A water conservancy project management method and system based on digital twin

Abstract

The present application relates to a water conservancy project management method and system based on digital twins, and to the technical field of water conservancy project management. The method includes: extracting time series features and performing cross-domain dynamic collaborative correlation analysis on the water discharge data by using data processing and analysis algorithms based on artificial intelligence and deep learning at the back end, thereby capturing the impact intensity time series semantics based on the water flow velocity time series features and the water flow density time series features, and automatically obtaining the damage score by cross-domain feature fusion query features between the impact intensity time series semantics and the water discharge time series semantics. In this way, it is possible to adaptively generate damage scores that conform to actual conditions according to the time series dynamic changes and collaborative correlations between water flow velocity, water flow density and water discharge by using artificial intelligence technology and deep learning algorithms, thereby improving the intelligence level and generalization ability of the water conservancy project management method.

Description

Technical Field

[0001] The present application relates to the technical field of water conservancy project management, and specifically, to a water conservancy project management method and system based on digital twins. Background Art

[0002] Reservoirs are one of the engineering measures used for flood control. Usually, a comprehensive utilization reservoir that can regulate and store flood water is built on the upstream river channel of the flood control area. The reservoir storage capacity is used to intercept and store flood water, and the peak flow entering the downstream river channel is reduced to achieve the purpose of reducing flood disasters. The outlets in water conservancy projects are prone to wear or damage due to long-term impact of water flow, which has a direct impact on the safety and service life of water conservancy facilities. However, the existing technology lacks safety monitoring of water outlets, especially monitoring of damage caused by water flow impact, making it difficult to timely detect and locate damaged outlets, thereby posing a safety hazard to reservoir operation.

[0003] In response to the above technical problems, Chinese patent CN116911692A discloses a water conservancy project management method and system based on digital twins, which can calculate the water outlet data of the target outlet based on the water inlet and outlet model, and then calculate the damage score of the target outlet based on the water outlet data, and finally generate an evaluation result based on the damage score. Among them, the calculation of the damage score involves factors such as water flow velocity, water flow density, and the direction of water flow, and the damage degree is quantified by the formula S=Fcosq, so as to monitor and evaluate the damage of the outlet in the water conservancy project.

[0004] However, in the above-mentioned water conservancy project management method, the damage score is mainly evaluated by relying on static physical parameters such as water velocity, water flow density and water output, and there is a lack of understanding of the time series analysis and dynamic interaction of these data. In other words, since the water velocity, water flow density and the flow direction of the water output are all data that change continuously in time series, the solution in the above-mentioned patent only uses a static formula to calculate the damage score, and does not pay attention to the dynamic changes and synergy between these data in time series. In this way, it is impossible to adapt to the complex and changeable changes in the outlet data in actual situations, resulting in low accuracy in damage monitoring and positioning.

[0005] Therefore, an optimized digital twin-based water conservancy project management solution is expected. Summary of the invention

[0006] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0007] In a first aspect, the present application provides a water conservancy project management method based on digital twins, the method comprising:

[0008] Calculate the water outlet data of the target water outlet based on the water inlet and the water outlet model, wherein the water outlet model includes a plurality of water outlets, and the target water outlet is at least one water outlet in the water outlet model;

[0009] Calculating the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density, and a time queue of water outlet volume;

[0010] generating an assessment result based on the damage score;

[0011] Wherein, calculating the damage score of the target water outlet based on the water outlet data includes:

[0012] Performing water discharge sequence encoding on the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water discharge volume respectively to obtain a water flow velocity time series associated feature vector, a water flow density time series associated feature vector and a water discharge volume time series associated feature vector;

[0013] Inputting the water velocity time series associated feature vector and the water flow density time series associated feature vector into a feature local-global interactive response module based on feature decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix;

[0014] Taking the water discharge time series correlation feature vector as the query feature vector and the impact strength local-global multi-scale interactive response representation matrix as the initial key matrix, the query feature vector and the initial key matrix are input into a cross-domain feature fusion query encoding module with key matrix local correlation enhancement to obtain a damage multi-modal time series correlation representation vector;

[0015] The impairment multi-modal temporal correlation representation vector is input into a decoder-based impairment assessor to obtain a decoded value of the impairment score.

[0016] Optionally, the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water output are respectively encoded with water output sequence to obtain the water flow velocity time series correlation feature vector, the water flow density time series correlation feature vector and the water output time series correlation feature vector, including: inputting the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water output into a water output sequence encoder based on a Bi-LSTM model respectively to obtain the water flow velocity time series correlation feature vector, the water flow density time series correlation feature vector and the water output time series correlation feature vector.

[0017] Optionally, the water flow velocity time series correlation feature vector and the water flow density time series correlation feature vector are input into a feature local-global interactive response module based on feature decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix, including: performing feature decoupling on the water flow density time series correlation feature vector to obtain a sequence of water flow density local time series correlation feature vectors; performing nonlinear transformation on the water flow velocity time series correlation feature vector to obtain a nonlinearly transformed water flow velocity time series correlation feature vector; calculating a correlation matrix between the nonlinearly transformed water flow velocity time series correlation feature vector and each water flow density local time series correlation feature vector in the sequence of the water flow density local time series correlation feature vector to obtain a water flow density-water flow velocity time series correlation feature matrix. The invention relates to a method for preparing a sequence of normalized water flow density-water flow velocity time series interaction correlation matrices; calculating the characteristic interaction energy factors of each water flow density-water flow velocity time series interaction correlation matrix in the sequence of water flow density-water flow velocity time series interaction correlation matrix to obtain a sequence of water flow density-water flow velocity time series characteristic interaction energy factors; inputting the sequence of water flow density-water flow velocity time series characteristic interaction energy factors into a Sigmoid function to obtain a sequence of normalized water flow density-water flow velocity time series characteristic interaction energy factors; using the sequence of normalized water flow density-water flow velocity time series characteristic interaction energy factors as a sequence of weights, calculating the position-weighted sum of the sequence of water flow density-water flow velocity time series interaction correlation matrices to obtain the local-global multi-scale interaction response representation matrix of the impact intensity.

[0018] Optionally, a nonlinear transformation is performed on the water flow velocity time series associated characteristic vector to obtain the water flow velocity time series associated characteristic vector after nonlinear transformation, including: calculating the multiplication of the water flow velocity time series associated characteristic vector and the nonlinear transformation weight matrix, and then performing multiplication with the nonlinear transformation bias vector according to the position point to obtain the water flow velocity time series associated characteristic vector after nonlinear transformation.

[0019] Optionally, the characteristic interaction energy factors of each water flow density-water flow velocity time series interaction correlation matrix in the sequence of the water flow density-water flow velocity time series interaction correlation matrix are calculated to obtain a sequence of water flow density-water flow velocity time series characteristic interaction energy factors, including: calculating the kurtosis of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction kurtosis value; calculating the global characteristic mean and variance of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction characteristic mean and the water flow density-water flow velocity time series interaction variance; calculating the water flow density-water flow velocity time series interaction variance and The sum of the predetermined hyperparameters is used to obtain the water flow density-water flow velocity time series interaction variance modulation value; after calculating the square of the difference between the water flow density-water flow velocity time series interaction kurtosis value and the water flow density-water flow velocity time series interaction feature mean, the obtained value is added with twice the water flow density-water flow velocity time series interaction variance and twice the predetermined hyperparameter to obtain the water flow density-water flow velocity time series interaction statistical feature sum value; the division between the water flow density-water flow velocity time series interaction variance modulation value and the water flow density-water flow velocity time series interaction statistical feature sum value is calculated to obtain the water flow density-water flow velocity time series feature interaction energy factor.

[0020] Optionally, the kurtosis of the water flow density-water flow velocity time series interaction correlation matrix is ​​calculated to obtain the water flow density-water flow velocity time series interaction kurtosis value, including: calculating the fourth power of the position difference between the water flow density-water flow velocity time series interaction correlation matrix and the water flow density-water flow velocity time series interaction feature mean, and then calculating the expectation of the obtained feature matrix to obtain the water flow density-water flow velocity time series interaction correlation expectation value; calculating the division between the water flow density-water flow velocity time series interaction correlation expectation value and the square of the water flow density-water flow velocity time series interaction variance to obtain the water flow density-water flow velocity time series interaction kurtosis value.

[0021] Optionally, the water yield time series correlation feature vector is used as a query feature vector and the impact strength local-global multi-scale interactive response representation matrix is ​​used as an initial key matrix, and the query feature vector and the initial key matrix are input into a cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix to obtain a damage multi-modal time series correlation representation vector, including: calculating the feature distribution difference energy coefficient between the water yield time series correlation feature vector and each row vector in the initial key matrix to obtain a set of impact strength local-global multi-scale interactive response feature distribution difference energy coefficients; comparing each impact strength local-global multi-scale interactive response feature distribution difference energy coefficient in the set of impact strength local-global multi-scale interactive response feature distribution difference energy coefficients with a predetermined threshold, and in response to the impact strength local-global multi-scale interactive response feature distribution difference energy coefficient being less than the predetermined threshold, using the impact strength local-global multi-scale interactive response feature distribution difference energy coefficient as the query feature vector and the initial key matrix to obtain a damage multi-modal time series correlation representation vector. The mean vector between the row vector in the initial key matrix corresponding to the interactive response feature distribution difference energy coefficient and the water output time series associated feature vector is used as the updated row vector to obtain the updated key matrix; the matrix multiplication between the water output time series associated feature vector and the value embedding matrix is ​​calculated, and then the water output time series associated value vector is added positionally; the water output time series associated feature vector is used as the query vector, each updated row vector in the updated key matrix is ​​used as the key vector, and the water output time series associated value vector is used as the value vector, and the query vector, the key vector and the value vector are input into the cross-domain attention gate query module based on the Transformer structure to obtain a sequence of impact strength local-global multi-scale interactive response cross-domain query attention vectors; the positional mean vector of the sequence of impact strength local-global multi-scale interactive response cross-domain query attention vectors is calculated to obtain the damage multimodal time series associated representation vector.

[0022] Optionally, the characteristic distribution difference energy coefficient between the water yield time series correlation feature vector and each row vector in the initial key matrix is ​​calculated to obtain a set of characteristic distribution difference energy coefficients of the impact strength local-global multi-scale interactive response, including: respectively calculating the position difference, position point multiplication and position addition between the water yield time series correlation feature vector and the row vector in the initial key matrix to obtain the water yield time series correlation-impact strength local-global multi-scale interactive response difference feature vector, the water yield time series correlation-impact strength local-global multi-scale interactive response point multiplication feature vector and Water yield time series correlation-impact intensity local-global multi-scale interactive response sum feature vector; the water yield time series correlation-impact intensity local-global multi-scale interactive response differential feature vector, the water yield time series correlation-impact intensity local-global multi-scale interactive response dot product feature vector and the water yield time series correlation-impact intensity local-global multi-scale interactive response sum feature vector are cascaded and processed through a one-dimensional convolution layer, and the obtained feature vector is subjected to maximum pooling processing to obtain a water yield time series correlation-impact intensity local-global multi-scale interactive response feature correlation representation vector; extract the maximum value of the feature in the water yield time series association-impact intensity local-global multi-scale interactive response feature association representation vector to obtain the water yield time series association-impact intensity local-global multi-scale interactive response association maximum representation value; calculate the mean and variance of the row vector in the initial key matrix to obtain the impact intensity local-global multi-scale interactive response feature mean and the impact intensity local-global multi-scale interactive response feature variance; add the impact intensity local-global multi-scale interactive response feature variance to the hyperparameter to obtain the first feature distribution difference energy factor; calculate the square of the difference between the water yield time series association-impact intensity local-global multi-scale interactive response maximum representation value and the impact intensity local-global multi-scale interactive response feature mean to obtain the semantic feature difference value; multiply the impact intensity local-global multi-scale interactive response feature variance by a constant of two and then add it to the semantic feature difference value and the hyperparameter to obtain the second feature distribution difference energy factor; divide the first feature distribution difference energy factor by the second feature distribution difference energy factor to obtain the impact intensity local-global multi-scale interactive response feature distribution difference energy coefficient.

[0023] Optionally, the water output time series associated feature vector is used as a query vector, each updated row vector in the update key matrix is ​​used as a key vector, and the water output time series associated value vector is used as a value vector, and the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of impact intensity local-global multi-scale interactive response cross-domain query attention vectors, including: after calculating the product between the query vector and the transposed vector of the key vector, the obtained query-key feature association value is divided by the square root of the length of the key vector to obtain a query-key feature association coefficient; the query-key feature association coefficient is input into a softmax function to obtain a query-key feature association weight value; the query-key feature association weight value is used as a weight to weight the transposed vector of the value vector to obtain an impact intensity local-global multi-scale interactive response cross-domain query attention vector.

[0024] In a second aspect, the present application provides a water conservancy project management system based on digital twins, the system comprising:

[0025] A water outlet data calculation module for a target water outlet, used to calculate the water outlet data of the target water outlet based on the water inlet and the water outlet model, wherein the water outlet model includes a plurality of water outlets, and the target water outlet is at least one water outlet in the water outlet model;

[0026] a damage score calculation module for a target water outlet, used to calculate the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density, and a time queue of water output;

[0027] An assessment result generating module, used for generating an assessment result based on the damage score;

[0028] Wherein, the damage score calculation module of the target water outlet includes:

[0029] A water discharge sequence encoding unit, used for performing water discharge sequence encoding on the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water discharge volume respectively to obtain a water flow velocity time series correlation feature vector, a water flow density time series correlation feature vector and a water discharge volume time series correlation feature vector;

[0030] A characteristic local-global interactive response unit, used for inputting the water velocity time series associated characteristic vector and the water flow density time series associated characteristic vector into a characteristic local-global interactive response module based on characteristic decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix;

[0031] A cross-domain feature fusion query encoding unit is used to use the water discharge time series correlation feature vector as a query feature vector and the impact strength local-global multi-scale interactive response representation matrix as an initial key matrix, and input the query feature vector and the initial key matrix into a cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix to obtain a damage multi-modal time series correlation representation vector;

[0032] The decoding unit is used to input the impairment multimodal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of the impairment score.

[0033] By adopting the above technical solution, the data processing and analysis algorithm based on artificial intelligence and deep learning is used in the back end to extract the time series features and perform cross-domain dynamic collaborative correlation analysis on the water discharge data, so as to capture the time series semantics of impact intensity based on the time series features of water flow velocity and water flow density, and automatically obtain the damage score by cross-domain feature fusion query features between the time series semantics of impact intensity and the time series semantics of water discharge. In this way, it is possible to adaptively generate damage scores that conform to the actual situation according to the time series dynamic changes and collaborative correlation between water flow velocity, water flow density and water discharge by using artificial intelligence technology and deep learning algorithms, thereby improving the intelligence level and generalization ability of water conservancy project management methods.

[0034] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings:

[0036] Figure 1 It is a flow chart of a water conservancy project management method based on digital twin according to an exemplary embodiment.

[0037] Figure 2 It is a block diagram of a water conservancy project management system based on digital twin according to an exemplary embodiment.

[0038] Figure 3 It is a block diagram of an electronic device according to an exemplary embodiment.

[0039] Figure 4 It is an application scenario diagram of a water conservancy project management method based on digital twins according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0041] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0042] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0043] It should be noted that the concepts such as "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0044] It should be noted that the modifications of "one" and "plurality" mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0045] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0046] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

[0047] Figure 1 is a flow chart of a water conservancy project management method based on digital twins according to an exemplary embodiment. Figure 1 As shown, the method includes:

[0048] Step S101, calculating the water outlet data of the target water outlet based on the water inlet and the water outlet model, wherein the water outlet model includes a plurality of water outlets, and the target water outlet is at least one water outlet in the water outlet model;

[0049] Step S102, calculating the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density, and a time queue of water outlet volume;

[0050] Step S103, generating an assessment result based on the damage score;

[0051] Wherein, step S102, calculating the damage score of the target water outlet based on the water outlet data, includes:

[0052] Step S1021, respectively perform water outflow sequence encoding on the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water outflow to obtain a water flow velocity time series correlation feature vector, a water flow density time series correlation feature vector and a water outflow time series correlation feature vector;

[0053] Step S1022, inputting the water velocity time series associated feature vector and the water flow density time series associated feature vector into a feature local-global interactive response module based on feature decoupling to obtain an impact intensity local-global multi-scale interactive response representation matrix;

[0054] Step S1023: using the water output time series correlation feature vector as a query feature vector and the impact strength local-global multi-scale interactive response representation matrix as an initial key matrix, inputting the query feature vector and the initial key matrix into a cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix to obtain a damage multi-modal time series correlation representation vector;

[0055] Step S1024: input the impairment multimodal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of the impairment score.

[0056] In response to the above technical problems, in the technical solution of the present application, a water conservancy project management method based on digital twins is proposed, which includes: calculating the water outlet data of the target water outlet based on the water inlet and outlet model, the water outlet model includes multiple water outlets, and the target water outlet is at least one water outlet in the water outlet model; calculating the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density and a time queue of water outlet; generating an evaluation result based on the damage score.

[0057] In particular, in the process of calculating the damage score of the target water outlet based on the water outlet data, since the collected water outlet data includes the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water output, the technical concept of the present application is to extract the time series features and perform cross-domain dynamic collaborative association analysis on the water outlet data by using the data processing and analysis algorithm based on artificial intelligence and deep learning at the back end, so as to capture the impact intensity time series semantics based on the water flow velocity time series features and the water flow density time series features, and automatically obtain the damage score by cross-domain feature fusion query features between the impact intensity time series semantics and the water output time series semantics. In this way, it is possible to adaptively generate damage scores that conform to the actual situation according to the time series dynamic changes and collaborative associations between the water flow velocity, water flow density and water output by using artificial intelligence technology and deep learning algorithms, thereby improving the intelligence level and generalization ability of the water conservancy project management method.

[0058] Specifically, in the technical solution of the present application, first, a time queue of water flow velocity, a time queue of water flow density and a time queue of water output are obtained, and the time queue of water flow velocity, the time queue of water flow density and the time queue of water output are respectively input into a water output sequence encoder based on a Bi-LSTM model for feature extraction, so as to respectively extract the temporal forward and backward long-distance dependency correlation feature information of the water flow velocity, the water flow density and the water output in the time dimension, thereby obtaining a water flow velocity time series correlation feature vector, a water flow density time series correlation feature vector and a water output time series correlation feature vector.

[0059] In one embodiment of the present application, the time queue of the water flow velocity, the time queue of the water flow density, and the time queue of the water output are respectively encoded with water output sequences to obtain the water flow velocity time series correlation feature vector, the water flow density time series correlation feature vector, and the water output time series correlation feature vector, including: inputting the time queue of the water flow velocity, the time queue of the water flow density, and the time queue of the water output into a water output sequence encoder based on a Bi-LSTM model respectively to obtain the water flow velocity time series correlation feature vector, the water flow density time series correlation feature vector, and the water output time series correlation feature vector.

[0060] Then, considering that the water velocity and water flow density not only have their own time series dynamic change characteristics in the time dimension, but also there is a time series interaction and correlation relationship between the two, this correlation interaction relationship can well reflect the impact intensity of the water flow on the outlet. Based on this, in order to better understand the complex relationship between the water flow velocity and the water flow density, so as to more accurately evaluate the impact intensity of the water flow on the outlet, in the technical solution of the present application, the water flow velocity time series correlation feature vector and the water flow density time series correlation feature vector are further input into the feature local-global interactive response module based on feature decoupling to obtain the impact intensity local-global multi-scale interactive response representation matrix. In particular, the feature local-global interactive response module based on feature decoupling includes processing steps such as feature decoupling, feature multi-scale interaction and feature interaction metric learning, which is used to deeply analyze and integrate the information in the feature vectors of different modal data features. This method not only considers the importance of local features between feature vectors, but also integrates the information of all local features from a global perspective, thereby obtaining a more comprehensive feature representation of impact intensity.

[0061] In one embodiment of the present application, the water flow velocity time series correlation feature vector and the water flow density time series correlation feature vector are input into a feature local-global interactive response module based on feature decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix, including: feature decoupling the water flow density time series correlation feature vector to obtain a sequence of water flow density local time series correlation feature vectors; nonlinearly transforming the water flow velocity time series correlation feature vector to obtain a nonlinearly transformed water flow velocity time series correlation feature vector; calculating the correlation matrix between the nonlinearly transformed water flow velocity time series correlation feature vector and each water flow density local time series correlation feature vector in the sequence of the water flow density local time series correlation feature vector to obtain a water flow density-water A sequence of flow velocity time series interaction correlation matrices; calculating the characteristic interaction energy factors of each water flow density-water flow velocity time series interaction correlation matrix in the sequence of water flow density-water flow velocity time series interaction correlation matrix to obtain a sequence of water flow density-water flow velocity time series characteristic interaction energy factors; inputting the sequence of water flow density-water flow velocity time series characteristic interaction energy factors into a Sigmoid function to obtain a sequence of normalized water flow density-water flow velocity time series characteristic interaction energy factors; using the sequence of normalized water flow density-water flow velocity time series characteristic interaction energy factors as a sequence of weights, calculating the position-weighted sum of the sequence of water flow density-water flow velocity time series interaction correlation matrices to obtain the local-global multi-scale interaction response representation matrix of the impact intensity.

[0062] Specifically, first, the water flow density time series correlation feature vector is decomposed into a series of local feature vector sequences through a decoupling operation. This decomposition helps to reveal the different dimensions or aspects within the water flow density time series correlation feature vector, providing more detailed input for subsequent processing. Decoupling enables the model to independently process and analyze each local time series region of the water flow density time series correlation feature vector, thereby enhancing the ability to capture the dynamic local time series characteristics of water flow density. Next, the water flow velocity time series correlation feature vector is nonlinearly transformed in order to introduce more complex feature interactions and enhance the expressive power of the model. Nonlinear transformation can help the model capture more complex dynamic change patterns of water flow velocity in time series, which is crucial for understanding the subtle differences in water flow behavior. Subsequently, the correlation matrix between the water flow velocity time series correlation feature vector after nonlinear transformation and each local feature vector in the decoupled water flow density local feature vector sequence is calculated to form a sequence of water flow density-water flow velocity time series interaction correlation matrices. The correlation matrix captures the interaction between the water flow velocity time series correlation feature vector and the water flow density local time series correlation feature vector after nonlinear transformation, and provides a way to quantify the relationship between the water flow velocity time series features and the water flow density time series features. Then, the characteristic interaction energy factors of each correlation matrix are extracted from the sequence of the water flow density-water flow velocity time series interaction correlation matrix. These characteristic interaction energy interaction factors reflect the interaction intensity and importance between the water flow density time series features and the water flow velocity time series features. The extraction of characteristic interaction energy factors helps to identify and strengthen those characteristic interactions that are most critical to model prediction. It should be understood that the extraction of characteristic interaction energy factors helps to identify and strengthen those characteristic interactions that are most critical to model processing and damage score calculation tasks. The sequence of the water flow density-water flow velocity time series characteristic interaction energy factors is then normalized by the Sigmoid function so that its value range falls within the interval (0, 1) and is converted into attention weights. The normalization process ensures the balance of the weights, so that the model can more reasonably allocate the attention to the interaction of different water flow densities and local time series features of water flow velocities. Finally, the normalized water flow density-water flow velocity time series interaction energy factor is used as the weight to perform weighted summation on the sequence of the water flow density-water flow velocity time series interaction correlation matrix to generate the impact intensity local-global multi-scale interaction response representation matrix. This response representation matrix integrates the interaction information of the local time series characteristics between water flow density and water flow velocity, and provides a comprehensive response representation at the global level, which enhances the model's understanding of the overall time series feature interaction pattern.

[0063] In one embodiment of the present application, a nonlinear transformation is performed on the water flow velocity time series associated characteristic vector to obtain the water flow velocity time series associated characteristic vector after the nonlinear transformation, including: calculating the multiplication of the water flow velocity time series associated characteristic vector and the nonlinear transformation weight matrix, and then performing multiplication with the nonlinear transformation bias vector according to the position point to obtain the water flow velocity time series associated characteristic vector after the nonlinear transformation.

[0064] Further, in one embodiment of the present application, the characteristic interaction energy factor of each water flow density-water flow velocity time series interaction correlation matrix in the sequence of the water flow density-water flow velocity time series interaction correlation matrix is ​​calculated to obtain a sequence of water flow density-water flow velocity time series characteristic interaction energy factors, including: calculating the kurtosis of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction kurtosis value; calculating the global characteristic mean and variance of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction characteristic mean and the water flow density-water flow velocity time series interaction variance; calculating the water flow density-water flow velocity The sum of the time series interaction variance and the predetermined hyperparameter is used to obtain the water flow density-water flow velocity time series interaction variance modulation value; after calculating the square of the difference between the water flow density-water flow velocity time series interaction kurtosis value and the water flow density-water flow velocity time series interaction feature mean, the obtained value is added with twice the water flow density-water flow velocity time series interaction variance and twice the predetermined hyperparameter to obtain the water flow density-water flow velocity time series interaction statistical feature sum value; the division between the water flow density-water flow velocity time series interaction variance modulation value and the water flow density-water flow velocity time series interaction statistical feature sum value is calculated to obtain the water flow density-water flow velocity time series feature interaction energy factor.

[0065] Furthermore, in one embodiment of the present application, the kurtosis of the water flow density-water flow velocity time series interaction correlation matrix is ​​calculated to obtain the water flow density-water flow velocity time series interaction kurtosis value, including: calculating the fourth power of the position difference between the water flow density-water flow velocity time series interaction correlation matrix and the water flow density-water flow velocity time series interaction feature mean, and then calculating the expectation of the obtained feature matrix to obtain the water flow density-water flow velocity time series interaction correlation expectation value; calculating the division between the water flow density-water flow velocity time series interaction correlation expectation value and the square of the water flow density-water flow velocity time series interaction variance to obtain the water flow density-water flow velocity time series interaction kurtosis value.

[0066] Specifically, the water velocity time series associated feature vector and the water flow density time series associated feature vector are input into the feature local-global interactive response module based on feature decoupling for processing using the following feature interactive response formula to obtain the impact intensity local-global multi-scale interactive response representation matrix;

[0067] Wherein, the characteristic interaction response formula is:

[0068]

[0069] in, is the time series correlation characteristic vector of the water flow density, represents the feature decoupling operation, are the first, second, and third in the sequence of local time series correlation feature vectors of water flow density. and The local time series correlation eigenvector of water flow density, is the water velocity time series correlation feature vector, and are the nonlinear transformation weight matrix and nonlinear transformation bias vector respectively, is the time series correlation feature vector of water velocity after nonlinear transformation, is vector multiplication, For the The water flow density-water flow velocity time series interaction correlation matrix corresponding to the local time series correlation eigenvector of water flow density is: and are the global characteristic mean and variance of the water flow density-water flow velocity time series interaction matrix, To predetermine the hyperparameters, is the interaction energy factor of water flow density-water flow velocity time series characteristics, is the kurtosis of the water flow density-water flow velocity time series interaction matrix, To calculate the expectation, is the eigenvalue of each position in the water flow density-water flow velocity time series interaction correlation matrix, for function, is the interaction energy factor of the normalized water flow density-water flow velocity time series characteristics, is the local-global multi-scale interaction response representation matrix of the impact intensity.

[0070] It should be understood that since the water output time series correlation feature vector is used as a query feature vector and the impact intensity local-global multi-scale interactive response representation matrix contains information about the water output time series correlation feature and the impact intensity time series multi-scale response feature, respectively, and since the water output and the impact intensity have mutual effects and influences, this is crucial for damage assessment. Based on this, in order to better integrate and analyze the water output time series feature information and the impact intensity time series feature information to achieve an accurate assessment of the damage to the water outlet in the water conservancy project, in the technical solution of the present application, the water output time series correlation feature vector is further used as a query feature vector and the impact intensity local-global multi-scale interactive response representation matrix is ​​used as an initial key matrix, and the query feature vector and the initial key matrix are input into the key matrix local correlation enhanced cross-domain feature fusion query encoding module to obtain a multi-modal time series correlation representation vector for damage.

[0071] In one embodiment of the present application, the water yield time series correlation feature vector is used as a query feature vector and the impact intensity local-global multi-scale interactive response representation matrix is ​​used as an initial key matrix, and the query feature vector and the initial key matrix are input into a cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix to obtain a damage multi-modal time series correlation representation vector, including: calculating the feature distribution difference energy coefficient between the water yield time series correlation feature vector and each row vector in the initial key matrix to obtain a set of impact intensity local-global multi-scale interactive response feature distribution difference energy coefficients; comparing each impact intensity local-global multi-scale interactive response feature distribution difference energy coefficient in the set of impact intensity local-global multi-scale interactive response feature distribution difference energy coefficients with a predetermined threshold, and in response to the impact intensity local-global multi-scale interactive response feature distribution difference energy coefficient being less than the predetermined threshold, using the impact intensity local-global multi-scale interactive response feature distribution difference energy coefficient as the query feature vector and the initial key matrix. The mean vector between the row vector in the initial key matrix corresponding to the local multi-scale interactive response feature distribution difference energy coefficient and the water yield time series associated feature vector is used as the updated row vector to obtain the updated key matrix; the matrix multiplication between the water yield time series associated feature vector and the value embedding matrix is ​​calculated, and then the water yield time series associated value vector is added positionally; the water yield time series associated feature vector is used as the query vector, each updated row vector in the updated key matrix is ​​used as the key vector, and the water yield time series associated value vector is used as the value vector, and the query vector, the key vector and the value vector are input into the cross-domain attention gate query module based on the Transformer structure to obtain a sequence of cross-domain query attention vectors of local-global multi-scale interactive response of impact strength; the positional mean vector of the sequence of cross-domain query attention vectors of local-global multi-scale interactive response of impact strength is calculated to obtain the multi-modal time series associated representation vector of damage.

[0072] Further, in one embodiment of the present application, the characteristic distribution difference energy coefficient between the water output time series correlation feature vector and each row vector in the initial key matrix is ​​calculated to obtain a set of impact strength local-global multi-scale interactive response characteristic distribution difference energy coefficients, including: respectively calculating the position difference, position point multiplication and position addition between the water output time series correlation feature vector and the row vector in the initial key matrix to obtain the water output time series correlation-impact strength local-global multi-scale interactive response difference feature vector, ... coefficient, water output time series correlation-impact strength local-global multi-scale interactive response difference feature The mutual response dot product feature vector and the water output time series correlation-impact intensity local-global multi-scale interactive response sum feature vector; the water output time series correlation-impact intensity local-global multi-scale interactive response differential feature vector, the water output time series correlation-impact intensity local-global multi-scale interactive response dot product feature vector and the water output time series correlation-impact intensity local-global multi-scale interactive response sum feature vector are cascaded and processed through a one-dimensional convolution layer, and the obtained feature vector is subjected to maximum pooling processing to obtain the water output time series correlation-impact intensity local-global multi-scale interactive response A feature association representation vector; extracting the feature maximum value in the water yield time series association-impact intensity local-global multi-scale interactive response feature association representation vector to obtain the water yield time series association-impact intensity local-global multi-scale interactive response association maximum representation value; calculating the mean and variance of the row vectors in the initial key matrix to obtain the impact intensity local-global multi-scale interactive response feature mean and the impact intensity local-global multi-scale interactive response feature variance; adding the impact intensity local-global multi-scale interactive response feature variance to the hyperparameter to obtain the first feature distribution difference energy factor; calculating the square of the difference between the water yield time series association-impact intensity local-global multi-scale interactive response association maximum representation value and the impact intensity local-global multi-scale interactive response feature mean to obtain the semantic feature difference value; multiplying the impact intensity local-global multi-scale interactive response feature variance by a constant of two and then adding it to the semantic feature difference value and the hyperparameter to obtain the second feature distribution difference energy factor; dividing the first feature distribution difference energy factor by the second feature distribution difference energy factor to obtain the impact intensity local-global multi-scale interactive response feature distribution difference energy coefficient.

[0073] Specifically, first, the energy coefficient of the feature distribution difference between the query feature vector (i.e., the water output time series associated feature vector) and each row vector in the initial key matrix (i.e., the impact intensity local-global multi-scale interactive response representation matrix) is calculated. The purpose of this step is to quantify the feature distribution difference between the query feature vector and each feature vector in the key matrix. Next, the energy coefficient of the feature distribution difference of each impact intensity local-global multi-scale interactive response is compared with a predetermined threshold. If the energy coefficient of the feature distribution difference of a certain impact intensity local-global multi-scale interactive response is less than the predetermined threshold, it is considered that the query feature vector, i.e., the water output time series associated feature vector, is relatively similar to the key vector in the corresponding key matrix. At this time, the mean vector of the query feature vector and the corresponding row vector is used to update the row vector in the key matrix. Here, the update operation strengthens the row vector that is more similar to the query vector, i.e., the water output time series associated feature vector. Through the local association strengthening mechanism, the pertinence and feature expression quality of the key matrix are improved. Furthermore, on the basis of updating the key matrix, after linear transformation of the water yield time series associated feature vector, the water yield time series associated feature vector is used as the query vector, each updated row vector in the updated key matrix is ​​used as the key vector, and the water yield time series associated value vector is used as the value vector, and they are input into the cross-domain attention gate query module based on the Transformer structure, so as to utilize the cross-domain attention mechanism, combine the query vector, key vector and value vector to perform feature self-attention calculation, and realize cross-domain information integration, thereby improving the richness and accuracy of the representation of the local-global multi-scale time series features of the impact strength and the water yield time series features of the support concentration responsiveness query attention features. Furthermore, by calculating the mean of the sequence of the local-global multi-scale interactive response cross-domain query attention vector of the impact strength by position, the final damage multi-modal time series associated representation vector is obtained. The representation vector integrates the information of all cross-domain query attention vectors in the sequence and is a comprehensive feature representation. It helps to improve the model's ability to understand the temporal responsiveness interaction characteristics between the impact intensity temporal semantics and the water output temporal characteristics in complex scenarios, thereby improving the model's adaptability and generalization capabilities for complex scenarios and improving the accuracy of subsequent damage assessment tasks.

[0074] In one embodiment of the present application, the water output time series associated feature vector is used as a query vector, each updated row vector in the update key matrix is ​​used as a key vector, and the water output time series associated value vector is used as a value vector. The query vector, the key vector, and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of impact intensity local-global multi-scale interactive response cross-domain query attention vectors, including: after calculating the product between the query vector and the transposed vector of the key vector, the obtained query-key feature association value is divided by the square root of the length of the key vector to obtain a query-key feature association coefficient; the query-key feature association coefficient is input into a softmax function to obtain a query-key feature association weight value; the query-key feature association weight value is used as a weight to weight the transposed vector of the value vector to obtain an impact intensity local-global multi-scale interactive response cross-domain query attention vector.

[0075] Then, the damage multimodal temporal correlation representation vector is input into the damage assessor based on the decoder to obtain the decoded value of the damage score. In other words, the damage multimodal temporal correlation feature representation is decoded and regressed to decode and generate the damage score. In this way, the damage score that conforms to the actual situation can be adaptively generated according to the temporal dynamic changes and synergistic correlation between the water velocity, water flow density and water output by using artificial intelligence technology and deep learning algorithms, thereby improving the intelligence level and generalization ability of the water conservancy project management method.

[0076] In a preferred example, inputting the impairment multimodal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of the impairment score comprises:

[0077] Determine the maximum eigenvalue of the damage multimodal temporal association of the damage multimodal temporal association representation vector and the minimum eigenvalue of the damage multimodal temporal association, and calculate the damage multimodal temporal association mean and damage multimodal temporal association standard deviation of the feature set of the damage multimodal temporal association representation vector;

[0078] Calculating the quotient of the damage multimodal time series association mean and the damage multimodal time series association standard deviation to obtain a damage multimodal time series association statistical normalization value;

[0079] Calculate the inverse of each eigenvalue of the damage multimodal time series association representation vector, multiply it by the difference between the maximum eigenvalue of the damage multimodal time series association and the minimum eigenvalue of the damage multimodal time series association, and then subtract it from the damage multimodal time series association statistical normalization value to obtain a damage multimodal time series association distribution approximation vector;

[0080] Calculate an exponential function with a natural constant as the base and each eigenvalue of the damage multimodal time series correlation distribution approximation vector as the exponent to obtain a damage multimodal time series correlation distribution class approximation vector;

[0081] Performing point addition on the damage multimodal temporal association distribution class approximation vector and the damage multimodal temporal association statistical normalization value, and calculating the base-two logarithm of the absolute value of each eigenvalue of the point-added vector to obtain an optimized damage multimodal temporal association representation vector; and

[0082] The optimized impairment multi-modal temporal correlation representation vector is input into a decoder-based impairment assessor to obtain a decoded value of the impairment score.

[0083] Among them, the damage multi-modal temporal correlation representation vector is denoted as The optimization expression is:

[0084]

[0085] is the statistical normalization value of the multi-modal temporal correlation of damage, is the multi-modal temporal correlation mean of the damage, is the standard deviation of the multimodal temporal correlation of the injury, and is the difference between the maximum eigenvalue of the damage multimodal time series association and the minimum eigenvalue of the damage multimodal time series association, is the logarithm to base 2, is the minimum eigenvalue of the damage multimodal temporal correlation representation vector of the damage multimodal temporal correlation, is the maximum eigenvalue of the damage multimodal temporal correlation representation vector of the damage multimodal temporal correlation, is the multi-modal temporal correlation mean of the damage, is the standard deviation of the multimodal temporal correlation of the injury, is the inverse of each eigenvalue of the damage multimodal temporal correlation representation vector, is the multi-modal temporal correlation representation vector of the damage, It is an exponential function that calculates the exponential of each eigenvalue of a vector with a natural constant as the base. is the approximate vector of the multi-modal temporal correlation distribution of damage, is the optimized damage multi-modal temporal correlation representation vector, is the base-two logarithm, It is the point multiplication by position. It is added by location point. It is subtracted by position point.

[0086] Here, in the preferred example, since the water yield time series correlation feature vector and each row vector in the impact intensity local-global multi-scale interactive response representation matrix represent the water yield time series correlation feature and the water flow velocity-water flow density time series correlation feature respectively, when they are input into the cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix, the semantic feature representation difference will have different cross-domain attention query weights based on the local correlation difference of the key matrix, so that the damage multimodal time series correlation representation vector will also have a diversified set expression distribution of cross-semantic space aggregation features. Therefore, it is expected to improve the balance between the regression mapping accuracy and completeness of the damage multimodal time series correlation representation vector when the damage evaluator based on the decoder is input for decoding regression, thereby improving the accuracy of the decoded value of the obtained damage score.

[0087] Therefore, by performing random statistical normalization on the diversified feature set of the damage multimodal time series association representation vector, the confidence space constructed based on the overall eigenvalue of the damage multimodal time series association representation vector is approximated by the standardized continuous probability density distribution relative to the response hypothesis test of each eigenvalue of the damage multimodal time series association representation vector, thereby establishing the target reachability of the diversified feature distribution of the damage multimodal time series association representation vector to the unified regression target, so as to achieve the balanced executableness between mapping accuracy and mapping completeness in the decoding regression process based on the diversified feature distribution of the damage multimodal time series association representation vector, and improve the accuracy of the decoded value of the damage score obtained by the damage evaluator based on the decoder of the damage multimodal time series association representation vector input. In this way, a more accurate damage score that conforms to the actual situation can be adaptively generated, thereby improving the intelligence level and generalization ability of the water conservancy project management method.

[0088] In summary, the above scheme is adopted to extract the time series features and perform cross-domain dynamic collaborative correlation analysis on the water discharge data by using data processing and analysis algorithms based on artificial intelligence and deep learning at the back end, so as to capture the time series semantics of impact intensity based on the time series features of water flow velocity and water flow density, and automatically obtain the damage score by cross-domain feature fusion query features between the time series semantics of impact intensity and the time series semantics of water discharge. In this way, it is possible to adaptively generate damage scores that conform to the actual situation according to the time series dynamic changes and collaborative correlation between water flow velocity, water flow density and water discharge by using artificial intelligence technology and deep learning algorithms, thereby improving the intelligence level and generalization ability of water conservancy project management methods.

[0089] Figure 2 FIG. 1 is a block diagram of a water conservancy project management system based on digital twins according to an exemplary embodiment. Figure 2 As shown, the water conservancy project management system 200 includes:

[0090] The target water outlet water data calculation module 201 is used to calculate the water outlet data of the target water outlet based on the water inlet and the water outlet model, wherein the water outlet model includes multiple water outlets, and the target water outlet is at least one water outlet in the water outlet model;

[0091] A damage score calculation module 202 for a target water outlet is used to calculate the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density, and a time queue of water output;

[0092] An assessment result generating module 203, used to generate an assessment result based on the damage score;

[0093] The damage score calculation module 202 of the target water outlet includes:

[0094] The water discharge sequence encoding unit 2021 is used to perform water discharge sequence encoding on the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water discharge volume respectively to obtain a water flow velocity time series correlation feature vector, a water flow density time series correlation feature vector and a water discharge volume time series correlation feature vector;

[0095] A characteristic local-global interactive response unit 2022 is used to input the water velocity time series associated characteristic vector and the water flow density time series associated characteristic vector into a characteristic local-global interactive response module based on characteristic decoupling to obtain an impact intensity local-global multi-scale interactive response representation matrix;

[0096] A cross-domain feature fusion query encoding unit 2023 is used to use the water discharge time series correlation feature vector as a query feature vector and the impact strength local-global multi-scale interactive response representation matrix as an initial key matrix, and input the query feature vector and the initial key matrix into a cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix to obtain a damage multi-modal time series correlation representation vector;

[0097] The decoding unit 2024 is used to input the impairment multi-modal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of the impairment score.

[0098] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present application. The terminal device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0099] like Figure 3 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0100] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0101] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present application are executed.

[0102] It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0103] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0104] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0105] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0106] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0107] The modules involved in the embodiments described in this application can be implemented by software or hardware. The name of the module does not limit the module itself in some cases. For example, the test parameter acquisition module can also be described as a "module for acquiring device test parameters corresponding to the target device".

[0108] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0109] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0110] Figure 4 FIG. 1 is an application scenario diagram of a water conservancy project management method based on digital twins according to an exemplary embodiment. Figure 4 As shown, in this application scenario, first, the damage score of the target water outlet is calculated based on the water outlet data, wherein the water outlet data includes a time sequence of water flow velocity (for example, Figure 4 C1 as shown in FIG. 1 ), the time sequence of the water flow density (e.g., Figure 4 C2) and the time sequence of the water output (e.g., as Figure 4 Then, the acquired time queue of water flow velocity, the time queue of water flow density and the time queue of water output are input into a server (for example, Figure 4 In S) shown in , the server is capable of processing the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water output based on the water conservancy project management algorithm of the digital twin to damage the decoded value of the score.

[0111] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

[0112] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0113] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely exemplary forms of implementation. With respect to the apparatus in the above-mentioned embodiments, the specific manner in which each module performs the operation has been described in detail in the embodiments related to the method, and will not be described in detail here.

Claims

1. A water conservancy project management method based on digital twins, characterized in that: include: Calculate the water outlet data of the target water outlet based on the water inlet and the water outlet model, wherein the water outlet model includes a plurality of water outlets, and the target water outlet is at least one water outlet in the water outlet model; Calculating the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density, and a time queue of water outlet volume; generating an assessment result based on the damage score; Wherein, calculating the damage score of the target water outlet based on the water outlet data includes: Performing water discharge sequence encoding on the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water discharge volume respectively to obtain a water flow velocity time series associated feature vector, a water flow density time series associated feature vector and a water discharge volume time series associated feature vector; Inputting the water velocity time series associated feature vector and the water flow density time series associated feature vector into a feature local-global interactive response module based on feature decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix; Taking the water discharge time series correlation feature vector as the query feature vector and the impact strength local-global multi-scale interactive response representation matrix as the initial key matrix, the query feature vector and the initial key matrix are input into a cross-domain feature fusion query encoding module with key matrix local correlation enhancement to obtain a damage multi-modal time series correlation representation vector; Inputting the impairment multimodal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of an impairment score; Inputting the impairment multimodal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of an impairment score includes: Determine the maximum eigenvalue of the damage multimodal temporal association of the damage multimodal temporal association representation vector and the minimum eigenvalue of the damage multimodal temporal association, and calculate the damage multimodal temporal association mean and damage multimodal temporal association standard deviation of the feature set of the damage multimodal temporal association representation vector; Calculating the quotient of the damage multimodal time series association mean and the damage multimodal time series association standard deviation to obtain a damage multimodal time series association statistical normalization value; Calculate the inverse of each eigenvalue of the damage multimodal time series association representation vector, multiply it by the difference between the maximum eigenvalue of the damage multimodal time series association and the minimum eigenvalue of the damage multimodal time series association, and then subtract it from the damage multimodal time series association statistical normalization value to obtain a damage multimodal time series association distribution approximation vector; Calculate an exponential function with a natural constant as the base and each eigenvalue of the damage multimodal time series correlation distribution approximation vector as the exponent to obtain a damage multimodal time series correlation distribution class approximation vector; Performing point addition on the damage multimodal temporal association distribution class approximation vector and the damage multimodal temporal association statistical normalization value, and calculating the base-two logarithm of the absolute value of each eigenvalue of the point-added vector to obtain an optimized damage multimodal temporal association representation vector; and Inputting the optimized impairment multimodal temporal association representation vector into a decoder-based impairment assessor to obtain a decoded value of an impairment score; Among them, the damage multi-modal temporal correlation representation vector is denoted as The optimization expression is: ； is the statistical normalization value of the multi-modal temporal correlation of damage, is the multi-modal temporal correlation mean of the damage, is the standard deviation of the multimodal temporal correlation of the injury, and is the difference between the maximum eigenvalue of the damage multimodal time series association and the minimum eigenvalue of the damage multimodal time series association, is the logarithm to base 2, is the minimum eigenvalue of the damage multimodal temporal correlation representation vector of the damage multimodal temporal correlation, is the maximum eigenvalue of the damage multimodal temporal correlation representation vector of the damage multimodal temporal correlation, is the inverse of each eigenvalue of the damage multimodal temporal correlation representation vector, is the multi-modal temporal correlation representation vector of the damage, It is an exponential function that calculates the exponential of each eigenvalue of a vector with a natural constant as the base. is the approximate vector of the multi-modal temporal correlation distribution of damage, is the optimized damage multi-modal temporal correlation representation vector, It is the point multiplication by position. It is added by location point. It is subtracted by position point; The water velocity time series associated feature vector and the water flow density time series associated feature vector are input into a feature local-global interactive response module based on feature decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix, including: Performing feature decoupling on the water flow density time series correlation feature vector to obtain a sequence of water flow density local time series correlation feature vectors; Performing a nonlinear transformation on the water flow velocity time series associated characteristic vector to obtain a nonlinearly transformed water flow velocity time series associated characteristic vector; Calculating the correlation matrix between the water flow velocity time series correlation feature vector after the nonlinear transformation and each water flow density local time series correlation feature vector in the sequence of the water flow density local time series correlation feature vector to obtain a sequence of water flow density-water flow velocity time series interactive correlation matrices; Calculating the characteristic interaction energy factor of each water flow density-water flow velocity time series interaction correlation matrix in the sequence of water flow density-water flow velocity time series interaction correlation matrix to obtain a sequence of water flow density-water flow velocity time series characteristic interaction energy factors; Inputting the sequence of water flow density-water flow velocity time series characteristic interaction energy factors into a Sigmoid function to obtain a sequence of normalized water flow density-water flow velocity time series characteristic interaction energy factors; Taking the sequence of the normalized water flow density-water flow velocity time series characteristic interaction energy factors as the sequence of weights, calculating the position-weighted sum of the sequence of the water flow density-water flow velocity time series interaction correlation matrix to obtain the impact intensity local-global multi-scale interaction response representation matrix; Calculating the characteristic interaction energy factor of each water flow density-water flow velocity time series interaction correlation matrix in the sequence of water flow density-water flow velocity time series interaction correlation matrix to obtain a sequence of water flow density-water flow velocity time series characteristic interaction energy factors, including: Calculating the kurtosis of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction kurtosis value; Calculating the global feature mean and variance of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction feature mean and the water flow density-water flow velocity time series interaction variance; Calculating the sum of the water flow density-water flow velocity time series interaction variance and a predetermined hyperparameter to obtain a water flow density-water flow velocity time series interaction variance modulation value; After calculating the square of the difference between the water flow density-water flow velocity time series interaction kurtosis value and the water flow density-water flow velocity time series interaction feature mean, the obtained value is added to twice the water flow density-water flow velocity time series interaction variance and twice the predetermined hyperparameter to obtain the water flow density-water flow velocity time series interaction statistical feature sum value; Calculate the division between the water flow density-water flow velocity time series interaction variance modulation value and the water flow density-water flow velocity time series interaction statistical feature sum value to obtain the water flow density-water flow velocity time series feature interaction energy factor; The water discharge time series correlation feature vector is used as a query feature vector and the impact strength local-global multi-scale interactive response representation matrix is ​​used as an initial key matrix, and the query feature vector and the initial key matrix are input into a cross-domain feature fusion query encoding module with key matrix local correlation enhancement to obtain a damage multi-modal time series correlation representation vector, including: Calculating the characteristic distribution difference energy coefficients between the water output time series correlation characteristic vector and each row vector in the initial bond matrix to obtain a set of impact strength local-global multi-scale interactive response characteristic distribution difference energy coefficients; Comparing each impact intensity local-global multi-scale interactive response characteristic distribution difference energy coefficient in the set of impact intensity local-global multi-scale interactive response characteristic distribution difference energy coefficients with a predetermined threshold, and in response to the impact intensity local-global multi-scale interactive response characteristic distribution difference energy coefficient being less than the predetermined threshold, taking the mean vector between the row vector in the initial key matrix corresponding to the impact intensity local-global multi-scale interactive response characteristic distribution difference energy coefficient and the water output time series correlation characteristic vector as the updated row vector to obtain an updated key matrix; Calculate the matrix multiplication between the water output time series correlation feature vector and the value embedding matrix, and then add the matrix to the value embedding vector by position to obtain the water output time series correlation value vector; Taking the water output time series associated feature vector as a query vector, each updated row vector in the update key matrix as a key vector and the water output time series associated value vector as a value vector, the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of cross-domain query attention vectors of impact intensity local-global multi-scale interactive response; The position-wise mean vector of the sequence of the impact intensity local-global multi-scale interactive response cross-domain query attention vector is calculated to obtain the damage multi-modal temporal association representation vector.

2. The water conservancy project management method based on digital twin according to claim 1 is characterized in that: The time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water output are respectively encoded with water output sequence to obtain the water flow velocity time series associated feature vector, the water flow density time series associated feature vector and the water output time series associated feature vector, including: inputting the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water output into a water output sequence encoder based on a Bi-LSTM model to obtain the water flow velocity time series associated feature vector, the water flow density time series associated feature vector and the water output time series associated feature vector.

3. The water conservancy project management method based on digital twin according to claim 2 is characterized in that: The water flow velocity time series associated characteristic vector is subjected to a nonlinear transformation to obtain the water flow velocity time series associated characteristic vector after the nonlinear transformation, including: calculating the multiplication of the water flow velocity time series associated characteristic vector and the nonlinear transformation weight matrix, and then performing multiplication with the nonlinear transformation bias vector according to the position point to obtain the water flow velocity time series associated characteristic vector after the nonlinear transformation.

4. The water conservancy project management method based on digital twin according to claim 3 is characterized in that: Calculating the kurtosis of the water flow density-water flow velocity time series interaction correlation matrix to obtain the water flow density-water flow velocity time series interaction kurtosis value includes: After calculating the fourth power of the position difference between the water flow density-water flow velocity time series interaction correlation matrix and the water flow density-water flow velocity time series interaction characteristic mean, the expectation of the obtained characteristic matrix is ​​calculated to obtain the water flow density-water flow velocity time series interaction correlation expectation value; The division between the expected value of the water flow density-water flow velocity time series interaction correlation and the square of the water flow density-water flow velocity time series interaction variance is calculated to obtain the water flow density-water flow velocity time series interaction kurtosis value.

5. The water conservancy project management method based on digital twin according to claim 4 is characterized in that: Calculating the characteristic distribution difference energy coefficients between the water output time series correlation characteristic vector and each row vector in the initial bond matrix to obtain a set of impact strength local-global multi-scale interactive response characteristic distribution difference energy coefficients, including: Respectively calculating the position difference, position point product and position addition between the water yield time series correlation feature vector and the row vector in the initial key matrix to obtain the water yield time series correlation-impact intensity local-global multi-scale interactive response difference feature vector, the water yield time series correlation-impact intensity local-global multi-scale interactive response point product feature vector and the water yield time series correlation-impact intensity local-global multi-scale interactive response sum feature vector; The water yield time series association-impact intensity local-global multi-scale interactive response differential feature vector, the water yield time series association-impact intensity local-global multi-scale interactive response dot product feature vector and the water yield time series association-impact intensity local-global multi-scale interactive response sum feature vector are cascaded and processed through a one-dimensional convolution layer, and the obtained feature vector is subjected to maximum pooling processing to obtain a water yield time series association-impact intensity local-global multi-scale interactive response feature association representation vector; Extracting the maximum value of the characteristic in the water yield time series correlation-impact intensity local-global multi-scale interactive response characteristic correlation representation vector to obtain the maximum representation value of the water yield time series correlation-impact intensity local-global multi-scale interactive response correlation; Calculating the mean and variance of the row vectors in the initial bond matrix to obtain the mean of the local-global multi-scale interactive response characteristics of the impact intensity and the variance of the local-global multi-scale interactive response characteristics of the impact intensity; Adding the impact intensity local-global multi-scale interactive response characteristic variance to the hyperparameter to obtain a first characteristic distribution difference energy factor; Calculate the square of the difference between the maximum representation value of the water output time series correlation-impact intensity local-global multi-scale interactive response correlation and the mean value of the impact intensity local-global multi-scale interactive response feature to obtain a semantic feature difference value; The impact intensity local-global multi-scale interactive response feature variance is multiplied by a constant of two and then added to the semantic feature difference value and the hyperparameter to obtain a second feature distribution difference energy factor; The first characteristic distribution difference energy factor is divided by the second characteristic distribution difference energy factor to obtain the impact strength local-global multi-scale interactive response characteristic distribution difference energy coefficient.

6. The water conservancy project management method based on digital twin according to claim 5 is characterized in that: The water output time series associated feature vector is used as a query vector, each updated row vector in the update key matrix is ​​used as a key vector, and the water output time series associated value vector is used as a value vector. The query vector, the key vector, and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of cross-domain query attention vectors of impact intensity local-global multi-scale interactive response, including: After calculating the product between the query vector and the transposed vector of the key vector, dividing the obtained query-key feature association value by the square root of the length of the key vector to obtain a query-key feature association coefficient; Inputting the query-key feature association coefficient into a softmax function to obtain a query-key feature association weight value; The query-key feature association weight value is used as a weight to weight the transposed vector of the value vector to obtain an impact intensity local-global multi-scale interactive response cross-domain query attention vector.

7. A management system using the water conservancy project management method based on digital twins according to claim 1, characterized in that: include: A water outlet data calculation module for a target water outlet, used to calculate the water outlet data of the target water outlet based on the water inlet and the water outlet model, wherein the water outlet model includes a plurality of water outlets, and the target water outlet is at least one water outlet in the water outlet model; a damage score calculation module for a target water outlet, used to calculate the damage score of the target water outlet based on the water outlet data, wherein the water outlet data includes a time queue of water flow velocity, a time queue of water flow density, and a time queue of water output; An assessment result generating module, used for generating an assessment result based on the damage score; Wherein, the damage score calculation module of the target water outlet includes: A water discharge sequence encoding unit, used for performing water discharge sequence encoding on the time queue of the water flow velocity, the time queue of the water flow density and the time queue of the water discharge volume respectively to obtain a water flow velocity time series correlation feature vector, a water flow density time series correlation feature vector and a water discharge volume time series correlation feature vector; A characteristic local-global interactive response unit, used for inputting the water velocity time series associated characteristic vector and the water flow density time series associated characteristic vector into a characteristic local-global interactive response module based on characteristic decoupling to obtain an impact strength local-global multi-scale interactive response representation matrix; A cross-domain feature fusion query encoding unit is used to use the water discharge time series correlation feature vector as a query feature vector and the impact strength local-global multi-scale interactive response representation matrix as an initial key matrix, and input the query feature vector and the initial key matrix into a cross-domain feature fusion query encoding module with local correlation enhancement of the key matrix to obtain a damage multi-modal time series correlation representation vector; The decoding unit is used to input the damage multi-modal temporal association representation vector into a damage assessor based on a decoder to obtain a decoded value of the damage score.

Images