the service status of dams within the range in which each strongly similar set is located, Step S4 involves calculating the degree of discreteness and quantitative characteristics (quantitative properties) of the analysis results of the service status of each element in the relevant hierarchy (the hierarchy being processed), constructing a decision matrix based on the calculated degree of discreteness and quantitative characteristics of the analysis results of the service status of each element, and determining the weights of the influence that the analysis results of the service status of each strongly similar set of dams have on the upper (higher hierarchy) part. Step S5 includes sequentially hierarchically merging the analysis results of each strongly similar set upwards to obtain the analysis results of the service status of each part in the part layer and the analysis results of the service status of the entire dam.
[0006] Furthermore, the step of employing a strongly similar set composition tracking analysis method for changes in the time-series data of the measured values in step S2 is as follows: S201: Perform normalization on the time series data of the measured values at each measurement point in a strongly similar set. S202: Determine the sliding time interval l according to the actual situation (variability) of the change in measured values at each measurement point in the strongly similar set. S203: Within a sliding time period length l, calculate the time series data of the TWED distance between each measurement point and other measurement points in the strongly similar set for each time interval, and construct the time series data of the total distance in the strongly similar set for each measurement point. S204: Maximum value of the total distance X within a fixed-length time period. mi The sample space will be used to extract data and calculate a discriminant index for the relationship of each measurement point to a strongly similar set. S205: Based on a typical small probability method, a discriminant index X is used to determine the relationship of each measurement point to a strongly similar set under a certain small probability. m Calculating, S206: For the measurement value of the i-th measurement point in the strongly similar set at the judgment time t', the total distance D' of the strongly similar set of that measurement point. i,t,l The discriminant index X is calculated to determine the relationship of the measurement point to the strongly similar set. m By comparing this with D', we obtain the initial determination result regarding the relationship to strongly similar sets. i,t,l ≤X m In this case, at time t', the measurement point is still considered to belong to this strongly similar set, D' i,t,l >X m In this case, at time t', the measurement point is considered not to belong to this strongly similar set. S207: For measurement points that are determined not to belong to this strongly similar set in the initial determination result regarding the attribution relationship, a significant error identification process (gross difference identification process) is performed on the measurement value, significant errors (gross differences) are removed, and then the analysis is performed again to obtain a supplementary determination result regarding the attribution relationship. S208: The measurement points determined to belong to the original strong similarity set are composed as a subset of the strong similarity set. For the measurement points determined not to belong to the original strong similarity set in the supplementary discrimination result regarding the belonging relationship, it is discriminated one by one whether they belong to this subset of the strong similarity set at present (at this point). If it is determined that they belong to this strong similarity subset, at time t', it is regarded that the measurement point still belongs to the original strong similarity set. If it is determined that they do not belong to this strong similarity subset, at time t', it is regarded that the measurement point does not belong to this strong similarity set. S209: Integrate the discrimination results regarding the belonging relationship of all measurement points in the strong similarity set, and obtain the discrimination result of the composition status of the strong similarity set at the current time under the change of the time-series data of the measured values. This includes.
[0007] Furthermore, the hierarchical analysis system of the dam service state divides the entire dam into four hierarchies: the measurement point layer, the same-site strong similarity set layer, the site layer, and the entire dam layer from bottom to top. Each element in the same hierarchy has the same rank, and the elements belonging to two adjacent upper and lower layers have a relationship of whole and part.
[0008] Furthermore, for an element in this hierarchy, the set of l corresponding elements of a certain element c_j in the upper hierarchy is
Number
Number
[0009] Furthermore, the step S3 is S301: Perform normalization processing on the monitoring values of the monitoring effect index quantities at each measurement point of the strong similarity set within the analysis time period from 1 to t'. S302: Inputting normalized multidimensional data into a dimensionality reduction algorithm using local linear embeddings, adjusting the number of neighbors k, and obtaining the initial dimensionality reduction result. S303: Adjust the range of the value based on the range of change of the time series data of the average value measured at each measurement point of a strongly similar set, and obtain time series data Y of the overall effect amount that reflects the change in the service status of the dam in the area where the strongly similar set is located during that time period. S304: Time series data of the total effect size within the time zone 1~(t'-1) is input into the inverse cloud generator, and three numerical properties Ex, En, and He are obtained. Ex, En, and He are the expected value, entropy, and hyperentropy in the cloud model of the time series data of the total effect size of a strongly similar set, respectively. S305: Input the numerical characteristics Ex, En, and He into a positive cloud generator and calculate and obtain a cloud model that reflects the overall effect size distribution. S306: Based on the "3En" criteria of the cloud, the upper limit T_u and lower limit T of the state discrimination index corresponding to time t'. l To set, S307: Total effect size Y of the strongly similar set at time t' t’ and state discrimination index value T u and T l By comparing this with the previous set, we obtain the initial analysis results of the service status of dams within the range where this strongly similar set is located at that point in time. S308: Formula
number
[0010] Furthermore, the degree of discreteness C of the analysis results of the service status of each element in the hierarchy one level below that corresponds to the element in that hierarchy. Di Calculate, Quantitative features C of the element in the next lower level that corresponds to the element in the current level. Qi Calculate, C Di and CQi Based on this, construct the decision matrix D, The following formula is used to determine the weight W of the impact of the analysis results of the service status of the elements at the relevant hierarchical level at time T on the upper levels. S (T) Calculate,
number
number
number
[0011] Furthermore, at time point T, element c in the hierarchy one level higher j Based on the analysis results of the service status of each corresponding element in that hierarchy, and the weight of the influence each element in that hierarchy has on the hierarchy above, it merges with the element c in the hierarchy above. j Analysis results of the service status c j (T) The formula obtained is,
number
[0012] Furthermore, the step of sequentially hierarchically merging the analysis results of each strongly similar set upwards to obtain the analysis results of the service status of each part in the part layer and the analysis results of the service status of the entire dam is as follows: At point t, the set of analysis results for the service status of each element in the next lower level.
number
[0013] Beneficial Effects: Compared to the prior art, the present invention has the following advantages: (1) In order to ensure the real-time effectiveness of the mining results of strongly similar sets, the present invention considers the change in similarity between measured values at measurement points, studies a time deviation editing distance method by sliding calculation for each time interval between a measurement point and other measurement points in the strongly similar set, constructs a total distance that quantitatively shows the relationship between a measurement point and the strongly similar set, and constructs a relationship that quantitatively shows the relationship between a measurement point and the strongly similar set using this method, constructs a discriminant index for the relationship to the strongly similar set based on a typical small probability method, proposes a method for tracking and analyzing the composition status of the strongly similar set under changes in time-series data of measured values, tracks changes in the composition status of the strongly similar set in real time, detects changes in the relationship between a measurement point and the strongly similar set in a timely manner, and ensures the real-time effectiveness of the mining results of the strongly similar set. (2) From the perspective of information fusion, dimensionality reduction processing is performed on multiple measurement point data within a strongly similar set based on a local linear embedding method, a state discrimination index is set in combination with cloud model theory, a method for analyzing the service state of a dam within the range of a single strongly similar set is proposed, multiple measurement point information within the strongly similar set is fused, a comprehensive effect size that reflects the service state of the dam within the local range is obtained, and analysis of the service state of the dam within the range of a single strongly similar set is realized. (3) In order to analyze changes in the service state of each part or the entire dam, the concept of hierarchical analysis is used, a hierarchical analysis system of dam service states based on multiple strongly similar sets is proposed, a judgment matrix fusion method based on importance is constructed, and a sequential hierarchical fusion analysis method of dam service states within the range of multiple strongly similar sets is proposed, enabling analysis of the service state of the dam from local to whole, obtaining analysis results of the service state of each part and the entire dam, identifying abnormal parts and monitoring them intensively.
[0014] This invention provides a new analytical method for dam safety monitoring by comprehensively reflecting changes in the service status of dams through mining of strongly similar sets and multi-layer fusion analysis. [Brief explanation of the drawing]
[0015] [Figure 1] This is a schematic diagram of the structure of the sequential hierarchical fusion analysis method for the service state of dams within a range of multiple strongly similar sets according to the present invention. [Figure 2] This invention relates to a method for tracking and analyzing the composition of strongly similar sets under changes in time-series data of measured values. [Figure 3] This invention provides a hierarchical analysis system for dam service states based on multiple strongly similar sets. [Figure 4] This is a flowchart for analyzing the service status of dams within a single strongly similar set according to the present invention. [Modes for carrying out the invention]
[0016] The technical solutions of the present invention will be described in detail below with reference to the attached drawings, but the scope of protection of the present invention is not limited to the following embodiments.
[0017] Example 1 The sequential hierarchical fusion analysis method for the service status of dams within the range of multiple strongly similar sets shown in Figure 1 includes the following specific implementation steps. S1: Based on the actual conditions of the dam, the dam is divided into E sections, and monitoring data is acquired from multiple measurement points in each section. S2: By performing strongly similar set mining on monitoring data from multiple measurement points within the same area, a strongly similar set encompassing all measurement points is obtained. A tracking analysis method for the composition status of the strongly similar set is then applied to the changes in the time-series data of the measured values. The determination result of the composition status of the strongly similar set at the current point in time under the changes in the time-series data of the measured values is obtained, thereby constructing a hierarchical analysis system for the dam service status. S3: Using a dam service status analysis method for the range of a single strongly similar set, the analysis results s of the service status of dams within the range in which each strongly similar set is located are obtained. S4: Based on the normalized data of the measured values of the monitoring effect size (effect index) at multiple measurement points within each strongly similar set, the weight W of the influence of the analysis results of the service status of the dam in each strongly similar set on the upper part. s To decide. S5: Perform a fusion calculation (a calculation incorporating weights) to obtain the QZ analysis result of the dam's service status at each part. S6: Based on the analysis results of the service status of each strongly similar set of dams within each section, the weight W of the influence of the analysis results of the service status of each section of the dam on the overall service status of the dam. QZ To decide. S7: Perform a fusion calculation to obtain the analysis result Z of the service status of the entire dam.
[0018] In step S2, the time deviation edit distance similarity analysis method is used to analyze monitoring data from multiple measurement points of the dam, a time deviation edit distance similarity matrix is obtained, the time deviation edit distance similarity matrix is converted into an adjacency matrix of multiple measurement point graphs, the measured information of monitoring effect amounts (monitoring indicators) at multiple measurement points and the adjacency matrix are jointly represented as multiple measurement point graphs, and the strong similarity set mining is performed on the multiple measurement point graphs using the greedy module maximization strong similarity set mining method to obtain the results of the subgraph partitioning of the multiple measurement point graphs.
[0019] When the dam change state changes, causing a change in the similarity relationship between the measured values of the monitoring effect quantities at each measurement point, the measured value at a certain measurement point in a strongly similar set may no longer have strong similarity with the measured values of other measurement points in the original strongly similar set. In other words, it no longer belongs to that strongly similar set. At this time, the TWED distance between the time series data of the measured values of other measurement points in the original strongly similar set, calculated by the time deviation editing distance method, increases. Utilizing this property, a total distance (hereinafter abbreviated as the total distance of the strongly similar set) can be constructed between the measured value at the measurement point and the measured values of all other measurement points in the strongly similar set. Based on this, the relationship between the measurement point and the strongly similar set can be quantitatively shown. On this foundation, based on the typical small probability method, a discriminant index for the relationship of belonging to the strongly similar set is constructed and used as the basis for discriminating the relationship between the measurement point and the strongly similar set. The basic flow of the tracking analysis method for the composition status of the strongly similar set under changes in the time series data of the measured values is shown in Figure 2, and the specific implementation steps are as follows. S201: Performs normalization on the time series data of the measured values at each measurement point in a strongly similar set. S202: The sliding time interval length l is determined according to the actual situation of the change in the measured value at each measurement point in the strongly similar set. S203: Within the sliding time period length l, time-series data of the TWED distance between each measurement point and other measurement points in the strongly similar set is calculated for each time interval, and based on this, time-series data of the total distance in the strongly similar set for each measurement point is constructed. S204: Maximum value of the total distance X within a fixed-length time period. mi The data is statistically analyzed (extracted or collected), and in the time series data of the total distance between the measured value of the measurement point and the measured values of other measurement points in the strongly similar set, the maximum value of the total distance within each fixed-length time period is X. mi We extract each measurement point one by one and construct a sample space X by calculating a discriminant index for the relationship of each measurement point to a strongly similar set. S205:X belongs to the small sample size space and represents the statistical (extracted) features of the sample space X.
number
number
number
[0020] In the formula, m is the number of measurement points in this strongly similar set, n is the length of the time series data of the measurement values, t is the current calculation time, l is the sliding time period length, and D i,j,t,l D is the TWED distance between the measurement point and the measurement value of another measurement point at time t and within the sliding time period length l prior to that time, and D i,t,l This is the total distance of strongly similar sets within the sliding time period l at time t and prior to it at the measurement point.
[0021] In the process of constructing the overall distance of a strongly similar set, first, the TWED distance is calculated by sliding the measured value of the measurement point with the measured values of other measurement points in the strongly similar set for each time interval, and then the mean square is applied to this, thereby obtaining the overall distance between the measured values of the measurement point and other measurement points in the strongly similar set, and the discriminant index X of the relationship of the measurement point to the strongly similar set is determined. m By comparing this with the initial determination result regarding the assignment relationship to strongly similar sets, D i,t,l ≤Xm In this case, at time t', the measurement point is still considered to belong to this strongly similar set, D i,t,l >X m In this case, at time t', the measurement point is considered not to belong to this strongly similar set. S207: For measurement points that are determined not to belong to this strongly similar set in the initial determination result regarding the association, a significant error identification process is performed on the measurement value to remove significant errors, and then the data is re-analyzed to obtain a supplementary determination result regarding the association. S208: Measurement points determined to belong to the original strongly similar set are composed or constructed as a subset of the strongly similar set. For measurement points determined not to belong to the original strongly similar set in the supplementary determination result regarding the belonging relationship, it is determined one by one whether they belong to this subset of the strongly similar set at that time. If it is determined that they belong to this strongly similar subset, then at time t', the measurement point is still considered to belong to the original strongly similar set. If it is determined that they do not belong to this strongly similar subset, then at time t', the measurement point is considered not to belong to this strongly similar set. S209: The determination results regarding the belonging relationships of all measurement points in the strongly similar set are integrated, and the determination result regarding the current composition of the strongly similar set under changes in the time-series data of the measured values is obtained.
[0022] Multiple measurement points within a strongly similar set exhibit similar change properties, reflecting the service state changes of a dam in a particular area or within a certain range. By employing a local linear embedding method based on dimensionality reduction, the monitoring data of the monitoring effect quantities at multiple measurement points within the strongly similar set are subjected to dimensionality reduction processing and merged into a one-dimensional overall effect quantity that can reflect the service state changes of the dam within its range, thereby achieving the objective of information fusion analysis.
[0023] As shown in Figure 4, the following is a basic flowchart for implementing a service status analysis method for dams within a single strongly similar set. S301: Normalization is performed on the monitored values of the monitoring effect index at each measurement point in the strongly similar set within the analysis time period 1 to t'. S302: The normalized multidimensional data is input into a dimensionality reduction algorithm using local linear embeddings, and the number of neighbors k is adjusted to obtain the initial dimensionality reduction result. S303: Based on the range of change of the time series data of the average value of the measured values at each measurement point of the strongly similar set, the value range is adjusted, and time series data Y of the overall effect amount that reflects the change in the service status of the dam in the area where the strongly similar set is located during that time period is obtained. S304: The cloud generator includes a positive cloud generator and an inverse cloud generator. Time-series data of the total effect size within the time zone 1 to (t'-1) is input to the inverse cloud generator to obtain three numerical properties: Ex, En, and He. The expected value Ex is the expected distribution of cloud drops in the argument. As the cloud drop distribution approaches the expected value, it becomes more concentrated, indicating a greater agreement in the understanding of the concept. As the cloud drop distribution moves away from the expected value, it becomes more dispersed, indicating a greater difference in the understanding of the concept. Entropy En is used to measure the degree of ambiguity of the qualitative concept. The larger the entropy, the more ambiguous the qualitative concept is. The wider the range of quantitative values that the qualitative concept corresponds to in the argument, the greater the degree of discreteness of the cloud graph. Hyperentropy He measures the uncertainty of entropy and reflects the stability of the indicator well, i.e., it reflects the degree of cohesion of cloud drops in the cloud graph. Therefore, hyperentropy is also called the entropy of entropy and is jointly determined by the ambiguity and randomness of entropy.
[0024] S305: Input the numerical characteristics Ex, En, and He into the positive cloud generator and calculate a cloud model that reflects the overall effect size distribution. S306: Based on the "3En" criteria for the cloud, the upper limit of the state discrimination index T corresponding to time t'. u and lower limit T l Set it. S307: Total effect size Y of the strongly similar set at time t' t’ and state discrimination index value T u and T l By comparing this with the previous set, we obtain the initial analysis results of the service status of dams within the range where this strongly similar set is located at that point in time. S308: Formula
number
[0025] The hierarchical analysis system of dam service status based on multiple strongly similar sets divides the entire dam into four layers from bottom to top, with each part (hereinafter abbreviated as element) within the same layer having the same grade, and elements belonging to two adjacent layers above and below have a whole-part relationship. The main contents of each layer and the relationships between adjacent layers are shown in Figure 3 and include the following: (1) Measurement point layer: Represents the monitoring information of the monitoring effect amount at each individual measurement point, and is the basic layer of the entire hierarchical analysis system of multiple strongly similar sets of dam service status; (2) Strongly similar set layer of the same part: Mining of strongly similar sets is performed on the measurement values of each monitoring effect amount measurement value within the same part range, and each measurement point is divided into a certain strongly similar set, and the strong similarity within the strongly similar set can be utilized, and the analysis results of the dam service status within the range where a relatively accurate single strongly similar set is located can be obtained. For example, each measurement point near the top of the dam in the river channel section has a relatively strong change similarity, and the single strongly similar set of dam service status By performing a service state analysis, service state information reflecting the dam crest portion within the river channel can be obtained. (3) Part layer: These are the main parts that can be specifically divided in the dam, and key parts and cross-sections can be further subdivided according to the actual situation. In this layer, the service state analysis results for the relevant part can be obtained through the hierarchical fusion of the service state analysis results of each strongly similar set in the strongly similar set layer. (4) Whole dam layer: The target of the whole layer is the service state of the entire dam, and the service state analysis results for the entire dam can be obtained through the hierarchical fusion of the service state analysis results of each layer below it.
[0026] This analysis examines the hierarchical integration process of the service status analysis results of the dam, and identifies an element in a certain layer c in the layer above it. j The l element sets that correspond to this hierarchy are
number
number
number
[0027] Let D be the decision matrix between the analysis results of the service status of each element in the same layer, and its expression is as follows.
number
[0028] At time t, element n in that hierarchy i The closer the analysis results of the service status of each element in the lower layer corresponding to the first element are, that is, the smaller the overall degree of discreteness, the more reliable and trustworthy the analysis results of the service status are. This invention uses a standard deviation method to measure and, at time t, element n i The degree of discreteness c of the analysis results of the service state of all m lower-level elements corresponding to this. i (t) The formula for calculating this is as follows:
number
[0029] In the formula, p k (t) p in the lower layer kThis is the result of the analysis of the service state of the element at time t,
number
[0030] Based on this, within time zones 1 to T, element n i The mean discrete degree value C of the analysis results of the service status of the corresponding lower-level elements. Di The following applies:
number
[0031] C Di The smaller the value, the more elements n in that hierarchy. i Because the degree of discreteness of the analysis results of the service status of the corresponding lower-level elements becomes smaller, when hierarchically merging to the upper level, element n i It should be given greater weight.
[0032] The more elements there are, the more monitoring information they contain, and the better they can reflect the changes in the dam's service status. Therefore, the quantitative characteristics of the elements can be measured and used as one of the criteria for determining their weights. The n elements in the relevant hierarchy i If the number of corresponding sub-elements is m, then the number of measured features C in each strongly similar set. Qi Calculate,
number
[0033] C Qi The larger the value, the more elements n in that hierarchy. i Because the number of corresponding lower-level elements increases, when hierarchically merging them to the upper level, element n i It should be given greater weight.
[0034] Based on the degree of discreteness of the analysis results of the service states of each element in the lower layer and the quantity characteristics of the elements in the lower layer within time periods 1 to T, the judgment matrix D at time point T (T) Among them, for element n in this layer i of n j the relative importance a of the analysis result of the service state of the dam of the element ij (T) can be obtained, and its calculation formula is as follows.
Number
[0035] Based on this, at time point T, the weight W of the influence exerted on the upper layer by the analysis result of the service state of element n in this layer i can be obtained, and its calculation formula is as follows. S (T)
Number
Number
[0036] In the formula
Number
Number
[0037] At time point T, based on the analysis results of the service states of the corresponding elements of element c in the upper layer and the weights of the influences exerted on the upper layer by each element of this layer, the analysis result c of the service state of element c in the upper layer j is fused and can be obtained, and its calculation formula is as follows. j analysis result c of the service state of element c j (T)
Number
Number
[0038] Assume that the dam is divided into E parts, and the analysis result QZ of the service state of each part at the T-th time point (T) is sorted in descending order, and the sorting result is as follows.
Number
[0039] QZ (T) If > 1, it reflects that the analysis result of the service state of the corresponding part is abnormal. In particular, for some parts with relatively large analysis results of the service state in the sorting result, key monitoring should be carried out.
[0040] As described above, the present invention has been shown and described with reference to specific preferred embodiments, but it should not be construed as a limitation on the present invention itself. Various changes can be made in its form and details without departing from the spirit and scope of the present invention defined in the appended claims.
[0041] The present invention belongs to the technical field of dam operation performance analysis. Specifically, it relates to a dam operation performance analysis method based on a strongly similar set, and includes the following steps. Divide the dam into multiple parts, and obtain the monitoring data of multiple measurement points of each part. Perform mining of strongly similar sets on the monitoring data of multiple measurement points within the same part to obtain a strongly similar set that includes all measurement points. Utilize the dam operation performance analysis method within the range of a single strongly similar set to obtain the dam operation performance analysis results within the range of each strongly similar set. Determine the weight of the influence of the dam operation performance analysis results of each strongly similar set on the upper part, and perform upward transfer fusion layer by layer on the analysis results of each strongly similar set to obtain the operation performance analysis results of each part in the part layer and the operation performance analysis results of the entire dam.
Claims
1. Step S1 involves dividing the dam into multiple sections and acquiring monitoring data from multiple measurement points in each section. Step S2 involves performing strong similarity set mining on monitoring data from multiple measurement points within the same area to obtain a strong similarity set encompassing all measurement points, employing a tracking analysis method for the composition status of the strong similarity set in relation to changes in the time-series data of the measured values, obtaining the determination result of the composition status of the strong similarity set at the current point in time under changes in the time-series data of the measured values, and constructing a hierarchical analysis system of the dam's service status based on multiple strong similarity sets. Step S3 involves using a method for analyzing the service status of dams within the range of a single strongly similar set to obtain the analysis results of the service status of dams within the range in which each strongly similar set is located. Step S4 involves calculating the degree of discreteness and quantitative characteristics of the service status analysis results for each element of the relevant hierarchy, constructing a decision matrix based on the calculated degree of discreteness and quantitative characteristics of the service status analysis results for each element, and determining the weights of the influence that the service status analysis results of each strongly similar set of dams have on the upper layers. A method for analyzing the service status of a dam based on strongly similar sets, characterized by including step S5, which sequentially fuses the analysis results of each strongly similar set upward in a hierarchical manner to obtain the analysis results of the service status of each part in the part layer and the analysis results of the service status of the entire dam.
2. The step of employing a tracking analysis method for the composition status of strongly similar sets for changes in the time-series data of the measured values in step S2 is as follows: S201: Perform normalization on the time series data of measurement values at each measurement point in a strongly similar set. S202: Determine the sliding time interval l according to the actual situation of changes in the measured values at each measurement point in the strongly similar set. S203: Within the sliding time period length l, calculate the time series data of the TWED distance between each measurement point and other measurement points in the strongly similar set for each time interval, and construct the time series data of the total distance in the strongly similar set for each measurement point. S204: Maximum value X of total distance within a fixed-length time period. ｍｉ The sample space will be used to extract data and calculate a discriminant index for the relationship of each measurement point to a strongly similar set. S205: Based on a typical small probability method, a discriminant index X is used to determine the relationship of each measurement point to a strongly similar set under a certain small probability. ｍ Calculating, S206: The total distance D' of the strongly similar set of the measurement point at the ith measurement point of the strongly similar set at the time of determination t'. i,t,l The discriminant index X of the relationship between the measurement point and its association with a strongly similar set is calculated. ｍ By comparing this with D', we obtain the initial determination result regarding the relationship to strongly similar sets. i,t,l ≤ X ｍ In this case, at time t', the measurement point is still considered to belong to this strongly similar set, D' i,t,l >X ｍ In this case, at time t', the measurement point is considered not to belong to this strongly similar set. S207: For measurement points that are determined not to belong to this strongly similar set in the initial determination result regarding the belonging relationship, a significant error identification process is performed on the measurement value, significant errors are removed, and then the data is re-analyzed to obtain a supplementary determination result regarding the belonging relationship. S208: Measurement points determined to belong to the original strongly similar set are constructed as a subset of the strongly similar set. For measurement points determined not to belong to the original strongly similar set in the supplementary determination result regarding the belonging relationship, it is determined one by one whether they currently belong to this subset of the strongly similar set. If it is determined that they belong to this strongly similar subset, at time t', the measurement point is considered to still belong to the original strongly similar set. If it is determined that they do not belong to this strongly similar subset, at time t', the measurement point is considered not to belong to this strongly similar set. S209: A method for analyzing the service status of a dam based on a strongly similar set, characterized in that it includes integrating the determination results regarding the belonging relationships of all measurement points in the strongly similar set and obtaining the determination result regarding the composition status of the strongly similar set at the present time under changes in the time-series data of the measured values.
3. The hierarchical analysis system for the service status of the dam is characterized in that the entire dam is divided into four layers from bottom to top: a measurement point layer, a layer of strongly similar sets of identical parts, a part layer, and a layer of the entire dam, and each element within the same layer has the same grade, and elements belonging to two adjacent layers above and below have a whole-part relationship. This is the method for analyzing the service status of a dam based on strongly similar sets as described in claim 1.
4. For an element in this hierarchy, an element c in the hierarchy one level above it ｊ The set of l corresponding elements in this hierarchy is [Math 1] The element n in the hierarchy is ｉ The m element sets that correspond to it in the next lower level are [Math 2] The method for analyzing the service status of a dam based on a strongly similar set as described in claim 3.
5. Step S3 is, S301: Perform normalization on the monitored values of the monitoring effect index quantities at each measurement point in the strongly similar set within the analysis time period 1 to t'. S302: Input the normalized multidimensional data into a dimensionality reduction algorithm using local linear embeddings, adjust the number of neighbors k, and obtain the initial dimensionality reduction result. S303: Adjust the range of the value based on the range of change of the time series data of the average value of the measured values at each measurement point of the strongly similar set, and obtain time series data Y of the overall effect amount that reflects the change in the service status of the dam in the range where the strongly similar set is located during that time period. S304: Time series data of the total effect size within the time zone 1 to (t'-1) is input into the inverse cloud generator, and three numerical characteristics Ex, En, and He are obtained, where Ex, En, and He are the expected value, entropy, and hyperentropy in the cloud model of the time series data of the total effect size of a strongly similar set, respectively. S305: Input the numerical characteristics Ex, En, and He into the positive cloud generator and calculate and obtain a cloud model that reflects the overall effect size distribution. S306: Based on the "3En" criteria of the cloud, the upper limit value T of the state discrimination index corresponding to time t'. ｕ and lower limit T ｌ To set, S307: Total effect size Y of the strongly similar set at time t' ｔ’ and state discrimination index value T ｕ and T ｌ By comparing this with the previous set, we obtain the initial analysis results of the service status of dams within the range where this strongly similar set is located at that point in time. S308: Formula [Math 3] The method for analyzing the service status of a dam based on a strongly similar set, as described in claim 4, further comprising: using to perform quantitative processing on the initial analysis results and obtaining analysis results of the service status of dams within the range in which this strongly similar set is located at that point in time.
6. The degree of discreteness C of the analysis results of the service status of each element in the hierarchy one level below the element in the current hierarchy. Ｄｉ Calculate, The quantitative characteristic C of the element in the next lower level that corresponds to the element in the current level. Qｉ Calculate, C Ｄｉ and C Ｑｉ Based on this, construct the decision matrix D, The following formula is used to determine the weight W of the impact of the analysis results of the service status of the elements at the relevant hierarchical level at time T on the upper levels. Ｓ （Ｔ） Calculate, [Math 4] In the above formula, [Math 5] is element n ｉ Other elements of the same layer n ｊ This is the sum of the relative importance of the following: [Math 6] The method for analyzing the service status of a dam based on a strongly similar set according to claim 5, wherein l is the sum of each element in the decision matrix D, and l is the number of elements in the relevant hierarchy.
7. At point T, element c in the hierarchy one level above. ｊ Based on the analysis results of the service status of each corresponding element in that hierarchy, and the weight of the influence each element in that hierarchy has on the hierarchy above, it merges with the element c in the hierarchy above. ｊ Analysis results of the service status c ｊ （Ｔ） The formula obtained is, [Number 7] The method for analyzing the service status of a dam based on a strongly similar set as described in claim 6.
8. The step of sequentially integrating the analysis results of each strongly similar set upwards in a hierarchical manner to obtain the analysis results of the service status of each part in the part layer and the analysis results of the service status of the entire dam is as follows: At point t, the set of analysis results for the service status of each element in the next lower level. [Number 8] Obtain the element n of the relevant hierarchy, merge it, and ｉ Analysis results of the service status n ｉ （ｔ） It obtains and further merges it with the element c of the next higher level. ｊ Analysis results of the service status c ｊ （ｔ） A method for analyzing the service state of a dam based on a strongly similar set, characterized in that it obtains a set and thereby completes a process of sequentially hierarchically merging from bottom to top.