Water conservancy project monitoring method and system based on water wave analysis

Abstract

The invention relates to the technical field of mode recognition, in particular to a hydraulic engineering monitoring method and system based on water wave analysis, and the method comprises the following steps: based on a time sequence of a fluid particle motion trail, combining with the spatial change of a water flow velocity field, carrying out the difference calculation of the position and the velocity, and carrying out the superposition operation of fusing the curl and shear. According to the method, the local dynamic feature matrix is generated by combining entropy screening, and the local change capturing precision is improved. An overall dynamic feature matrix is generated based on convolution operation of time distribution and a position matrix, and the dynamic distribution and time correlation expression ability is enhanced. Efficient fusion of feature data sets is realized through sensitive weight distribution and distribution matching processing of eddy current distribution, and the accuracy of trend analysis is enhanced. A multi-scale dynamic rule is captured by using a rule matrix constructed by multi-frequency decomposition and a time window, dynamic trend prediction is completed by combining offset recursion extraction, and comprehensive analysis and high-precision prediction of water area changes are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of pattern recognition, and in particular to a water conservancy project monitoring method and system based on water pattern analysis. Background Art

[0002] The field of pattern recognition technology includes methods for analyzing, classifying and understanding data, aiming to automatically identify various data patterns through computers or other devices. The core content of pattern recognition includes feature extraction, data classification and clustering, and learning and modeling of patterns. Its application scope covers many fields such as image recognition, speech recognition, biometric recognition and environmental monitoring. In this technical field, by performing feature analysis on massive data and adopting classification or modeling methods, it is possible to accurately identify the laws or specific patterns contained in the data, thereby providing basic support for solving complex problems.

[0003] Among them, the water conservancy project monitoring method based on water ripple analysis refers to the monitoring and management of the operation status of water conservancy projects by analyzing and processing the dynamic change information of water ripples on the water surface. This method mainly obtains the original data of water surface movement through water ripple acquisition technology, extracts the specific characteristic parameters of water ripples through feature analysis technology, and analyzes and models the water ripple changes in combination with hydrodynamic analysis theory, thereby identifying the relevant status information of water conservancy projects. This method covers the links of water ripple data acquisition, feature extraction and parameterization, and status judgment based on water ripple features, aiming to achieve real-time monitoring and management of the status of water conservancy projects through systematic analysis.

[0004] The existing technology lacks in-depth analysis of the multi-level correlation between particle motion and velocity field in dynamic feature extraction, resulting in low accuracy in capturing local dynamic features. The overall feature expression method cannot fully correlate the changes in time and space, and there is a separation between dynamic distribution and overall trend analysis, which affects data correlation and overall monitoring effects. The lack of comprehensive application of multi-frequency decomposition and time window law analysis makes it impossible to effectively analyze multi-scale dynamic changes, resulting in a single law extraction. The prediction of offset lacks recursive calculation support, and trend prediction is not accurate enough, which limits the application adaptability in complex water conservancy project scenarios. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a water conservancy project monitoring method and system based on water pattern analysis.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a water conservancy project monitoring method based on water pattern analysis, comprising the following steps:

[0007] S1: Based on the time series of fluid particle motion trajectories, combined with the spatial variation of the water velocity field, through the differential calculation of position and velocity, the superposition operation of fusion curl and shear, all parameters are normalized and the entropy parameters are screened, and the local dynamic feature matrix is ​​generated through dynamic feature extraction;

[0008] S2: Based on the local dynamic feature matrix and in combination with the time distribution of water level changes, a position dynamic matrix is ​​formed by dependency value calculation, the position matrix is ​​convolved with the time-related parameter to form a time matrix, and dynamic distribution processing is performed after superimposing the position and time matrices to generate an overall dynamic feature matrix;

[0009] S3: Based on the local dynamic feature matrix and the overall dynamic feature matrix, combined with the associated parameters in the eddy distribution, the matrix is ​​normalized and decomposed item by item through the local feature sensitivity weight allocation and the overall associated parameter superposition processing, and the dynamic trend is integrated through distribution matching to generate a water area dynamic feature data set;

[0010] S4: Based on the water area dynamic characteristic data set, combined with the signal parameters of the water body energy field, the data is divided by time window and multi-frequency decomposition is performed, the amplitude and trend of the change within the time period are extracted, the amplitude and trend are clustered and classified, and a matrix is ​​constructed by time scale to generate a multi-scale dynamic law matrix;

[0011] S5: Based on the multi-scale dynamic law matrix, the future change data of the water velocity field is intercepted through a sliding window, and the offset is differentially extracted by combining the recursive operation of the short-term amplitude and the long-term trend. The offset change range is combined with the trend matrix to generate a dynamic trend offset prediction result.

[0012] The local dynamic feature matrix includes position parameters, velocity parameters, curl parameters, and shear parameters. The position dynamic matrix includes position correlation values ​​and time correlation values. The overall dynamic feature matrix includes local feature sensitivity weights, overall correlation parameters, and dynamic distribution parameters. The water area dynamic feature data set specifically includes dynamic trend distribution parameters, eddy distribution parameters, and water body energy field signal parameters. The multi-scale dynamic law matrix includes time change amplitude, time change trend, and time scale matrix. The dynamic trend offset prediction results include short-term amplitude recursive results, long-term trend recursive results, and offset change range.

[0013] As a further solution of the present invention, the step of obtaining the local dynamic feature matrix is ​​specifically as follows:

[0014] S111: extracting the time series data of fluid particles, calculating the displacement difference of particles at continuous time points, and performing superposition operation in combination with the spatial change of the water flow velocity field to generate a dynamic change matrix of position and velocity;

[0015] S112: extracting the rotation and shear components from the position and velocity dynamic change matrix, calculating the combined superposition value of the rotation and shear components, and normalizing them to establish a superposition feature matrix;

[0016] S113: Calculate the entropy value of the superimposed feature matrix, select key entropy value parameters, set dynamic threshold screening conditions, and use the formula:

[0017]

[0018] Generate a local dynamic feature matrix;

[0019] Among them, H represents the entropy value calculation result, V i represents the value of the i-th component in the superimposed feature matrix, ω i represents the curl of the ith component, represents the time derivative of the curl, α represents the adjustment factor used to adjust the influence of the time derivative of the curl, represents the sum of all component values ​​of the superimposed feature matrix, represents the normalized logarithm of the ith component, and n is the total number of features contained in the superimposed feature matrix, which determines the dimension of the data to be processed and the range of entropy calculation.

[0020] As a further solution of the present invention, the step of obtaining the overall dynamic feature matrix is ​​specifically as follows:

[0021] S211: extracting parameter components from the local dynamic feature matrix, calculating dependency values ​​of multiple components in combination with the time distribution of water level changes, and forming a component dependency value matrix;

[0022] S212: calling the component dependency value matrix, performing differential calculation on the dependency value and the position parameter matrix, performing convolution calculation on the time correlation parameters in combination with the time distribution characteristics, and generating a time correlation matrix;

[0023] S213: Perform element-by-element superposition calculation on the time correlation matrix and the position matrix, using the formula:

[0024]

[0025] The dynamic distribution matrix of position and time is superimposed to generate the overall dynamic feature matrix;

[0026] Among them, P ij Represents the element value of the superimposed dynamic distribution matrix, V ij Represents the element value of the position matrix, T ij Represents the element value of the time-related matrix, S ij represents the local increment in the time correlation matrix, γ is the adjustment parameter, and β is the time distribution adjustment factor.

[0027] As a further solution of the present invention, the steps of acquiring the water area dynamic characteristic data set are specifically:

[0028] S311: using the data of the local dynamic characteristic matrix and the overall dynamic characteristic matrix, calculating the associated parameters in the eddy current distribution, taking the eddy current intensity and the flow direction as basic inputs for weight distribution, and generating an eddy current characteristic weighted matrix;

[0029] S312: combining the eddy current feature weighting matrix, performing sensitivity analysis on local features, calling weight values, and dynamically superimposing them with overall correlation parameters to generate a comprehensive feature dynamic matrix;

[0030] S313: normalizing the comprehensive feature dynamic matrix item by item, using the formula:

[0031]

[0032] Perform feature decomposition, fuse dynamic trends through distribution matching algorithm, and generate water area dynamic feature data set;

[0033] Among them, M ij Represents the normalized matrix element value, W ij Represents the elements of the eddy current feature weighting matrix, G ij Represents the elements of the overall correlation parameter matrix, and n represents the number of columns in the matrix.

[0034] As a further solution of the present invention, the step of obtaining the multi-scale dynamic law matrix is ​​specifically as follows:

[0035] S411: Based on the water body dynamic characteristic data set, call the water body energy field signal parameters, combine the signal amplitude and time window division, perform segmented calculation, and generate an energy field time distribution matrix;

[0036] S412: using the energy field time distribution matrix, performing a multi-frequency decomposition operation, extracting the signal change amplitude and trend characteristics in multiple time periods, performing segmented clustering calculations in combination with the local extreme value points of the signal amplitude, and generating an amplitude and trend clustering matrix;

[0037] S413: The amplitude and trend clustering matrix is ​​multi-dimensionally expanded in combination with the time scale parameter, using the formula:

[0038]

[0039] The time scale characteristic matrix is ​​established by fusing the amplitude and trend matrix, and then normalized to generate a multi-scale dynamic law matrix;

[0040] Among them, D ij Multi-scale dynamic law matrix, Ti represents the time scale factor, F j Represents the signal frequency factor, A ijk is the amplitude parameter of the clustering matrix, B ijk is the trend parameter of the clustering matrix, and n is the number of features in the time period.

[0041] As a further solution of the present invention, the steps for obtaining the dynamic trend shift prediction result are specifically as follows:

[0042] S511: Based on the multi-scale dynamic law matrix, a sliding window technology is used to intercept the future change data of the water flow velocity field in a short period of time to obtain the original input data of the future trend;

[0043] S512: performing recursive operations on the short-term amplitude and the long-term trend of the original input data of the future trend, calculating the amplitude and the trend in multiple time windows in the future, and generating trend analysis results;

[0044] S513: Perform differential extraction on the trend analysis results, analyze the offset change, and combine the multi-scale dynamic law matrix to use the formula:

[0045]

[0046] Integrate the change amplitude and trend of time series to generate dynamic trend shift prediction results;

[0047] Among them, ΔP t represents the offset prediction result, S t represents the trend data extracted from the multi-scale dynamic law matrix, α represents the weight factor of short-term changes, β represents the integral coefficient of long-term trends, represents the time derivative of trend data, It represents the integral of trend data within the time window T, reflecting the dynamic changes on the time scale.

[0048] A water conservancy project monitoring system based on water ripple analysis, the water conservancy project monitoring system based on water ripple analysis is used to execute the water conservancy project monitoring method based on water ripple analysis, the system comprises:

[0049] The local feature processing module extracts the spatial variation data of the water velocity field based on the time series of the fluid particle motion trajectory, calculates the difference between the position and the velocity, superimposes the curl and shear data and normalizes them, screens the entropy value and constructs a dynamic feature matrix, extracts the local area variation characteristics, and establishes a local dynamic feature matrix;

[0050] The position-time matrix module extracts the time distribution data of water level changes based on the local dynamic feature matrix, calculates the regional dynamic position value and generates a position dynamic matrix, convolves the position matrix and the time parameter to generate a time matrix, superimposes the position matrix and the time matrix, and performs distribution processing to establish an overall dynamic feature matrix;

[0051] The dynamic feature fusion module extracts eddy distribution related parameters based on the local dynamic feature matrix and the overall dynamic feature matrix, assigns local weights and superimposes overall data, decomposes the matrix and fuses trend parameters, extracts dynamic distribution matching data, and establishes a water area dynamic feature data set;

[0052] The multi-scale law extraction module extracts dynamic data within the time window based on the water area dynamic characteristic data set, decomposes the multi-frequency data and calculates the amplitude and trend, extracts the amplitude law and classifies and clusters it, establishes a time scale matrix in combination with the trend distribution parameters, and generates a multi-scale dynamic law matrix;

[0053] The trend shift prediction module extracts the future change data of the water velocity field based on the multi-scale dynamic law matrix, intercepts the sliding window data and calculates the amplitude and trend recursive value, extracts the offset range and combines the trend matrix analysis to generate the dynamic trend shift prediction result.

[0054] Compared with the prior art, the advantages and positive effects of the present invention are:

[0055] In the present invention, the dynamic characteristics of the fluid particle motion trajectory and the water velocity field are extracted through differential calculation and normalization processing, and the local dynamic feature matrix is ​​generated by combining entropy value screening, thereby improving the accuracy of capturing local changes. The overall dynamic feature matrix is ​​generated based on the convolution operation of the time distribution and position matrix, which strengthens the ability to express the dynamic distribution and time association. The efficient fusion of the feature data set is achieved through the sensitivity weight allocation and distribution matching processing of the eddy current distribution, thereby enhancing the accuracy of trend analysis. Multi-frequency decomposition and time window are used to construct a regularity matrix to capture multi-scale dynamic laws, and the dynamic trend prediction is completed by combining the offset recursive extraction, thereby achieving a comprehensive analysis and high-precision prediction of water area changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0057] Figure 2 Flow chart of the steps of obtaining the local dynamic feature matrix of the present invention;

[0058] Figure 3 The flowchart of the steps of obtaining the overall dynamic feature matrix of the present invention;

[0059] Figure 4 A flowchart of the steps for obtaining a water area dynamic characteristic data set of the present invention;

[0060] Figure 5 It is a flow chart of the steps of obtaining the multi-scale dynamic law matrix of the present invention;

[0061] Figure 6 The present invention is a flowchart of the steps for obtaining the dynamic trend offset prediction results. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0063] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0064] Embodiment 1

[0065] See also Figure 1 The present invention provides a technical solution: a water conservancy project monitoring method based on water pattern analysis, comprising the following steps:

[0066] S1: Based on the time series of fluid particle motion trajectories, combined with the spatial variation of the water velocity field, through the differential calculation of position and velocity, the superposition operation of fusion curl and shear, all parameters are normalized and the entropy parameters are screened, and the local dynamic feature matrix is ​​generated through dynamic feature extraction;

[0067] S2: Based on the local dynamic feature matrix and the time distribution of water level changes, the position dynamic matrix is ​​formed by dependency value calculation, the position matrix is ​​convolved with the time-related parameters to form a time matrix, and the position and time matrices are superimposed and dynamically distributed to generate the overall dynamic feature matrix;

[0068] S3: Based on the local dynamic feature matrix and the overall dynamic feature matrix, combined with the associated parameters in the eddy distribution, the matrix is ​​normalized and decomposed item by item through the local feature sensitivity weight allocation and the overall associated parameter superposition processing, and the dynamic trend is integrated through distribution matching to generate the water area dynamic feature data set;

[0069] S4: Based on the water body dynamic characteristic data set, combined with the signal parameters of the water body energy field, the data is divided by time window and multi-frequency decomposition is performed to extract the amplitude and trend of changes within the time period, cluster the amplitude and trend, and construct a matrix through the time scale to generate a multi-scale dynamic law matrix;

[0070] S5: Based on the multi-scale dynamic law matrix, the future change data of the water velocity field is intercepted through a sliding window. The offset is differentially extracted by combining the recursive operation of the short-term amplitude and the long-term trend. The offset change range is combined with the trend matrix to generate the dynamic trend offset prediction result.

[0071] The local dynamic feature matrix includes position parameters, velocity parameters, curl parameters, and shear parameters. The position dynamic matrix includes position correlation values ​​and time correlation values. The overall dynamic feature matrix includes local feature sensitivity weights, overall correlation parameters, and dynamic distribution parameters. The water area dynamic feature data set specifically includes dynamic trend distribution parameters, eddy distribution parameters, and water body energy field signal parameters. The multi-scale dynamic law matrix includes time change amplitude, time change trend, and time scale matrix. The dynamic trend offset prediction results include short-term amplitude recursive results, long-term trend recursive results, and offset change range.

[0072] See also Figure 2 , the steps to obtain the local dynamic feature matrix are as follows:

[0073] S111: extracting the time series data of fluid particles, calculating the displacement difference of particles at continuous time points, and performing superposition operation in combination with the spatial change of the water flow velocity field to generate a dynamic change matrix of position and velocity;

[0074] This process is based on precise data collection methods to ensure that the differential calculation of displacement is accurate, and then combined with the spatial changes of the water velocity field to perform superposition operations. The key to this step is the accurate description of the velocity field and the high-precision measurement of spatial variables. The superposition operation emphasizes the real-time tracking and analysis of dynamic changes. This process requires not only high-frequency updates of data collection, but also efficient data processing capabilities to ensure feasibility and efficiency in practical applications. The generation of the dynamic change matrix will provide basic data support for subsequent analysis, ensuring that each data update can accurately reflect the actual dynamic state of the fluid particles.

[0075] S112: extracting the rotation and shear components from the position and velocity dynamic change matrices, calculating the combined superposition value of the rotation and shear components, and normalizing them to establish a superposition feature matrix;

[0076] The algorithm is used to accurately distinguish the physical meaning of each matrix element. The calculation of shear and curl must be based on the basic principles of fluid dynamics to ensure the scientificity and accuracy of the calculation results. When calculating the combined superposition value of curl and shear components, the viscosity characteristics of the fluid and the change in flow rate need to be considered. Normalization is to eliminate the influence of dimension and ensure the comparability of data under different conditions. The established superposition characteristic matrix should be able to reflect the dynamic characteristics of the fluid in detail and provide a basis for subsequent fluid analysis. The normalized data is more universal and comparable, which is conducive to direct application in different experiments and practical applications.

[0077] S113: Calculate the entropy value of the superimposed feature matrix, select key entropy value parameters, set dynamic threshold screening conditions, and use the formula:

[0078]

[0079] Generate a local dynamic feature matrix;

[0080] Among them, H represents the entropy value calculation result, V i represents the value of the i-th component in the superimposed feature matrix, ω i represents the curl of the ith component, represents the time derivative of the curl, α represents the adjustment factor used to adjust the influence of the time derivative of the curl, represents the sum of all component values ​​of the superimposed feature matrix, represents the normalized logarithm of the ith component, and n is the total number of features contained in the superimposed feature matrix, which determines the dimension of the data to be processed and the range of entropy calculation.

[0081] formula:

[0082]

[0083] The benefit of the formula is that by combining entropy calculation and curl change, the accuracy of capturing the dynamic characteristics of the fluid is improved, making the description of complex fluid motion more comprehensive and accurate.

[0084] Detailed explanation of the formula and the process of formula calculation and derivation:

[0085] Assume there are 3 data points, V = [2, 3, 5], n = 3, the curl change rate ω = [0.1, 0.2, 0.15], the time derivatives are [0.01, 0.02, 0.015], and α = 0.5. First calculate the overall flow Then calculate the entropy contribution of each term,

[0086] Item 1:

[0087] Item 2:

[0088] Item 3:

[0089] H=0.3268+0.3712+0.35325=1.05125;

[0090] The results show that considering the contribution of flow rate and curl change, the total entropy value H is 1.05125, which means that the system is highly disordered in the current state. The generation result of the corresponding step is the local dynamic characteristic matrix. This entropy value can be used to evaluate the complexity of fluid motion. Further deduction can obtain a detailed description and prediction of fluid particle motion.

[0091] See also Figure 3 , the steps to obtain the overall dynamic feature matrix are as follows:

[0092] S211: extracting parameter components in the local dynamic feature matrix, combining the time distribution of water level changes, calculating the dependency values ​​of multiple components, and forming a component dependency value matrix;

[0093] Time series analysis methods are used to capture the dynamic changes of fluid particles at different time points. By extracting the characteristic matrix of the local area, the position and velocity differences of the particles at continuous time points are calculated, and the differential results are combined with the water level change data to adjust the matrix parameters. This process involves detailed data analysis of the physical properties of the particles and their time distribution. During the calculation process, the velocity and position data of the particles are obtained through real-time monitoring, and the water level changes are calculated by combining previous data with current measurements. Through these detailed data analyses, more accurate dependent values ​​are obtained. These dependent values ​​can represent the motion state of the particles under different time and space conditions, and generate a component dependency value matrix with time sensitivity.

[0094] S212: calling the component dependency value matrix, performing differential calculation on the dependency value and the position parameter matrix, performing convolution calculation on the time correlation parameters in combination with the time distribution characteristics, and generating a time correlation matrix;

[0095] This calculation process uses two mathematical methods, namely, difference and convolution, to further refine the correlation between time and position. The calculation of time correlation parameters involves an in-depth analysis of time series data to ensure that each element in the matrix can accurately reflect the dynamic changes in time and space. Through the convolution operation, the influence of time distribution can be closely combined with position data to generate a time correlation matrix that can describe the dynamic behavior of the fluid. This matrix is ​​the basis for further calculations and provides detailed spatiotemporal background information for dynamic characteristics.

[0096] S213: Perform element-by-element superposition calculation on the time correlation matrix and the position matrix, using the formula:

[0097]

[0098] The dynamic distribution matrix of position and time is superimposed to generate the overall dynamic feature matrix;

[0099] Among them, P ij Represents the element value of the superimposed dynamic distribution matrix, V ij Represents the element value of the position matrix, T ij Represents the element value of the time-related matrix, S ij represents the local increment in the time correlation matrix, γ is the adjustment parameter, and β is the time distribution adjustment factor.

[0100] formula:

[0101]

[0102] The benefit of the formula is that it allows the element values ​​of the dynamic distribution matrix to be calculated by combining the changes in position and time, allowing the complex interactions in physical processes to be directly simulated mathematically, where the changes in time and position are obtained through actual measurements and prediction models. This calculation provides a method that can quantify complex fluid dynamics phenomena.

[0103] Detailed explanation of the formula and the process of formula calculation and derivation:

[0104] Set a specific value, V ij =2.0, T ij =3.0, S ij =1.5, γ=0.5, β=2.0, substitute into the formula for calculation:

[0105]

[0106] The results show that the calculated P ij The value of 5.57 reflects how the position and time factors under given parameters work together to affect the dynamic distribution of the fluid. This result further confirms the effectiveness and practicality of the formula used, which can actually reflect the dynamic characteristics of the fluid under specific conditions.

[0107] See also Figure 4 ,The specific steps for obtaining the water dynamic characteristic dataset are:

[0108] S311: using the data of the local dynamic characteristic matrix and the overall dynamic characteristic matrix, calculating the associated parameters in the eddy current distribution, taking the eddy current intensity and the flow direction as basic inputs for weight distribution, and generating an eddy current characteristic weighted matrix;

[0109] This step involves the quantification of vortex intensity and flow direction. The initial weights are set using existing fluid dynamics data and historical vortex observations. The weight distribution of each vortex feature is then calculated based on these data. This calculation involves the cross-analysis of multiple physical quantities, such as the speed, direction, and duration of the vortex. The size and shape changes of the vortex must also be considered during the calculation process, which requires dynamic adjustment of weights to reflect changes in real-time data. The weight distribution parameters are dynamically adjusted using data provided by the real-time monitoring system to ensure the accuracy and real-time nature of the weight distribution. This dynamic weight adjustment strategy can more accurately simulate the behavior of vortices under different conditions, ensuring the flexibility and adaptability of the model. The resulting vortex feature weighting matrix can be directly used for subsequent sensitivity analysis and feature synthesis.

[0110] S312: Combined with the eddy current feature weighted matrix, by performing sensitivity analysis on the local features, calling the weight value, and dynamically superimposing the overall correlation parameters, a comprehensive feature dynamic matrix is ​​generated;

[0111] This step requires determining which features are most sensitive to eddy dynamic changes based on the data in the weighted matrix. The data involved include, but are not limited to, the eddy rotation speed, vortex diameter, and eddy movement speed. The sensitivity analysis of these parameters will rely on specific experimental data and records. By comparing the eddy performance under different conditions, the sensitivity weights can be adjusted to ensure that the model can accurately reflect the eddy behavior in actual waters. This sensitivity analysis helps to optimize the model's predictive capabilities and enhance its accuracy in predicting future eddy behavior. After the sensitivity weights are adjusted, these weights are dynamically superimposed with the overall associated parameters to generate a comprehensive feature dynamic matrix.

[0112] S313: Normalize the comprehensive feature dynamic matrix item by item, using the formula:

[0113]

[0114] Perform feature decomposition, fuse dynamic trends through distribution matching algorithm, and generate water area dynamic feature data set;

[0115] Among them, M ij Represents the normalized matrix element value, W ij Represents the elements of the eddy current feature weighting matrix, G ij Represents the elements of the overall correlation parameter matrix, and n represents the number of columns in the matrix.

[0116] formula:

[0117]

[0118] The benefit of the formula is that it integrates local features and global dynamics in the form of exponential weights, which can effectively normalize the data while maintaining its characteristics, thus improving the ability to handle complex data.

[0119] Detailed explanation of the formula and the process of formula calculation and derivation:

[0120] Assume that the weighted characteristic W of a vortex is given ij = 0.2 and the overall correlation parameter G ij =0.8, for a simple 4-element matrix, the calculation is as follows:

[0121]

[0122] Calculating the terms in the denominator, we get:

[0123] e 0.2 ≈1.221,e 0.8 ≈2.225,e 1.0 ≈2.718,e 0.5 ≈1.649;

[0124]

[0125] The result is:

[0126]

[0127] The results show that the normalized matrix element values ​​reflect the combined influence of eddy characteristics and overall correlation parameters, have high response sensitivity, and are important indicators for measuring the dynamic characteristics of water areas.

[0128] See also Figure 5 , the specific steps for obtaining the multi-scale dynamic law matrix are:

[0129] S411: Based on the water body dynamic characteristic data set, call the water body energy field signal parameters, combine the signal amplitude and time window division, perform segmented calculation, and generate the energy field time distribution matrix;

[0130] Signal parameters are adjusted by analyzing dynamic changes in the water area, such as flow rate and temperature differences. The adjustment of these parameters is based on real-time data collection and historical trend analysis. In the example, the time window is set to 10 minutes, and the FFT algorithm is used to process the signal data to obtain preliminary frequency domain characteristics. The time distribution matrix is ​​then obtained by analyzing the change in signal amplitude. The segmented calculation calculates the total energy in each time window through the numerical integration method. In order to accurately adjust the signal parameters, the energy output of different time windows is compared, and parameters such as frequency and phase are dynamically adjusted to adapt to environmental changes. The parameters involved in this process include signal frequency, amplitude and their changing trends. These parameters are calculated based on real-time monitoring data through statistical methods such as moving average or exponential smoothing. The generated energy field time distribution matrix is ​​the result of dynamic adjustment based on continuous monitoring data and environmental feedback. This result can provide a detailed view of the energy distribution in the water area and provide a basis for subsequent energy management and monitoring.

[0131] S412: using the energy field time distribution matrix, performing multi-frequency decomposition operation, extracting the signal change amplitude and trend characteristics in multiple time periods, combining the local extreme value points of the signal amplitude to perform segmented clustering calculation, and generating an amplitude and trend clustering matrix;

[0132] First, the energy field time distribution matrix is ​​decomposed in the frequency domain through wavelet transform to extract the energy change information on different time scales. This step is achieved through the wavelet toolbox in MATLAB or Python. In the specific example, an appropriate mother wavelet such as Daubechies is selected for decomposition. According to the energy of each frequency band obtained, the short-term and long-term energy fluctuation characteristics are analyzed. Then, clustering calculations are performed in combination with the local extreme points of the signal amplitude. Different energy change patterns are identified using algorithms such as K-means clustering. The parameters include the values ​​of local extreme points and their positions in the time series. Through real-time analysis of these parameters, the energy change trends in the water body can be dynamically identified and classified. The generated amplitude and trend clustering matrix provides a quantitative view for further analysis and management of the dynamic changes of water body energy. This result shows the specific pattern of energy change in different time periods, providing important data support for energy management and optimization.

[0133] S413: The amplitude and trend clustering matrix is ​​multi-dimensionally expanded with the time scale parameter using the formula:

[0134]

[0135] The time scale characteristic matrix is ​​established by fusing the amplitude and trend matrix, and then normalized to generate a multi-scale dynamic law matrix;

[0136] Among them, D ij Multi-scale dynamic law matrix, Ti represents the time scale factor, F j Represents the signal frequency factor, A ijk is the amplitude parameter of the clustering matrix, B ijk is the trend parameter of the clustering matrix, and n is the number of features in the time period.

[0137] formula:

[0138]

[0139] The formula is useful in that it incorporates the time scale factor T i and frequency factor F j , taking into account the importance and contribution of signals at different times and frequencies, thus more accurately reflecting the multi-scale characteristics of water body dynamics.

[0140] Detailed explanation of the formula and the process of formula calculation and derivation:

[0141] Assume that at a certain time scale T i = 0.8 and frequency F j = 0.5, the parameter A of the clustering matrix ijk and B ijk 2 and 0.1 respectively. Substituting these specific values ​​into the formula for calculation, first calculate the square root of the frequency-time product and get:

[0142]

[0143] Next, we calculate the logarithmic term:

[0144] ln(1+0.1)=0.0953;

[0145] Finally, substituting these values ​​into the formula, we get:

[0146]

[0147] The results show that for this specific time scale and frequency, the system's dynamic law strength is 0.3013, a value that can help understand the main characteristics of the water body's dynamic behavior under these conditions and how to use these characteristics to predict future dynamic changes.

[0148] See also Figure 6 , the specific steps for obtaining the dynamic trend shift prediction results are:

[0149] S511: Based on the multi-scale dynamic law matrix, the sliding window technology is used to intercept the future change data of the water flow velocity field in the short term to obtain the original input data of the future trend;

[0150] This process first involves obtaining relevant water flow velocity field data from the water area dynamic characteristic data set, and then filtering out the corresponding data set by setting the time window. In this process, it is necessary to ensure the integrity and representativeness of the data to ensure the accuracy of trend prediction. Next, use data preprocessing technology to prepare the data, which includes denoising, standardization, and determining the time series characteristics of the data. These steps are all aimed at improving the effectiveness and reliability of the data in subsequent analysis. Through these detailed data processing steps, it can be ensured that the original input data obtained contains both the necessary time markers and retains the key changes in water flow velocity, which lays a solid foundation for the next step of trend analysis.

[0151] S512: performing recursive operations on the short-term amplitude and long-term trend of the original input data of the future trend, calculating the amplitude and trend in multiple time windows in the future, and generating trend analysis results;

[0152] This processing step is mainly achieved through mathematical models, which involve complex statistical analysis and time series analysis methods, such as autoregressive models and moving average models. These models can help identify and extract patterns and trends in time series data, which is particularly critical for predicting future changes in water flow velocity fields. By using these models, short-term and long-term change trends can be accurately depicted numerically, which not only improves the accuracy of the prediction, but also provides a scientific basis for subsequent data analysis and decision-making. The trend analysis results obtained will directly affect the understanding and prediction of future water flow behavior.

[0153] S513: Perform differential extraction on the trend analysis results, analyze the offset change, and combine the multi-scale dynamic law matrix using the formula:

[0154]

[0155] Integrate the change amplitude and trend of time series to generate dynamic trend shift prediction results;

[0156] Among them, ΔP t represents the offset prediction result, S t represents the trend data extracted from the multi-scale dynamic law matrix, α represents the weight factor of short-term changes, β represents the integral coefficient of long-term trends, represents the time derivative of trend data, It represents the integral of trend data within the time window T, reflecting the dynamic changes on the time scale.

[0157] formula:

[0158]

[0159] The benefit of the formula is that it can combine short-term velocity changes and long-term cumulative trends. By properly adjusting the values ​​of α and β, the sensitivity of the model to future events can be fine-tuned, thereby optimizing the accuracy of the forecast results.

[0160] Detailed explanation of the formula and the process of formula calculation and derivation:

[0161] First, we need to extract trend data S from the multi-scale dynamic law matrix t For example, in the past week, the average water velocity may be 0.5m / s, and its rate of change By calculating the speed change over two consecutive days, assuming that the speed is 0.45 m / s on the previous day and 0.55 m / s on the next day, then for

[0162] The calculation of the long-term cumulative trend involves integration. If the time window is set to one week, and the daily speed value increases continuously from 0.45m / s to 0.55m / s, the integral result is It can be obtained by numerical integration method, for example, using the trapezoidal rule, and calculated to be ≈3.5m / s. If α=0.7 and β=0.3, then ΔP t =0.7×0.1+0.3×3.5=1.15m / s.

[0163] The results show that the predicted offset is the increase in water velocity over the next week, that is, the water velocity will increase by 1.15m / s, which is helpful for predicting and preparing for possible high flow events.

[0164] A water conservancy project monitoring system based on water ripple analysis, the water conservancy project monitoring system based on water ripple analysis is used to execute the water conservancy project monitoring method based on water ripple analysis, the system comprises:

[0165] The local feature processing module extracts the spatial variation data of the water velocity field based on the time series of the fluid particle motion trajectory, calculates the difference between the position and the velocity, superimposes the curl and shear data and normalizes them, screens the entropy value and constructs a dynamic feature matrix, extracts the local area variation characteristics, and establishes a local dynamic feature matrix;

[0166] The position-time matrix module extracts the time distribution data of water level changes based on the local dynamic feature matrix, calculates the regional dynamic position value and generates the position dynamic matrix, convolves the position matrix and time parameters to generate the time matrix, superimposes the position matrix and the time matrix, and performs distribution processing to establish the overall dynamic feature matrix;

[0167] The dynamic feature fusion module extracts eddy distribution correlation parameters based on the local dynamic feature matrix and the overall dynamic feature matrix, assigns local weights and superimposes overall data, decomposes the matrix and fuses trend parameters, extracts dynamic distribution matching data, and establishes a water area dynamic feature data set;

[0168] The multi-scale law extraction module extracts dynamic data within the time window based on the water dynamic characteristic data set, decomposes the multi-frequency data and calculates the amplitude and trend, extracts the amplitude law and classifies and clusters it, establishes the time scale matrix in combination with the trend distribution parameters, and generates a multi-scale dynamic law matrix;

[0169] The trend shift prediction module is based on a multi-scale dynamic law matrix, extracts future change data of the water velocity field, intercepts sliding window data and calculates the amplitude and trend recursive value, extracts the offset range and combines it with trend matrix analysis to generate dynamic trend shift prediction results.

[0170] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A water conservancy project monitoring method based on water pattern analysis, characterized in that: The following steps are involved: S1: Based on the time series of fluid particle motion trajectories, combined with the spatial variation of the water velocity field, through the differential calculation of position and velocity, the superposition operation of fusion curl and shear, all parameters are normalized and the entropy parameters are screened, and the local dynamic feature matrix is ​​generated through dynamic feature extraction; S2: Based on the local dynamic feature matrix and in combination with the time distribution of water level changes, a position dynamic matrix is ​​formed by dependency value calculation, the position matrix is ​​convolved with the time-related parameter to form a time matrix, and dynamic distribution processing is performed after superimposing the position and time matrices to generate an overall dynamic feature matrix; S3: Based on the local dynamic feature matrix and the overall dynamic feature matrix, combined with the associated parameters in the eddy distribution, the matrix is ​​normalized and decomposed item by item through the local feature sensitivity weight allocation and the overall associated parameter superposition processing, and the dynamic trend is integrated through distribution matching to generate a water area dynamic feature data set; S4: Based on the water area dynamic characteristic data set, combined with the signal parameters of the water body energy field, the data is divided by time window and multi-frequency decomposition is performed, the amplitude and trend of the change within the time period are extracted, the amplitude and trend are clustered and classified, and a matrix is ​​constructed by time scale to generate a multi-scale dynamic law matrix; S5: Based on the multi-scale dynamic law matrix, the future change data of the water velocity field is intercepted through a sliding window, and the offset is differentially extracted by combining the recursive operation of the short-term amplitude and the long-term trend. The offset change range is combined with the trend matrix to generate a dynamic trend offset prediction result.

2. The water conservancy project monitoring method based on water pattern analysis according to claim 1 is characterized in that: The local dynamic feature matrix includes position parameters, velocity parameters, curl parameters, and shear parameters. The position dynamic matrix includes position correlation values ​​and time correlation values. The overall dynamic feature matrix includes local feature sensitivity weights, overall correlation parameters, and dynamic distribution parameters. The water area dynamic feature data set specifically includes dynamic trend distribution parameters, eddy distribution parameters, and water body energy field signal parameters. The multi-scale dynamic law matrix includes time change amplitude, time change trend, and time scale matrix. The dynamic trend offset prediction results include short-term amplitude recursive results, long-term trend recursive results, and offset change range.

3. The water conservancy project monitoring method based on water pattern analysis according to claim 2 is characterized in that: The steps of obtaining the local dynamic feature matrix are specifically as follows: S111: extracting the time series data of fluid particles, calculating the displacement difference of particles at continuous time points, combining the spatial change of the water flow velocity field with superposition operation, and generating a dynamic change matrix of position and velocity; S112: extracting the rotation and shear components from the position and velocity dynamic change matrix, calculating the combined superposition value of the rotation and shear components, and normalizing them to establish a superposition feature matrix; S113: Calculate the entropy value of the superimposed feature matrix, select key entropy value parameters, set dynamic threshold screening conditions, and use the formula: Generate a local dynamic feature matrix; Among them, H represents the entropy value calculation result, V i represents the value of the i-th component in the superimposed feature matrix, ω i represents the curl of the ith component, represents the time derivative of the curl, α represents the adjustment factor used to adjust the influence of the time derivative of the curl, represents the sum of all component values ​​of the superimposed feature matrix, represents the normalized logarithm of the ith component, and n is the total number of features contained in the superimposed feature matrix, which determines the dimension of the data to be processed and the range of entropy calculation.

4. The water conservancy project monitoring method based on water pattern analysis according to claim 3 is characterized in that: The steps for obtaining the overall dynamic feature matrix are specifically as follows: S211: extracting parameter components from the local dynamic feature matrix, calculating dependency values ​​of multiple components in combination with the time distribution of water level changes, and forming a component dependency value matrix; S212: calling the component dependency value matrix, performing differential calculation on the dependency value and the position parameter matrix, performing convolution calculation on the time correlation parameters in combination with the time distribution characteristics, and generating a time correlation matrix; S213: Perform element-by-element superposition calculation on the time correlation matrix and the position matrix, using the formula: The dynamic distribution matrix of position and time is superimposed to generate the overall dynamic feature matrix; Among them, P ij Represents the element value of the dynamic distribution matrix after superposition, V ij Represents the element value of the position matrix, T ij Represents the element value of the time-related matrix, S ij represents the local increment in the time correlation matrix, γ is the adjustment parameter, and β is the time distribution adjustment factor.

5. The water conservancy project monitoring method based on water pattern analysis according to claim 4 is characterized in that: The steps for obtaining the water area dynamic characteristic data set are specifically as follows: S311: using the data of the local dynamic characteristic matrix and the overall dynamic characteristic matrix, calculating the associated parameters in the eddy current distribution, taking the eddy current intensity and the flow direction as basic inputs for weight distribution, and generating an eddy current characteristic weighted matrix; S312: combining the eddy current feature weighting matrix, performing sensitivity analysis on local features, calling weight values, and dynamically superimposing them with overall correlation parameters to generate a comprehensive feature dynamic matrix; S313: normalizing the comprehensive feature dynamic matrix item by item, using the formula: Perform feature decomposition, fuse dynamic trends through distribution matching algorithm, and generate water area dynamic feature data set; Among them, M ij Represents the normalized matrix element value, W ij Represents the elements of the eddy current feature weighting matrix, G ij Represents the elements of the overall correlation parameter matrix, and n represents the number of columns in the matrix.

6. The water conservancy project monitoring method based on water pattern analysis according to claim 5 is characterized in that: The steps for obtaining the multi-scale dynamic law matrix are specifically as follows: S411: Based on the water body dynamic characteristic data set, call the water body energy field signal parameters, combine the signal amplitude and time window division, perform segmented calculation, and generate an energy field time distribution matrix; S412: using the energy field time distribution matrix, performing a multi-frequency decomposition operation, extracting the signal change amplitude and trend characteristics in multiple time periods, performing segmented clustering calculations in combination with the local extreme value points of the signal amplitude, and generating an amplitude and trend clustering matrix; S413: The amplitude and trend clustering matrix is ​​multi-dimensionally expanded in combination with the time scale parameter, using the formula: The time scale characteristic matrix is ​​established by fusing the amplitude and trend matrix, and then normalized to generate a multi-scale dynamic law matrix; Among them, D ij Multi-scale dynamic law matrix, T i represents the time scale factor, F j Represents the signal frequency factor, A ijk is the amplitude parameter of the clustering matrix, B ijk is the trend parameter of the clustering matrix, and n is the number of features in the time period.

7. The water conservancy project monitoring method based on water pattern analysis according to claim 6 is characterized in that: The steps for obtaining the dynamic trend shift prediction result are specifically as follows: S511: Based on the multi-scale dynamic law matrix, a sliding window technology is used to intercept the future change data of the water flow velocity field in a short period of time to obtain the original input data of the future trend; S512: performing recursive operations on the short-term amplitude and the long-term trend of the original input data of the future trend, calculating the amplitude and the trend in multiple time windows in the future, and generating trend analysis results; S513: Perform differential extraction on the trend analysis results, analyze the offset change, and combine the multi-scale dynamic law matrix to use the formula: Integrate the change amplitude and trend of time series to generate dynamic trend shift prediction results; Among them, ΔP t represents the offset prediction result, S t represents the trend data extracted from the multi-scale dynamic law matrix, α represents the weight factor of short-term changes, β represents the integral coefficient of long-term trends, represents the time derivative of trend data, It represents the integral of trend data within the time window T, reflecting the dynamic changes on the time scale.

8. A water conservancy project monitoring system based on water pattern analysis, characterized in that: According to the water conservancy project monitoring method based on water pattern analysis according to any one of claims 1 to 7, the system comprises: The local feature processing module extracts the spatial variation data of the water velocity field based on the time series of the fluid particle motion trajectory, calculates the difference between the position and the velocity, superimposes the curl and shear data and normalizes them, screens the entropy value and constructs a dynamic feature matrix, extracts the local area variation characteristics, and establishes a local dynamic feature matrix; The position-time matrix module extracts the time distribution data of water level changes based on the local dynamic feature matrix, calculates the regional dynamic position value and generates a position dynamic matrix, convolves the position matrix and the time parameter to generate a time matrix, superimposes the position matrix and the time matrix, and performs distribution processing to establish an overall dynamic feature matrix; The dynamic feature fusion module extracts eddy distribution related parameters based on the local dynamic feature matrix and the overall dynamic feature matrix, assigns local weights and superimposes overall data, decomposes the matrix and fuses trend parameters, extracts dynamic distribution matching data, and establishes a water area dynamic feature data set; The multi-scale law extraction module extracts dynamic data within the time window based on the water area dynamic characteristic data set, decomposes the multi-frequency data and calculates the amplitude and trend, extracts the amplitude law and classifies and clusters it, establishes a time scale matrix in combination with the trend distribution parameters, and generates a multi-scale dynamic law matrix; The trend shift prediction module extracts the future change data of the water flow velocity field based on the multi-scale dynamic law matrix, intercepts the sliding window data and calculates the amplitude and trend recursive value, extracts the offset range and combines the trend matrix analysis to generate the dynamic trend shift prediction result.

Images