A pump station unit flow field rapid identification system and method based on digital twin

Abstract

The present invention discloses a system and method for rapid identification of flow fields of pump station units based on digital twins. The system comprises: a data acquisition module, a pump station drive model construction module, a flow field identification model construction module and a flow field identification module; the data acquisition module is used to install sensors on the pump station units and collect pump station unit data according to the sensors; the pump station drive model construction module is used to process the pump station unit data, and construct a pump station drive model based on the processed pump station unit data, and use the pump station drive mechanism to simulate the pump station unit data; the flow field identification model construction module is used to perform preprocessing based on the simulation data to obtain preprocessed data, and construct a digital twin degradation model based on the preprocessed data and the pump station drive machine model; the flow field identification module is used to perform flow field identification on the flow field of the pump station unit based on the digital twin degradation model.

Description

Technical Field

[0001] The present invention relates to the technical field of pump stations, and in particular to a system and method for rapid identification of flow fields of pump station units based on digital twins. Background Art

[0002] As one of the most important components of the South-to-North Water Diversion Project, pump stations are arranged in tiers at intervals of kilometers to advance water sources and control the operating parameters of the entire system. As the "heart" of the pump station, the transient flow characteristics of the liquid in the pump put forward higher requirements on the safety and reliability of the pump station unit under the premise of meeting different water diversion needs. In recent years, the concept of "digital twin" has been proposed and expanded from the aerospace field to the field of water conservancy engineering. Its broad meaning is that by creating a digital twin model of the actual pump station and performing digital simulation of the physical system, the state characteristics can be output in real time to solve the problem of data blind spots in traditional monitoring.

[0003] At the same time, the current digital twin models are mainly divided into data-driven models and physical-driven models. The data-driven model mainly uses the historical operation data of the pumping station as the main guide for black box testing. However, it is difficult for most pumping stations to obtain complete historical data. At the same time, the credibility of the data-driven model is low and does not conform to the actual operation rules. The physical-driven model mainly models the operation mechanism of the pumping station unit. Although it conforms to the operation logic, its modeling difficulty is high and the calculation time is long. At the same time, the parameters involved in the modeling are not easy to obtain. Therefore, combining the two methods can not only meet the operation mechanism requirements, but also solve the problem of insufficient data.

[0004] During the long-term operation of the pump station unit, the flow field characteristics will appear to varying degrees due to the influence of various environments and loads. Timely identification and analysis of the flow field characteristics can ensure the normal and safe operation of the pump station unit while reducing the loss of personnel and property. The flow field identification of the pump station unit is an important part of the digital twin pump station empowerment. Its purpose is to identify the non-steady-state characteristic structure in the flow field at the earliest possible stage to prevent further deterioration and affect the safe operation of the pump station unit. Summary of the invention

[0005] In order to overcome the problems that most existing digital twin models adopt a single driving method and data redundancy causes update lag, making it difficult to achieve online real-time flow field health monitoring, the present invention constructs a digital twin degradation model of a pump station unit, and updates the basis function after intrinsic orthogonal decomposition in real time. The basis function can quickly reflect the internal flow field of the pump station unit, and can also "preview" the flow field identification in different scenarios. The present invention proposes a pump station unit flow field rapid identification system and method based on digital twins, and the system specifically includes: a data acquisition module, a pump station drive model construction module, a flow field identification model construction module and a flow field identification module;

[0006] The data acquisition module is used to install sensors on the pump station units and collect pump station unit data according to the sensors;

[0007] The pump station drive model building module is used to process the pump station unit data, and build a pump station drive model based on the processed pump station unit data, and use the pump station drive mechanism to simulate the pump station unit data;

[0008] The flow field identification model building module is used to perform preprocessing based on simulation data to obtain preprocessed data, and to build a digital twin degradation model based on the preprocessed data and a pump station driver model;

[0009] The flow field identification module is used to perform flow field identification on the flow field of the pump station unit based on the digital twin degradation model.

[0010] Optionally, the data acquisition module includes a device installation submodule and a data transmission submodule;

[0011] The equipment installation submodule is used to install the sensor at a preset position of the pump station unit;

[0012] The data transmission submodule is used to build a matching model based on sensor data and sensor position, and obtain the collected pump station unit data based on the matching model.

[0013] Optionally, the pump station drive model building module includes a drive mechanism submodule and a data drive submodule;

[0014] The driving mechanism submodule is used to construct a mechanism based on the fluid mechanics and internal environmental parameters in the operation of the pump station, and to perform mechanism simulation according to the constructed mechanism;

[0015] The data driving submodule is used for performing data preprocessing based on the driving mechanism submodule and the data acquisition module.

[0016] Optionally, the driving mechanism submodule includes a conservation mechanism unit, an environmental mechanism unit and a simulation mechanism unit;

[0017] The conservation mechanism unit is used for the fluid mechanics theory involved in the operation of the pump station unit to construct the conservation mechanism of the pump station unit;

[0018] The environmental mechanism unit is used to construct the environmental mechanism of the pump station unit according to the variable parameters of the pump station unit;

[0019] The simulation mechanism unit is used to determine the pump station simulation model by combining the conservation mechanism and the environmental mechanism.

[0020] Optionally, the data driven submodule includes: a data dimension reduction submodule, a data reconstruction submodule and a data update submodule;

[0021] The data dimension reduction submodule is used to simplify the data simulated by the driving mechanism submodule by using the method of intrinsic orthogonal decomposition, thereby achieving dimension reduction of the unsteady flow field pressure data and calculating the dimension reduction eigenvalue;

[0022] The data reconstruction submodule is used to restore and reconstruct the flow field unit based on the dimension reduction eigenvalue;

[0023] The data updating submodule is used to update the reconstructed data using a Kalman filter algorithm.

[0024] Optionally, the working process of the data dimension reduction submodule specifically includes:

[0025] The pressure U(x,t i ) is recorded as a set of instantaneous pressure matrices. The pressures at different times constitute the instantaneous pressure matrix U(x, t), where x represents the position vector and t represents the time. The instantaneous pressure matrix can be expressed as the sum of the average pressure matrix U(x, t) and the pulsating pressure matrix U'(x, t):

[0026]

[0027] Using the separation of variables method, the pulsating pressure matrix is ​​decomposed into two parts that are independent of each other in time and space, and a series of time-related coefficients a i (t) and a series of spatial orthogonal basis functions The product of:

[0028]

[0029] In the formula, is the POD space mode, also called POD basis; a i (t) is the modal coefficient corresponding to each mode, also known as the time coefficient.

[0030] Optionally, the working process of the data reconstruction submodule specifically includes:

[0031]

[0032] In the formula represents the approximate deviation matrix of pulsating pressure, N is the total number of intercepted modes;

[0033] In order to ensure the accuracy of the modal truncation process, the selection of the first N modes should make the transient pressure matrix U'(x, t) and the approximate deviation matrix The error between them is the smallest, and the second-order norm is used to construct the error expression to satisfy the following formula:

[0034]

[0035] Where || ||2 Represents the second norm of a vector.

[0036] In order to ensure the accuracy of pressure field reconstruction, the difference between the pulsating pressure matrix and the approximate deviation matrix is ​​minimized; the error between the truncated intrinsic orthogonal mode and the pulsating pressure matrix can be expressed as:

[0037]

[0038] Bring in get

[0039]

[0040] Let R=(U'U' T ), from linear algebra we know that R is a symmetric matrix. By constructing the Lagrangian function to find the extreme value, the error can be expressed as:

[0041]

[0042] where λ i is the eigenvalue of the real symmetric matrix R, is the eigenvector of the real symmetric matrix R.

[0043] Optionally, the working process of the data updating submodule specifically includes:

[0044] Update the first few order basis functions of the intrinsic orthogonal decomposition and reconstruction, and arrange the first few order basis functions in column form as column vectors. The evolution of the column vectors over time can be expressed as follows:

[0045]

[0046] Where Φ l,k The first few order basis functions represented by can change over time; considering the possible system evolution, the evolution of the first few order function space over time can be expressed as:

[0047] Φ l,k =Φ l,k-1 +w k

[0048] Where w k is a zero-mean Gaussian white noise, whose time-invariant covariance is adjusted to obtain unbiased estimates of the first few order basis functions;

[0049] Use an observation method to supplement and update the first few order function spaces according to the measured values:

[0050] y k,r =H k,r Φ l,k +v k,r

[0051] Where H k,r is a transformation matrix that combines the first few order basis function spaces with the observation equations, and updates the first few order basis functions of the flow field through the pressure response, v k,r represents the observation noise;

[0052] Since the change of the state vector is linear, the Kalman filter method is used to recursively update the basis function space; the initial value of the basis function space and its initial covariance are defined as and P0;

[0053] Prediction: Prior

[0054] Prior error covariance

[0055] In the formula represents the basis function state prediction error covariance matrix;

[0056] Correction: Kalman gain

[0057] Posterior Estimation

[0058] Update error covariance

[0059] In the formula is the updated state estimate, P k is the filtering (a posteriori) error covariance matrix.

[0060] The present invention also discloses a method for rapid identification of flow field of pump station units based on digital twins, the method comprising:

[0061] Install sensors on the pump station units and collect data on the pump station units based on the sensors;

[0062] Process the pump station unit data, build a pump station drive model based on the processed pump station unit data, and use the pump station drive mechanism to simulate the pump station unit data;

[0063] Preprocessing is performed based on the simulation data to obtain preprocessed data, and a digital twin degradation model is constructed based on the preprocessed data and the pump station driver model;

[0064] The digital twin degradation model is used to identify the flow field of the pump station unit.

[0065] Optionally, a sensor is installed on the pump station unit, and the method for collecting the data of the pump station unit according to the sensor specifically includes:

[0066] Install sensors at preset locations of pump station units;

[0067] A matching model is constructed based on the sensor data and the sensor position, and the collected pump station unit data is obtained based on the matching model.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] The present invention constructs a digital twin degradation model of the pump station unit through digital twin technology, integrates the advantages of intrinsic orthogonal decomposition and Kalman filtering technology, and realizes the identification of the flow field by only using the first few order basis functions with a relatively large energy share. By introducing the Kalman filtering algorithm, the online update pressure of the flow field characteristic parameters is increased, and the online identification efficiency of the flow field is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0071] Figure 1 This is a system structure diagram of a pump station unit flow field rapid identification system based on digital twins according to an embodiment of the present invention;

[0072] Figure 2 This is a flow chart of a method for constructing a digital twin degradation model of a pump station unit according to an embodiment of the present invention;

[0073] Figure 3 It is a flow chart of the driving mechanism submodule of the pump station unit according to an embodiment of the present invention;

[0074] Figure 4 This is a flow chart of the pump station unit data driving submodule according to an embodiment of the present invention;

[0075] Figure 5 This is a flow chart of a method for rapid identification of flow fields of pump station units based on digital twins according to an embodiment of the present invention;

[0076] Figure 6 A flow chart for constructing a flow field identification model of a pump station unit based on digital twins according to an embodiment of the present invention. Specific implementation methods

[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0079] Embodiment 1

[0080] A rapid identification system of flow field of pump station units based on digital twins, such as Figure 1 As shown, it specifically includes: a data acquisition module, a pump station drive model construction module, a flow field identification model construction module and a flow field identification module.

[0081] The data acquisition module is used to install sensors on the pump station unit and collect pump station unit data according to the sensors; the data acquisition module includes an equipment installation submodule and a data transmission submodule; the equipment installation submodule is used to install sensors at the preset position of the pump station unit; in the sensor module establishment stage, sensors are installed in certain parts of the pump station unit according to different data collection purposes, such as installing displacement and pressure sensors on the machine base, mainly to monitor whether the vibration of the machine base is offset beyond the safety value; installing pressure sensors at the guide vanes of the unit, the purpose is to monitor the pressure changes at the target position; installing liquid level sensors upstream and downstream of the pump station unit, the purpose is to monitor the upstream and downstream water levels, and indirectly affect the head of the pump station unit operation. The data transmission submodule is used to build a matching model based on sensor data and sensor position, and obtain the collected pump station unit data based on the matching model. In the data transmission module establishment stage, a three-dimensional model that matches the sensor position and sensor data one by one is built according to the sensor installation position and LabVIEW programming software, in which the voltage sensor and the liquid level sensor use voltage output, and the pressure sensor uses current output. The data acquisition VI is used and the relevant parameters are set, including current and voltage ranges, physical channels, acquisition number, sampling rate, sampling time, etc.

[0082] like Figure 3 As shown, the pump station drive model construction module is used to process the pump station unit data, and build a pump station drive model based on the processed pump station unit data, and use the pump station drive mechanism to simulate the pump station unit data; the pump station drive model construction module includes a drive mechanism submodule and a data drive submodule; the drive mechanism submodule is used to construct a mechanism based on the fluid mechanics and internal environmental parameters in the operation of the pump station, and to simulate the mechanism according to the constructed mechanism; the data drive submodule is used to perform data preprocessing based on the drive mechanism submodule and the data acquisition module.

[0083] The driving mechanism submodule includes a conservation mechanism unit, an environmental mechanism unit and a simulation mechanism unit; the conservation mechanism unit is used to establish the conservation mechanism module of the pump station unit. Since only the internal flow of the pump station is analyzed, only the theoretical knowledge of fluid mechanics involved in the operation of the water pump is studied. The conservation mechanism module includes basic equations, kinematics and viscous flow, among which the basic equations include continuity equations, momentum equations, energy equations, Bernoulli equations, etc.; dynamics include micro-group motion, Reynolds transport, streamline coordinates, material derivatives, etc., viscous flow includes turbulent boundary layer, boundary layer separation, flow loss, boundary layer thickness, and at the same time, the pump station unit is composed of multiple water pumps in parallel, and the relationship between the parameters of its external characteristics is also included in the conservation mechanism.

[0084] The environmental mechanism unit is used to establish the environmental mechanism module of the pump station unit according to the different variable parameters of the pump station unit. The environmental mechanism mainly refers to the internal environmental parameters of the pump station unit during operation, including the type and number of grids, turbulence model, inlet and outlet parameters, and impeller blade angle. The grid types include tetrahedral grids and hexahedral grids, which are selected according to the degree of refinement and computational efficiency; the turbulence models are mainly divided into three categories, direct numerical simulation, Reynolds-averaged NS equations, and large eddy simulation; the inlet and outlet parameters are mainly divided into different combinations of pressure, pressure, and mass flow; the impeller blade angle is set according to the specific pump plant operation, which are 0°, ±2°, ±4°, and ±6°.

[0085] The simulation mechanism unit is used to establish the simulation mechanism model. In this stage, the simulation object is three-dimensionally modeled according to the actual simulation requirements, and the environmental parameters in the environmental mechanism are selected to form a separate simulation model. At the same time, fixed models are also provided for specific purposes, such as cavitation models and multiphase flow models.

[0086] like Figure 4 As shown, the data driving submodule includes: a data dimension reduction submodule, a data reconstruction submodule and a data updating submodule; the data dimension reduction submodule is used to simplify the data after the simulation of the driving mechanism submodule by using the method of intrinsic orthogonal decomposition, realize the dimension reduction of the non-steady-state flow field pressure data, and calculate the dimension reduction eigenvalue; the data dimension reduction submodule is also used to establish a data dimension reduction model according to the mechanism driving module, realize data dimension reduction, and retain the basis functions and time coefficients with a large energy proportion; the data reconstruction submodule is used to restore and reconstruct the flow field unit based on the dimension reduction eigenvalue; the data reconstruction submodule is also used to establish a data reconstruction module according to the data dimension reduction module, mainly including flow field data reconstruction, that is, the product of the basis function and the corresponding time coefficient; the data updating submodule is used to update the reconstructed data using the Kalman filtering algorithm. The data updating submodule is also used to establish a data updating module according to the basis function after data dimension reduction, so as to realize the real-time change of the flow field characteristics inside the pump station unit.

[0087] In the data dimension reduction stage, the data after the mechanism-driven model simulation is simplified. Due to the large number of grids, the number of variable data at the grid nodes is huge. The variables mainly depend on the research object, including velocity, pressure, entropy generation, vorticity, etc. If the intrinsic orthogonal decomposition method is used to reduce the dimension of the unsteady flow field pressure data, the pressure U(x,t i ) is recorded as a set of instantaneous pressure matrices. The pressures at different times constitute the instantaneous pressure matrix U(x, t), where x represents the position vector and t represents the time. The instantaneous pressure matrix can be expressed as the sum of the average pressure matrix U(x, t) and the pulsating pressure matrix U'(x, t), that is:

[0088]

[0089] Using the separation of variables method, the pulsating pressure matrix is ​​decomposed into two parts that are independent of each other in time and space, and a series of time-related coefficients a i (t) and a series of spatial orthogonal basis functions The product of:

[0090]

[0091] In the formula, is the POD space mode, also called POD basis; a i (t) is the modal coefficient corresponding to each mode, also known as the time coefficient.

[0092] In the data reconstruction stage, the data dimension reduction retains the data with a larger proportion and deletes some data with little impact on the overall situation. It is mainly manifested in sorting the eigenvalues ​​calculated in the above dimensionality reduction process, and the eigenvectors with large eigenvalues ​​are retained as reconstructed data. Based on this, the original flow field is restored to the greatest extent, and the calculation pressure is accelerated while retaining the most significant features of the flow field. The principle is as follows:

[0093]

[0094] In the formula represents the approximate deviation matrix of pulsating pressure, N is the total number of intercepted modes;

[0095] In order to ensure the accuracy of the modal truncation process, the selection of the first N modes should make the transient pressure matrix U'(x, t) and the approximate deviation matrix The error between is the smallest, and the second-order norm is used to construct the error expression, which satisfies the following formula:

[0096]

[0097] Where || || 2 Represents the second norm of a vector.

[0098] In order to ensure the accuracy of pressure field reconstruction, the difference between the pulsating pressure matrix and the approximate deviation matrix is ​​minimized; the error between the truncated intrinsic orthogonal mode and the pulsating pressure matrix can be expressed as:

[0099]

[0100] Bring in get:

[0101]

[0102] Let R=(U'U' T ), from linear algebra we know that R is a symmetric matrix. By constructing the Lagrangian function to find the extreme value, the error can be expressed as:

[0103]

[0104] where λ i is the eigenvalue of the real symmetric matrix R, is the eigenvector of the real symmetric matrix R.

[0105] In the data update stage, the Kalman filter algorithm is used to update the data obtained by data reconstruction. Since the Kalman filter algorithm mainly includes two steps: update and prediction, the main principle is as follows. For the operation of a specific pump station unit, the first few order basis functions of the intrinsic orthogonal decomposition and reconstruction are updated. First, the first few order basis functions are arranged in column form as column vectors, and the first few order basis functions of the intrinsic orthogonal decomposition and reconstruction are updated. The first few order basis functions are arranged in column form as column vectors. The evolution of the column vector over time can be expressed as follows:

[0106]

[0107] Where Φ l,k The first few order basis functions represented by can change over time; considering the possible system evolution, the evolution of the first few order function space over time can be expressed as:

[0108] Φ l,k =Φ l,k-1 +w k

[0109] Where w k is a zero-mean Gaussian white noise, whose time-invariant covariance is adjusted to obtain an unbiased estimate of the first few order basis functions; it is supplemented by an observation method and the first few order function spaces are updated according to the measured values:

[0110] y k,r =H k,r Φ l,k +v k,r

[0111] Where H k,r is a transformation matrix that combines the first few order basis function spaces with the observation equations, and updates the first few order basis functions of the flow field through the pressure response, v k,r represents the observation noise; since the change of the state vector is linear, the Kalman filter method is used to recursively update the basis function space; the initial value of the basis function space and its initial covariance are defined as and P0;

[0112] Prediction: Prior

[0113] Prior error covariance

[0114] In the formula represents the basis function state prediction error covariance matrix;

[0115] Correction: Kalman gain

[0116] Posterior Estimation

[0117] Update error covariance

[0118] In the formula is the updated state estimate, P k is the filtering (a posteriori) error covariance matrix.

[0119] The flow field identification model building module is used to preprocess the simulation data to obtain preprocessed data, and to build a digital twin degradation model based on the preprocessed data and the pump station driver model; Figure 2 The digital twin degradation model shown in Figure 5 The fast identification method shown in Figure 6 The pump station unit flow field rapid identification model based on digital twin shown in the figure specifically includes:

[0120] In the real-time flow field identification establishment stage, according to the established digital twin degradation model, the pressure simulation data is first calculated through the boundary parameter simulation of data acquisition, and the first few order basis functions with a high energy share are calculated through SVD using the intrinsic orthogonal decomposition. The basis function can represent different flow field characteristics. Using the prediction and correction steps in the above-mentioned Kalman filter algorithm, the initial value is set first. The Kalman filter algorithm automatically imports the basis function of the previous time step to calculate the basis function of the next time step, and determines whether there are non-steady-state flow field characteristics, such as wakes, cavitation, vortices of different scales, etc. If so, human intervention can be carried out immediately to identify the location and analyze its cause; if the flow field is normal, the next moment is updated.

[0121] The flow field identification module is used to identify the flow field of the pump station unit based on the digital twin degradation model. The simulation flow field identification establishment stage is similar to the real-time flow field identification. The difference is that the real-time flow field identification is the flow field identification result obtained by simulating the current environmental parameters. It is more real-time and practical, and can reflect the current state in real time. The simulation flow field identification is a preview of different environmental parameters. It can perceive the possible characteristics of the flow field of the pump station unit under different conditions in advance, and provide data basis for the later adjustment and efficient operation of the pump station unit. For example, during the flood season, the workload of the pump station unit is greater than usual and the speed is higher. In order to avoid the internal flow turbulence at that time, the corresponding mechanism driving model can be set in advance, and different degradation models can be constructed to obtain the flow field characteristics under this state.

[0122] Embodiment 2

[0123] A method for rapid identification of flow field of a pump station unit based on digital twin, the method comprising:

[0124] A sensor is installed on a pump station unit, and pump station unit data is collected according to the sensor; a method for installing a sensor on a pump station unit and collecting pump station unit data according to the sensor specifically includes: installing the sensor at a preset position of the pump station unit; building a matching model based on the sensor data and the sensor position, and obtaining the collected pump station unit data based on the matching model.

[0125] The pump station unit data is processed, and a pump station drive model is constructed based on the processed pump station unit data, and the pump station drive mechanism is used to simulate the pump station unit data; the use of the pump station drive mechanism to simulate the pump station unit data specifically includes: constructing a mechanism based on the fluid mechanics and internal environmental parameters in the operation of the pump station, and simulating the mechanism according to the constructed mechanism; preprocessing the data based on the drive mechanism submodule and the data acquisition module. The fluid mechanics theory involved in the operation of the pump station unit is used to construct the conservation mechanism of the pump station unit; the environmental mechanism of the pump station unit is constructed according to the variable parameters of the pump station unit; and the pump station simulation model is determined by combining the conservation mechanism and the environmental mechanism.

[0126] The data after simulation of the driving mechanism submodule is simplified by using the method of intrinsic orthogonal decomposition, so as to achieve dimensionality reduction of the non-steady-state flow field pressure data and calculate the dimensionality reduction eigenvalue; the flow field unit is restored and reconstructed based on the dimensionality reduction eigenvalue; and the reconstructed data is updated by using the Kalman filter algorithm. The data dimensionality reduction method specifically includes:

[0127] The pressure U(x,t i) is recorded as a set of instantaneous pressure matrices. The pressures at different times constitute the instantaneous pressure matrix U(x, t), where x represents the position vector and t represents the time. The instantaneous pressure matrix can be expressed as the sum of the average pressure matrix U(x, t) and the pulsating pressure matrix U'(x, t), that is:

[0128]

[0129] Using the separation of variables method, the pulsating pressure matrix is ​​decomposed into two parts that are independent of each other in time and space, and a series of time-related coefficients a i (t) and a series of spatial orthogonal basis functions The product of:

[0130]

[0131] In the formula, is the POD space mode, also called POD basis; a i (t) is the modal coefficient corresponding to each mode, also known as the time coefficient.

[0132] The data reconstruction methods specifically include:

[0133]

[0134] In the formula represents the approximate deviation matrix of pulsating pressure, and N is the total number of modes intercepted.

[0135] In order to ensure the accuracy of the modal truncation process, the selection of the first N modes should make the transient pressure matrix U'(x, t) and the approximate deviation matrix The error between is the smallest, and the second-order norm is used to construct the error expression, which satisfies the following formula:

[0136]

[0137] Where || || 2 Represents the second norm of a vector.

[0138] In order to ensure the accuracy of pressure field reconstruction, the difference between the pulsating pressure matrix and the approximate deviation matrix is ​​minimized; the error between the truncated intrinsic orthogonal mode and the pulsating pressure matrix can be expressed as:

[0139]

[0140] Bring in get:

[0141]

[0142] Let R=(U'U' T), from linear algebra we know that R is a symmetric matrix. By constructing the Lagrangian function to find the extreme value, the error can be expressed as:

[0143]

[0144] where λ i is the eigenvalue of the real symmetric matrix R, is the eigenvector of the real symmetric matrix R.

[0145] The data update methods specifically include:

[0146] Update the first few order basis functions of the intrinsic orthogonal decomposition and reconstruction, and arrange the first few order basis functions in column form as column vectors. The evolution of the column vectors over time can be expressed as follows:

[0147]

[0148] Where Φ l,k The first few order basis functions represented by can change over time; considering the possible system evolution, the evolution of the first few order function space over time can be expressed as:

[0149] Φ l,k =Φ l,k-1 +w k

[0150] Where w k is a zero-mean Gaussian white noise, whose time-invariant covariance is adjusted to obtain an unbiased estimate of the first few order basis functions; it is supplemented by an observation method and the first few order function spaces are updated according to the measured values:

[0151] y k,r =H k,r Φ l,k +v k,r

[0152] Where H k,r is a transformation matrix that combines the first few order basis function spaces with the observation equations, and updates the first few order basis functions of the flow field through the pressure response, v k,r represents the observation noise; since the change of the state vector is linear, the Kalman filter method is used to recursively update the basis function space; the initial value of the basis function space and its initial covariance are defined as and P0;

[0153] Prediction: Prior

[0154] Prior error covariance

[0155] In the formula represents the basis function state prediction error covariance matrix;

[0156] Correction: Kalman gain

[0157] Posterior Estimation

[0158] Update error covariance

[0159] In the formula is the updated state estimate, P k is the filtering (a posteriori) error covariance matrix.

[0160] Preprocessing is performed based on the simulation data to obtain preprocessed data, and a digital twin degradation model is constructed according to the preprocessed data and the pump station driver model; the digital twin degradation model performs flow field identification on the flow field of the pump station unit.

[0161] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A pump station unit flow field rapid identification system based on digital twin, characterized in that: The system comprises: a data acquisition module, a pump station drive model construction module, a flow field identification model construction module and a flow field identification module; The data acquisition module is used to install sensors on the pump station units and collect pump station unit data according to the sensors; The pump station drive model building module is used to process the pump station unit data, and build a pump station drive model based on the processed pump station unit data, and use the pump station drive mechanism to simulate the pump station unit data; The pump station drive model building module includes a drive mechanism submodule and a data drive submodule; The driving mechanism submodule is used to construct a mechanism based on the fluid mechanics and internal environmental parameters in the operation of the pump station, and to perform mechanism simulation according to the constructed mechanism; The data driving submodule is used for performing data preprocessing based on the driving mechanism submodule and the data acquisition module; The data driven submodule includes: a data dimension reduction submodule, a data reconstruction submodule and a data update submodule; The data dimension reduction submodule is used to simplify the data simulated by the driving mechanism submodule by using the method of intrinsic orthogonal decomposition, thereby achieving dimension reduction of the unsteady flow field pressure data and calculating the dimension reduction eigenvalue; The data reconstruction submodule is used to restore and reconstruct the flow field unit based on the dimension reduction eigenvalue; The data updating submodule is used to update the reconstructed data using a Kalman filter algorithm; The working process of the data dimension reduction submodule specifically includes: The pressure U(x,t i ) is recorded as a set of instantaneous pressure matrices. The pressures at different times constitute the instantaneous pressure matrix U(x,t), where x represents the position vector and t represents the time. The instantaneous pressure matrix can be expressed as the average pressure matrix And the sum of the pulsating pressure matrix U'(x,t): Using the separation of variables method, the pulsating pressure matrix is ​​decomposed into two parts that are independent of each other in time and space, and a series of time-related coefficients a i (t) and a series of spatial orthogonal basis functions The product of: In the formula, is the POD spatial mode; a i (t) is the modal coefficient corresponding to each mode; The flow field identification model building module is used to perform preprocessing based on simulation data to obtain preprocessed data, and to build a digital twin degradation model based on the preprocessed data and a pump station driver model; The flow field identification module is used to perform flow field identification on the flow field of the pump station unit based on the digital twin degradation model.

2. According to the digital twin-based pump station unit flow field rapid identification system according to claim 1, it is characterized in that: The data acquisition module includes a device installation submodule and a data transmission submodule; The equipment installation submodule is used to install the sensor at a preset position of the pump station unit; The data transmission submodule is used to build a matching model based on sensor data and sensor position, and obtain the collected pump station unit data based on the matching model.

3. According to the digital twin-based pump station unit flow field rapid identification system of claim 1, it is characterized in that: The driving mechanism submodule includes a conservation mechanism unit, an environmental mechanism unit and a simulation mechanism unit; The conservation mechanism unit is used to construct the conservation mechanism of the pump station unit based on the fluid mechanics theory involved in the operation of the pump station unit; The environmental mechanism unit is used to construct the environmental mechanism of the pump station unit according to the variable parameters of the pump station unit; The simulation mechanism unit is used to determine the pump station simulation model by combining the conservation mechanism and the environmental mechanism.

4. The pump station unit flow field rapid identification system based on digital twin according to claim 1 is characterized in that: The working process of the data reconstruction submodule specifically includes: In the formula, represents the approximate deviation matrix of pulsating pressure, N is the total number of intercepted modes; In order to ensure the accuracy of the modal truncation process, the selection of the first N modes should make the pulsating pressure matrix U'(x, t) and the approximate deviation matrix The error between them is the smallest, and the second-order norm is used to construct the error expression, which satisfies the following formula: Where || || 2 represents the second norm of a vector; In order to ensure the accuracy of pressure field reconstruction, the difference between the pulsating pressure matrix and the approximate deviation matrix is ​​minimized; the error between the truncated intrinsic orthogonal mode and the pulsating pressure matrix is ​​expressed as: Bring in get Let R=(U'U' T ), from linear algebra we know that R is a real symmetric matrix, by constructing the Lagrangian function to find the extreme value, the error is expressed as: where λ i is the eigenvalue of the real symmetric matrix R, is the eigenvector of the real symmetric matrix R.

5. A method for rapid identification of flow field of pumping station units based on digital twins, the method being used to implement the rapid identification system of flow field of pumping station units according to any one of claims 1 to 4, characterized in that: The method comprises: Install sensors on the pump station units and collect data on the pump station units based on the sensors; Process the pump station unit data, build a pump station drive model based on the processed pump station unit data, and use the pump station drive mechanism to simulate the pump station unit data; Preprocessing is performed based on the simulation data to obtain preprocessed data, and a digital twin degradation model is constructed based on the preprocessed data and the pump station driver model; The digital twin degradation model is used to identify the flow field of the pump station unit.

6. The method for rapid identification of flow field of pump station units based on digital twin according to claim 5 is characterized in that: The method of installing sensors on the pump station units and collecting data of the pump station units according to the sensors specifically includes: Install sensors at preset locations of pump station units; A matching model is constructed based on the sensor data and the sensor position, and the collected pump station unit data is obtained based on the matching model.

Images