Analysis method for hydrological data of water conservancy project

Abstract

The invention relates to the field of hydrological data processing, and particularly discloses an analysis method for hydrological data of a water conservancy project, which is used for improving the quality of the hydrological data and realizing optimized acquisition and analysis of the hydrological data of the water conservancy project so as to improve the precision of a hydrological test result. The method specifically comprises the following steps: selecting various types of water level error values of historically selected hydrological data under different acquisition frequencies in a fixed time period; the types of the water level error values comprise an average water level difference value, a highest water level difference value and a lowest water level difference value; the historical selected hydrological data with the mean value of the various types of water level error values exceeding the maximum value of the allowable error range is judged to be recorded as historical warning data; judging the historical selected hydrological data of which the mean value of the various types of water level error values does not exceed the maximum value of the allowable error range, and recording the historical selected hydrological data as historical normal data; and executing a hierarchical interpolation strategy on the historical normal data in combination with the historical warning data, and adjusting the acquisition frequency.

Description

Technical Field

[0001] This invention relates to the field of hydrological data processing, and more specifically to a method for analyzing hydrological data from water conservancy projects. Background Technology

[0002] Water conservancy projects are an important undertaking for humankind to utilize water resources, ensure water security, prevent floods and droughts, and improve the ecological environment. Hydrological data is the foundation for the design and management of water conservancy projects, and its accuracy and comprehensiveness directly affect the reliability and effectiveness of the projects.

[0003] Existing conventional methods for hydrological data processing include data quality inspection, data denoising, interpolation, and completion. By analyzing the processes and methods of data processing and preprocessing, the quality and integrity of the data can be ensured. In actual monitoring, hydrological data collected at fixed frequencies is complex and uncertain. Low-quality hydrological data can affect the accuracy of hydrological modeling and the efficiency of subsequent water conservancy project design and management, thereby reducing the accuracy of hydrological test results. Summary of the Invention

[0004] The purpose of this invention is to provide a method for analyzing hydrological data from water conservancy projects, thereby solving the following technical problems: How can we optimize the collection and analysis of hydrological data from water conservancy projects to improve the accuracy of hydrological measurement results?

[0005] The objective of this invention can be achieved through the following technical solutions: A method for analyzing hydrological data from water conservancy projects, comprising: Historical hydrological data within a fixed time period were selected, and water level error values ​​of various types were collected at different frequencies. The types of water level error values ​​included average water level difference, highest water level difference, and lowest water level difference. Historically selected hydrological data whose average water level error value for each type exceeds the maximum allowable error range are recorded as historical warning data. Historically selected hydrological data in which the mean value of each type of water level error does not exceed the maximum allowable error range are recorded as historical normal data. By combining historical warning data with historical normal data, a hierarchical interpolation strategy is implemented to adjust the data collection frequency.

[0006] Preferably, the method for selecting historical hydrological data is as follows: The system retrieves and records several sets of historical hydrological data on water quality levels from the database. The water quality levels are ranked from highest to lowest as Level 1, Level 2, and Level 3, with lower levels indicating poorer water quality. Historical hydrological data of several quality inspection levels are classified according to preset water level ranges; the preset water level ranges include warning water level ranges and non-warning water level ranges. Historical hydrological data that falls within the warning water level range will be marked as non-historical selected hydrological data. Historical hydrological data that fall within the non-warning water level range are marked as historical selected hydrological data.

[0007] Preferably, the method further includes: Historical warning data is divided into historical water quality warning data according to water quality level; Historical normal data are classified into historical normal water quality data according to water quality level; Historical water quality warning data and historical water quality normal data of the same water quality level were used as negative sample sets and positive sample sets, respectively, and then imported into the logistic regression model for training. The parameters of the logistic regression model are estimated, and the maximum likelihood estimation is used to solve the problem. The likelihood function is maximized to obtain the degree of water quality disturbance.

[0008] Preferably, the method for calculating the degree of water quality disturbance is as follows: For new data points The water quality grade model calculates the probability of disturbance.

[0009] in, This is a water quality status label, with a maximum value of 1; This is the feature vector of the new data point, which contains all feature values. The probability that the water quality is in a disturbed state given the conditions of a new data point; For the intercept term; This is a linear prediction value.

[0010] Preferably, the current level Interference level preset threshold range Comparison: like < If so, it is determined to be a low level of interference; like ≤ < If so, it is determined to be of medium interference level; like ≥ If so, it is determined to be a high level of interference.

[0011] Preferably, a hierarchical interpolation strategy is applied to historical normal data in conjunction with historical warning data. The hierarchical interpolation strategy is as follows: Historical alert data is divided into three levels according to the frequency of alert occurrence: high-frequency, medium-frequency, and low-frequency. Graded interpolation is then performed for each of the high-frequency, medium-frequency, and low-frequency zones. In the high-frequency region, interpolation and moving window weighting are performed based on time characteristics:

[0012] in, For time points interpolated value, Weights are based on historical data. For time points Historical observation values; The total number of time points, and ∈ ; This refers to the time period during which errors occur in the high-frequency region; Interpolation in the mid-frequency region is performed based on spatial correlation, using spatial kriging interpolation based on a variogram model.

[0013] in, For interpolation position interpolated value, The location to be interpolated; For the The weight of each monitoring station; This represents the total number of monitoring stations in space. For the Observations from each monitoring station; Low-frequency regions undergo seasonal decomposition based on statistical characteristics:

[0014] in, These are seasonally interpolated values. Time series Trend components; Time series Seasonal ingredients; Time series The residual components.

[0015] Preferably, the sampling frequency is adjusted and optimized parameters are generated, specifically including: Constructing a frequency adjustment model:

[0016] in, For the current point The value after adjusting the sampling frequency; The probability of the disturbance state; The rate of change of pollutant concentration; The standard deviation of historical concentrations; This represents the maximum value of the standard sampling frequency. This is the minimum standard sampling frequency. This is the initial sampling frequency.

[0017] Preferably, it also includes: determining the correlation of water quality disturbances based on multi-dimensional correlation analysis.

[0018] Preferably, multi-dimensional correlation analysis includes interference analysis through temporal correlation, spatial correlation, and indicator coupling: The time correlation is obtained by calculating the cross-correlation function:

[0019] in, The time-delay correlation coefficient; For the time lag step, and is an integer, and A value greater than 0 indicates the direction of pollution transmission. A value less than 0 indicates a reverse warning signal; Total number of points in time; The dominant sequence value; The time lag step is The lag sequence value at time, The mean of the dominant sequence, The mean of the lagged sequence; Standard deviation of the dominant sequence; The standard deviation of the lagged series; Spatial correlation is obtained by constructing the Moran index:

[0020] in, For spatial autocorrelation, if A value greater than 0 indicates a positive spatial correlation, suggesting the presence of diffuse pollution. If <0, it indicates a spatially negative correlation, which is a random distribution; , They are respectively in spatial locations and Water quality data at the location; This represents the average water quality across the entire region. This is the spatial weight matrix; This represents the total number of space monitoring stations. The coupling property of the index is obtained by establishing a cross-wavelet transform model:

[0021] in, The complex matrix of the cross-wavelet energy spectrum. As an indicator wavelet coefficients; As an indicator The conjugate of complex numbers; , These are temperature and flow rate indicators, respectively.

[0022] The beneficial effects of this invention are: This invention adjusts the collection frequency by combining historical warning data with historical normal data and implementing a hierarchical interpolation strategy. It also uses error patterns that have been identified as containing warning data to interpolate data at collection frequencies that have been marked as historical normal data, thereby reducing the actual collection frequency and ensuring cost or resource savings while filling in some data gaps. This invention determines the signal characteristics by combining historical warning data, ensuring that different warning frequencies are classified. The interpolation calculation method differs under different classifications, obtaining interpolation values ​​for different classifications. Based on the interpolation values, the corresponding acquisition frequency is adjusted. This achieves the goal of finding the most suitable actual acquisition frequency and calculation frequency for different hydrological stations, different monitoring targets (e.g., whether to focus on average values ​​or extreme values), and different time periods (flood season, non-flood season, special scheduling period) through a classified interpolation strategy based on warning data characteristics. This reduces the actual acquisition frequency to optimize resources. Furthermore, by maintaining and optimizing the key identified warning data, it ensures the dynamic adaptive frequency determination process. The method for analyzing hydrological data of water conservancy projects designed in this invention realizes hierarchical processing and optimization of hydrological data of water conservancy projects, improves the accuracy of hydrological measurement results and improves the quality of hydrological data. This design eliminates low-level data through water quality classification and filtering, and excludes data with warning water level information through water level interval screening, ensuring the stability of historical selected datasets.

[0023] Of course, any product implementing this invention does not necessarily need to achieve all the advantages described above at the same time. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating the steps of a method for analyzing hydrological data in water conservancy projects according to the present invention. Detailed Implementation

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0027] See also Figure 1 As shown, this invention provides a method for analyzing hydrological data from water conservancy projects. The method includes: Historical hydrological data within a fixed time period were selected, and water level error values ​​of various types were collected at different frequencies. The types of water level error values ​​included average water level difference, highest water level difference, and lowest water level difference. Historically selected hydrological data whose average water level error value for each type exceeds the maximum allowable error range are recorded as historical warning data. Historically selected hydrological data in which the mean value of each type of water level error does not exceed the maximum allowable error range are recorded as historical normal data. By combining historical warning data with historical normal data, a hierarchical interpolation strategy is implemented to adjust the data collection frequency.

[0028] In the above technical solution, to achieve hierarchical processing and optimization of hydrological data from water conservancy projects, improve the accuracy of hydrological measurement results, and enhance the quality of hydrological data, this design employs water quality grading and filtering to eliminate low-grade data, and excludes data containing warning water levels through water level interval screening, ensuring the stability of the selected historical dataset. This is specifically achieved through the following steps: First, select historical hydrological data within a fixed time period at different collection frequencies to calculate various types of water level error values. This is typically done within a flood season in the field of water conservancy engineering, or over a year or quarter. Historical data within this time period is collected, with the collection frequency based on the original baseline data (such as high-frequency raw data collected every 1 minute, 5 minutes, or 10 minutes) and calculated at different frequencies (such as hourly average, three-hourly average, six-hourly average, and daily average). The calculation of water level error values ​​involves calculating the difference between each calculation frequency (e.g., hourly average) and a reference benchmark (usually the highest frequency data collected during the same period or its calculation result). The types of water level error values ​​include average water level difference, highest water level difference, and lowest water level difference. The methods for calculating these three types of water level error values ​​are as follows: Average water level difference: The average (or a central trend value) of the differences between the calculated values ​​and the reference benchmark values ​​at all calculation points within the time period. Maximum water level difference: The difference between the highest water level calculated within a time period and its corresponding reference benchmark (actual highest water level); The difference between the lowest water level calculated within the lowest water level difference period and its corresponding reference benchmark (actual lowest water level).

[0029] Then, historical selected hydrological data whose average water level error value for each type exceeds the maximum allowable error range are recorded as historical warning data; next, historical selected hydrological data whose average water level error value for each type does not exceed the maximum allowable error range are recorded as historical normal data.

[0030] The above method uses error evaluation to determine the magnitude of water level errors. Specifically, the evaluation method involves determining whether the mean of the water level error values ​​exceeds the maximum allowable error range. For each type of water level error (average water level, highest water level, and lowest water level), an acceptable maximum threshold is set. These thresholds need to be scientifically determined based on the actual requirements of hydrological monitoring (such as flood warning accuracy and water resource allocation accuracy). The classification is based on the time period average: For the historical data corresponding to each calculation frequency (i.e., the error set calculated by each different frequency), the following calculations are performed: the time period average of all average water level differences at that frequency; the time period average of all highest water level differences at that frequency; and the time period average of all lowest water level differences at that frequency.

[0031] First, the determination method for historical warning data relies on classification rules. Classification is based on the overall average change of a set of historical error data corresponding to a specific calculation frequency, i.e., the time-period average. For data at a specific calculation frequency, if any of the "time-period average of the average water level difference, the time-period average of the highest water level difference, or the time-period average of the lowest water level difference" exceeds the error allowable range set for its type, then this period of historical hydrological data at that frequency is marked as "historical warning data". Secondly, for data at a specific calculation frequency, if the "time-period average of the average water level difference, the time-period average of the highest water level difference, and the time-period average of the lowest water level difference" all do not exceed the set error tolerance range, then this period of historical hydrological data at that frequency is marked as "historical normal data"; historical warning data indicates that the overall error level of the water level values ​​(at least one of the average, highest, and lowest values) calculated at that frequency exceeds the safe tolerance range; historical normal data indicates that the overall error level at that frequency is acceptable.

[0032] Finally, a tiered interpolation strategy is implemented on historical normal data based on historical warning data to adjust the acquisition frequency. The above utilizes the error patterns identified as containing warning data characteristics. In subsequent actual monitoring, data interpolation is performed on the acquisition frequencies marked as historical normal data. The purpose of data interpolation is to attempt to reduce the actual acquisition frequency while maintaining the basic hydrological accuracy requirements at the acquisition frequency, thereby saving costs or resources, or filling in some data gaps. By combining the signal characteristics of historical warning data, different levels are determined based on the frequency of warning occurrence. The interpolation calculation method differs under different levels, obtaining interpolation values ​​for different levels. The corresponding acquisition frequency is adjusted based on the interpolation values. This tiered interpolation strategy based on warning data characteristics helps find the most suitable actual acquisition frequency and calculation frequency for different hydrological stations, different monitoring targets (e.g., whether to focus on average values ​​or extreme values), and different time periods (flood season, non-flood season, special scheduling period). This reduces the actual acquisition frequency to optimize resources and also maintains and optimizes the key identified warning data, ensuring a dynamic adaptive frequency determination process.

[0033] As one embodiment of the present invention, the method for selecting historical hydrological data is as follows: The system retrieves and records several sets of historical hydrological data on water quality levels from the database. The water quality levels are ranked from highest to lowest as Level 1, Level 2, and Level 3, with lower levels indicating poorer water quality. Historical hydrological data of several quality inspection levels are classified according to preset water level ranges; the preset water level ranges include warning water level ranges and non-warning water level ranges. Historical hydrological data that falls within the warning water level range will be marked as non-historical selected hydrological data. Historical hydrological data that fall within the non-warning water level range are marked as historical selected hydrological data.

[0034] In the above technical solution, several sets of historical hydrological data are retrieved from the database. These historical hydrological data are grouped according to their quality inspection levels, which are pre-classified into Level 1, Level 2, and Level 3. The lower the level, the worse the water quality. That is, Level 1 represents the highest quality and most reliable hydrological data; Level 2 represents average quality, including some possible small errors and uncertainties; and Level 3 represents poor quality, low reliability, and the possibility of significant errors or problems. Furthermore, a preset water level range is defined, which includes a warning water level range and a non-warning water level range. The warning water level range refers to the water level range that is close to or exceeds the design safety limit of the water conservancy project, including the water level range that may cause flood risk or other disasters; the non-warning water level range refers to the water level range that is below the warning water level and is in a relatively safe or normal operating condition. Based on the group quality inspection level data obtained in the previous step, and according to the recorded water level height values, the data is divided into two preset water level height ranges: Historical hydrological data within the warning water level range are marked as non-historical selected hydrological data, while historical hydrological data outside the warning water level range are marked as historical selected hydrological data. The reason for excluding non-historical selected hydrological data is that high-risk hydrological data at these high water levels requires further processing. Excluding data with consistently serious quality problems facilitates relatively stable hydrological conditions during non-warning water level periods. The determination of historical selected hydrological data helps ensure that the acquisition system is less affected by external environmental factors such as water level issues and reduces the influence of mixed non-frequency factors, thereby optimizing hydrological acquisition under normal operating conditions. Furthermore, this acquisition method is more in line with the needs of routine water resource management and ensures reduced management costs. As one embodiment of the present invention, the method further includes: Historical warning data is divided into historical water quality warning data according to water quality level; Historical normal data are classified into historical normal water quality data according to water quality level; Historical water quality warning data and historical water quality normal data of the same water quality level were used as negative sample sets and positive sample sets, respectively, and then imported into the logistic regression model for training. The parameters of the logistic regression model are estimated, and the maximum likelihood estimation is used to solve the problem. The likelihood function is maximized to obtain the degree of water quality disturbance.

[0035] In the above technical solution, water quality level is used as a variable, and a logistic regression model is used to quantify the degree of interference of water quality on the reliability of current water level data. The physical characteristics of the water quality data are modeled and linked with historical warning data to ensure integrated water quality interference analysis. This allows for the refinement of warning data according to water quality level: historical warning data is divided into historical warning data and historical normal data into historical normal data. Historical warning data and historical normal data of the same water quality level are then used as negative and positive sample sets, respectively, and imported into the logistic regression model for training. For example, the negative sample set represents water quality level one. The system uses historical water quality warning data and a positive sample set of historical normal water quality data at water quality level 1. It establishes a link between data warning and normal quality inspection based on water quality levels. By constructing a training set for a logistic regression model, it ensures a training foundation for each specific water quality level. By treating water quality level as a categorical variable, the logistic regression model can capture the basic probability differences in warning data at different levels. When the output parameter estimates are correlated with water quality levels, the parameter estimates represent the quantification of water quality interference caused by that water quality level as a feature. By integrating water quality data with historical warning and normal data, the system achieves the analysis of water quality interference probability.

[0036] Specifically, as one embodiment of the present invention, parameter estimation is performed on the logistic regression model, and the likelihood function is maximized through maximum likelihood estimation to obtain the degree of water quality disturbance; the calculation method for the degree of water quality disturbance is as follows: For new data points The water quality grade model calculates the probability of disturbance.

[0037] in, This is a water quality status label, with a maximum value of 1; This is the feature vector of the new data point, which contains all feature values. The probability that the water quality is in a disturbed state given the conditions of a new data point; For the intercept term; This is a linear prediction value.

[0038] In the above technical solutions, among which This is a water quality status label with a maximum value of 1. It represents historical warning data when the water quality is in a disturbed state; and normal water quality when it is in a normal state. It should be 0, indicating historical normal data; for linear prediction calculations, Specifically:

[0039] in, Linear predicted value , The intercept term of the model is a constant; For corresponding features The regression coefficients; Represents a new point The feature vector contains an intercept term that ends in 1.

[0040] As one embodiment of the present invention, the current level Interference level preset threshold range Comparison: like < If so, it is determined to be a low level of interference; like ≤ < If so, it is determined to be of medium interference level; like ≥ If so, it is determined to be a high level of interference.

[0041] In the above technical solutions, the above-obtained To give a new data point The probability that the water quality is in a disturbed state under certain conditions, and The value is at this point The degree of water quality disturbance is greater than 0, and The larger the value, the greater the interference of water quality on data reliability, indicating a higher likelihood of being in a state of interference. The specific judgment condition is: if... < If it is, then it is determined to be a low level of interference; if ≤ < If so, it is determined to be a medium level of interference; if ≥ If so, it is determined to be a high level of interference.

[0042] Assume the coefficients of the trained water quality rating model are: (intercept); =1.5 (coefficient of Class II water quality relative to Class I); New data point characteristics: Water quality level: Class II → =1 Other features are omitted for now); Calculation process:

[0043]

[0044] make for The water quality interference level of the above secondary water quality data point is 37.75%, indicating a moderate probability of being in an interference state similar to historical warning data.

[0045] As one embodiment of the present invention, a hierarchical interpolation strategy is performed on historical normal data in conjunction with historical warning data. The hierarchical interpolation strategy is as follows: Historical alert data is divided into three levels based on the frequency of alert occurrence: high-frequency, medium-frequency, and low-frequency. Hierarchical interpolation is then performed for each of these three levels to accurately improve data integrity through risk stratification, while ensuring optimized accuracy and security in critical scenarios. The specific interpolation calculation method is as follows: Firstly, in the high-frequency region, interpolation and moving window weighting are performed based on time characteristics:

[0046] in, For time points interpolated value, For time points Historical observation values; The total number of time points, and ∈ ; The historical data weights are the decay weights; the calculation formula is:

[0047] in, This refers to the time period in which errors occur in the high-frequency area. By weighted fusion of data from adjacent periods during the time periods in which errors frequently occur (such as the gate control period during heavy rain), the interpolation results are avoided from being affected by a single outlier. This ensures that in high-risk scenarios (such as flood control scheduling), interpolation stability is achieved at the cost of sacrificing some timeliness, thus preventing systemic misjudgments caused by error propagation.

[0048] Secondly, in the mid-frequency region, spatial kriging interpolation based on a variogram model is performed based on spatial correlation.

[0049] in, For interpolation position interpolated value, The location to be interpolated; For the The weight of each monitoring station; This represents the total number of monitoring stations in space. For the The observed values ​​at the monitoring station; Weight of each monitoring station The calculation process is as follows:

[0050]

[0051] in, For monitoring sites With monitoring stations distance, It is a variogram; the constraint condition is... And the interpolation variance Minimize; by leveraging watershed topological relationships (such as the correlation between upstream and downstream water levels), effective information from discrete stations is transmitted to missing areas to achieve spatial collaborative completion, avoiding local data black holes caused by single-point failures and ensuring the spatial integrity of inputs to hydraulic models (such as hydrodynamic simulations).

[0052] Third, seasonal decomposition of the low-frequency region is performed based on statistical characteristics:

[0053] in, These are seasonally interpolated values. Time series Trend components; Time series Seasonal ingredients; Time series The residual components; by separating the long-term trend Seasonal cycles Then, only for low-impact residuals Interpolation is performed to eliminate redundant noise, ensuring that lightweight interpolation is used during low-risk periods (such as the dry season) and reducing the frequency of calls to high-cost spatial / real-time computing (such as satellite remote sensing and high-frequency sensing).

[0054] The above approach achieves Pareto optimality for both safety and efficiency by dynamically matching historical accident frequencies (high / medium / low frequency) with interpolation costs (real-time performance / computational load / accuracy requirements). For example, high-frequency regions utilize real-time sensor data windows for interpolation, while low-frequency regions reuse data from the previous month's cycle. The three-level zoning strategy enables calculation and analysis at different frequencies. High-frequency region window interpolation blocks the cross-period transmission of water level anomalies, reducing the risk of flood misjudgment by 23%. Mid-frequency regions employ spatial correlation-based interpolation using spatial Kriging interpolation based on a variogram model. Spatial reconstruction improves the Nash efficiency coefficient (NSE) of the watershed hydrodynamic model by 0.11. Seasonal interpolation in low-frequency regions reduces the data acquisition frequency during non-critical periods by 50%.

[0055] As one embodiment of the present invention, adjusting the acquisition frequency and generating optimized parameters specifically includes: Constructing a frequency adjustment model:

[0056] in, For the current point The value after adjusting the sampling frequency; The probability of the disturbance state; The rate of change of pollutant concentration; The standard deviation of historical concentrations; This represents the maximum value of the standard sampling frequency. This is the minimum standard sampling frequency. This is the initial sampling frequency.

[0057] In the above technical solution, when calculating the acquisition frequency information of the current monitoring point, the probability of interference state of the initial acquisition output value is determined. Is the size within range? If so, adjust the frequency value. If greater than ,but ;otherwise Historical concentration standard deviation ,in, This represents the total number of historical data collections. For the The pollutant concentration was collected in the second sample. This represents the historical average concentration of pollutants.

[0058] As one embodiment of the present invention, it also includes: judging the correlation of water quality interference based on multi-dimensional correlation analysis.

[0059] As one embodiment of the present invention, multi-dimensional correlation analysis includes interference analysis through time correlation, spatial correlation, and index coupling; the three-dimensional analysis shortens the time of traditional manual investigation and ensures a rapid improvement in the efficiency of pollution source tracing.

[0060] Firstly, the time correlation is obtained by calculating the cross-correlation function:

[0061] in, The time-delay correlation coefficient; For the time lag step, and is an integer, and A value greater than 0 indicates the direction of pollution transmission. A value less than 0 indicates a reverse warning signal; This represents the total number of time points; The dominant sequence value; The time lag step is The lag sequence value at time, The mean of the dominant sequence, The mean of the lagged sequence; Standard deviation of the dominant sequence; The standard deviation of the lag sequence is given; a lag timetable is established based on the pollution location and source tracing, as shown in Table 1 below; Table 1

[0062] Secondly, spatial correlation is obtained by constructing the Moran index:

[0063] in, For spatial autocorrelation, if A value greater than 0 indicates a positive spatial correlation, suggesting the presence of diffuse pollution. If <0, it indicates a spatially negative correlation, which is a random distribution; , They are respectively in spatial locations and Water quality data at the location; This represents the average water quality across the entire region. For the spatial weighting matrix, during the design process, when hydrological connectivity is maintained, i.e., the water flow is continuous, ;in, The distance is the hydrodynamic distance (km), obtained through GIS hydrological network analysis. This is the spatial attenuation coefficient (km), determined based on river type; when there is no hydrological connectivity, disconnectivity means that different watersheds are isolated. =0; The total number of space monitoring stations is shown in Table 2 below. The criteria for judging different spaces are shown in Table 2 below. Table 2

[0064] Third, the coupling property of the index is obtained by establishing a cross-wavelet transform model:

[0065] in, The complex matrix of the cross-wavelet energy spectrum. As an indicator wavelet coefficients; As an indicator The conjugate of complex numbers; , These are temperature and flow velocity indices, respectively. When water temperature rises, algae bloom, showing a homogeneous relationship; when flow velocity increases, sediment is released, showing an inverse relationship. Regional analysis using different energy spectrum characteristics of specific sizes can accurately determine this, as shown in Table 3 below. Table 3

[0066] The above process of joint analysis using three dimensions is as follows: Receive water quality anomaly alarms and determine the correlation of feedback time. Is it greater than 0.8? If so, then it is determined to be point source pollution diffusion; If not, then further determine the clustering of the feedback space. Is it greater than 0.7? If so, it is determined to be non-point source pollution input; If not, then determine the feedback cross wavelet energy. Whether it is centralized; If so, it is determined to be driven by a specific indicator; If not, it is determined to be a natural disturbance or measurement error; In actual experiments, the three-dimensional joint analysis significantly improved the pollution source tracing time, as shown in Table 4 below: Table 4

[0067] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0068] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended documents. In some cases, the actions or steps described in this application may be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this application, they should all fall within the protection scope of the present invention.

Claims

1. A method for analyzing hydrological data from water conservancy projects, characterized in that, The method includes: Historical hydrological data within a fixed time period were selected, and water level error values ​​of various types were collected at different frequencies. The types of water level error values ​​included average water level difference, highest water level difference, and lowest water level difference. Historically selected hydrological data whose average water level error value for each type exceeds the maximum allowable error range are recorded as historical warning data. Historically selected hydrological data in which the mean value of each type of water level error does not exceed the maximum allowable error range are recorded as historical normal data. By combining historical warning data with historical normal data, a hierarchical interpolation strategy is implemented to adjust the data collection frequency.

2. The method for analyzing hydrological data of water conservancy projects according to claim 1, characterized in that, The method for selecting the historical hydrological data is as follows: The system retrieves and records several sets of historical hydrological data on water quality inspection levels from the database. The water quality inspection levels are ranked from highest to lowest as Level 1, Level 2, and Level 3, with lower levels indicating poorer water quality. The historical hydrological data of the aforementioned several groups of quality inspection levels are classified according to a preset water level range; the preset water level range includes the warning water level range and the non-warning water level range. Historical hydrological data that fall within the aforementioned warning water level range will be marked as non-historical selected hydrological data. Historical hydrological data that fall within the aforementioned non-warning water level range are marked as historical selected hydrological data.

3. The method for analyzing hydrological data of water conservancy projects according to claim 2, characterized in that, The method further includes: Historical warning data is divided into historical water quality warning data according to water quality level; Historical normal data are classified into historical normal water quality data according to water quality level; Historical water quality warning data and historical water quality normal data of the same water quality level were used as negative sample sets and positive sample sets, respectively, and then imported into the logistic regression model for training. The parameters of the logistic regression model are estimated, and the maximum likelihood estimation is used to solve the problem. The likelihood function is maximized to obtain the degree of water quality disturbance.

4. The method for analyzing hydrological data of water conservancy projects according to claim 3, characterized in that, The method for calculating the degree of water quality disturbance is as follows: For new data points The water quality grade model calculates the probability of disturbance. in, This is a water quality status label, with a maximum value of 1; This is the feature vector of the new data point, which contains all feature values. The probability that the water quality is in a disturbed state given the conditions of a new data point; For the intercept term; This is a linear prediction value.

5. The method for analyzing hydrological data of water conservancy projects according to claim 4, characterized in that, The current level Interference level preset threshold range Comparison: like < If so, it is determined to be a low level of interference; like ≤ < If so, it is determined to be of medium interference level; like ≥ If so, it is determined to be a high level of interference.

6. The method for analyzing hydrological data of water conservancy projects according to claim 1, characterized in that, The step involves combining historical warning data with historical normal data to perform a hierarchical interpolation strategy. The hierarchical interpolation strategy is as follows: Historical alert data is divided into three levels according to the frequency of alert occurrence: high-frequency, medium-frequency, and low-frequency. Graded interpolation is then performed for each of the high-frequency, medium-frequency, and low-frequency zones. In the high-frequency region, interpolation and moving window weighting are performed based on time characteristics: in, For time points interpolated value, Weights are based on historical data. For time points Historical observation values; The total number of time points, and ∈ ; This refers to the time period during which errors occur in the high-frequency region; Interpolation in the mid-frequency region is performed based on spatial correlation, using spatial kriging interpolation based on a variogram model. in, For interpolation position interpolated value, The location to be interpolated; For the The weight of each monitoring station; This represents the total number of monitoring stations in space. For the Observations from each monitoring station; Low-frequency regions undergo seasonal decomposition based on statistical characteristics: in, These are seasonally interpolated values. Time series Trend components; Time series Seasonal ingredients; Time series The residual components.

7. The method for analyzing hydrological data of water conservancy projects according to claim 1, characterized in that, The adjustment of the acquisition frequency and the generation of optimized parameters specifically include: Constructing a frequency adjustment model: in, For the current point The value after adjusting the sampling frequency; The probability of the disturbance state; The rate of change of pollutant concentration; The standard deviation of historical concentrations; This represents the maximum value of the standard sampling frequency. This is the minimum standard sampling frequency. This is the initial sampling frequency.

8. The method for analyzing hydrological data of water conservancy projects according to claim 1, characterized in that, Also includes: The correlation between water quality disturbances is determined by conducting multi-dimensional correlation analysis.

9. The method for analyzing hydrological data of water conservancy projects according to claim 8, characterized in that, The multi-dimensional correlation analysis includes interference analysis through temporal correlation, spatial correlation, and indicator coupling: The time correlation is obtained by calculating the cross-correlation function: in, The time-delay correlation coefficient; For the time lag step, and is an integer, and A value greater than 0 indicates the direction of pollution transmission. A value less than 0 indicates a reverse warning signal; Total number of points in time; The dominant sequence value; The time lag step is The lag sequence value at time, The mean of the dominant sequence, The mean of the lagged sequence; Standard deviation of the dominant sequence; The standard deviation of the lagged series; Spatial correlation is obtained by constructing the Moran index: in, For spatial autocorrelation, if A value greater than 0 indicates a positive spatial correlation, suggesting the presence of diffuse pollution. If <0, it indicates a spatially negative correlation, which is a random distribution; , They are respectively in spatial locations and Water quality data at the location; This represents the average water quality across the entire region. This is the spatial weight matrix; This represents the total number of space monitoring stations. The coupling property of the index is obtained by establishing a cross-wavelet transform model: in, The complex matrix of the cross-wavelet energy spectrum. As an indicator wavelet coefficients; As an indicator The conjugate of complex numbers; , These are temperature and flow rate indicators, respectively.

Images