An online monitoring system and method for flow state based on FPGA integrated reduced order model

Abstract

The application relates to a flow state online monitoring system and method based on an FPGA integrated reduced-order model, which comprises a sensing parameter measurement module, an FPGA development board and a visual display module connected in sequence, the FPGA development board is provided with a data receiving and filtering module, a reduced-order algorithm integration module and an output signal conversion module, the sensing parameter measurement module is used for collecting flow state physical parameter signals in real time; the data receiving and filtering module is used for filtering the flow state physical parameter signals; the reduced-order algorithm integration module is used for combining the flow state physical parameters with the reduced-order model, obtaining the physical field parameters of the full three-dimensional space flow state by solving the reduced-order model; and the output signal conversion module is used for converting the physical field parameters of the full three-dimensional space flow state into corresponding visualized data and transmitting the visualized data to the visual display module for monitoring result display. Compared with the prior art, the application can effectively improve the timeliness and accuracy of full three-dimensional space flow state monitoring.

Description

Technical Field

[0001] This invention relates to the field of flow state monitoring technology, and in particular to an online flow state monitoring system and method based on an FPGA-integrated reduced-order model. Background Technology

[0002] Fluid dynamics is prevalent in industrial production processes such as aerospace, automotive electronics, and marine engineering. Various flow parameters are directly related to the production requirements and operational performance of core components. Therefore, rapid and accurate monitoring of the three-dimensional flow state is of great significance for safe production operations.

[0003] However, the flow process is accompanied by complex physical phenomena such as heat transfer, mass transfer, and vibration, which severely limits the timeliness and accuracy of flow parameter sensing and monitoring, making it difficult to establish accurate mechanistic models to describe flow behavior. Traditional techniques for monitoring flow conditions mainly fall into two categories:

[0004] 1) Sensor methods based on experimental measurements: With the continuous advancement of multi-sensor measurement technology, a large number of parameters containing flow state information can be sensed by experimental instruments. Through signal conversion and data analysis, basic flow states such as temperature, pressure, and flow rate can be directly obtained. However, due to the susceptibility of the measurement area to obstruction and blockage by geometric structures, the spatial dimension of sensing measurement is very limited. Even when combined with advanced three-dimensional PIV (Particle Image Velocimetry) measurement technology, its measurement space and flow velocity range are still constrained by various hardware aspects, making it difficult to achieve online sensing of complex internal regions and full three-dimensional space.

[0005] 2) Data assimilation techniques that integrate experimental measurements with predictive models. Data assimilation uses actual measured sensor data as constraints to correct fluid dynamics models, thereby accurately monitoring flow states. However, high-precision numerical models are built on high-resolution grids, resulting in extremely high computational costs, which greatly limits the timeliness of flow state monitoring.

[0006] Currently, some research considers the high connectivity, low latency, and parallel computing and transmission characteristics of FPGAs (Field Programmable Gate Arrays) to make them the best platform for high-speed sensor data transmission, which can basically meet the real-time and synchronization requirements of monitoring systems. However, in the process of flow state monitoring, FPGAs are only used for simple data input processing and output control. With their high parallelism of multiple inputs and outputs (I / O), they can quickly analyze the parameters collected by sensors. However, existing numerical models are difficult to integrate into FPGAs due to their computational complexity. Therefore, in practical applications, high-speed sensing systems based on FPGAs cannot achieve online monitoring of flow state in full three-dimensional space. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the existing technology and provide an online flow state monitoring system and method based on an FPGA-integrated reduced-order model, which can effectively improve the timeliness and accuracy of flow state monitoring in three-dimensional space.

[0008] The objective of this invention can be achieved through the following technical solution: an online monitoring system for flow state based on an FPGA-integrated reduced-order model, comprising a sensing parameter measurement module, wherein the sensing parameter measurement module is connected to an FPGA development board, the FPGA development board is connected to a visualization display module, and the FPGA development board is provided with a data receiving and filtering module, a reduced-order algorithm integration module and an output signal conversion module, wherein the sensing parameter measurement module is used to acquire flow state physical parameter signals in real time;

[0009] The data receiving and filtering module is used to receive and filter the collected flow state physical parameter signals.

[0010] The order reduction algorithm integration module is used to combine the physical parameters of the flow state with the order reduction model, and output the physical field parameters of the flow state in the full three-dimensional space by solving the order reduction model.

[0011] The output signal conversion module is used to convert the physical field parameters of the full three-dimensional spatial flow state into corresponding visual data and transmit them to the visualization display module;

[0012] The visualization module is used to display the monitoring results of the flow state in full three-dimensional space online.

[0013] Furthermore, the sensing parameter measurement module includes, but is not limited to, temperature sensors, speed sensors, pressure sensors, and flow sensors.

[0014] Furthermore, the visualization display module is specifically a host computer.

[0015] A method for online monitoring of flow state based on an FPGA-integrated reduced-order model includes the following steps:

[0016] S1. Real-time acquisition of physical parameter signals of flow state;

[0017] S2. Filter the collected flow state physical parameter signals;

[0018] S3. Based on the filtered physical parameters of the flow state, a reduced-order model is constructed and solved to obtain the physical field parameters of the flow state in the full three-dimensional space.

[0019] S4. After converting the physical field parameters of the full three-dimensional spatial flow state into corresponding visual data, the data is transmitted to the visualization display module for display.

[0020] Furthermore, the physical parameter signals of the flow state in step S1 include, but are not limited to, temperature, velocity, pressure, and flow rate.

[0021] Furthermore, step S2 specifically involves using a Kalman filter to filter the acquired flow state physical parameter signals in order to eliminate noise signals.

[0022] Furthermore, step S3 specifically includes the following steps:

[0023] S31. Based on the physical parameters of the flow state after filtering, a numerical model simulating the flow state is established using computational fluid dynamics.

[0024] S32. Using a reduction algorithm, construct a corresponding reduced-order model based on the numerical model established in step S31;

[0025] S33. Calculate and solve the reduced-order model to obtain the physical field parameters of the flow state in the full three-dimensional space.

[0026] Furthermore, step S31 specifically involves using the filtered flow state physical parameters as boundary conditions for the fluid dynamics physical equations, and establishing a numerical model simulating the flow state using computational fluid dynamics.

[0027] Furthermore, the specific process of step S32 is as follows: a linear array structure is used to design the matrix multiplication and vector multiplication calculations in the dimension reduction projection in parallel, and the dimension reduction projection is applied to realize the dimension reduction processing of discrete variables, thereby carrying out dimension reduction projection on the basis of the numerical model and constructing a reduced-order model.

[0028] Furthermore, the specific process of calculating and solving the reduced-order model in step S33 is as follows: the conjugate gradient algorithm is used to calculate and solve the reduced-order model.

[0029] Compared with the prior art, the present invention has the following advantages:

[0030] I. This invention connects a sensing parameter measurement module to an FPGA development board and the FPGA development board to a visualization display module. The FPGA development board is equipped with a data receiving and filtering module, a reduction algorithm integration module, and an output signal conversion module. The data receiving and filtering module receives and filters the flow state physical parameter signals collected by the sensing parameter measurement module. The reduction algorithm integration module combines the flow state physical parameters with a reduction model, solves the reduction model, and outputs the physical field parameters of the flow state in three-dimensional space. The output signal conversion module converts the physical field parameters of the flow state in three-dimensional space into corresponding visual data and transmits it to the visualization display module for displaying the monitoring results. Therefore, based on FPGA, basic modules such as signal acquisition, signal processing, embedded model, online calculation, and status output are fully integrated, breaking the limitations between circuit hardware and algorithm software. It can further integrate reduced-order models based on local sensing information, thereby monitoring the physical state of locations that cannot be measured by sensors or even in the complete three-dimensional space. This breaks the limitations of sensing spatial dimension, effectively improves the comprehensiveness and accuracy of monitoring, and the reduced-order model embedded in FPGA solves the problem of computational complexity of traditional numerical models, greatly ensuring the timeliness of status monitoring.

[0031] Second, in this invention, a reduced-order model is constructed and solved based on the filtered physical parameters of the flow state, thereby obtaining the physical field parameters of the flow state in full three-dimensional space. On one hand, the physical parameters collected by sensors are used as boundary conditions for the fluid dynamics physical equations, and a numerical model simulating the flow state is established using computational fluid dynamics. On the other hand, a basic algorithm for matrix-vector operations in FPGA is established: a linear array structure is used to design the matrix multiplication and vector multiplication calculations in the dimensionality reduction projection in parallel, and the dimensionality reduction projection is applied to achieve dimensionality reduction of discrete variables. This ensures the accuracy of the constructed model while significantly reducing the computational complexity.

[0032] Third, this invention uses sensor measurement data as boundary parameters to establish a numerical model, which can flexibly select different numerical models according to the actual physical state. It can be applied to condition monitoring in multiple scenarios and is widely applicable to online monitoring of flow states such as heat transfer and mass transfer. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0034] Figure 2 This is a schematic diagram of the method flow of the present invention;

[0035] Figure 3 A schematic diagram illustrating matrix order reduction projection calculation for an FPGA development board;

[0036] Figure 4 A schematic diagram illustrating vector order reduction projection calculation for an FPGA development board;

[0037] Figure 5 A schematic diagram illustrating the solution of a system of equations on an FPGA development board;

[0038] The markings in the diagram are as follows: 1. Sensor parameter measurement module; 2. FPGA development board; 3. Visualization display module; 201. Data receiving and filtering module; 202. Order reduction algorithm integration module; 203. Output signal conversion module. Detailed Implementation

[0039] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0040] Example

[0041] like Figure 1 As shown, an online flow state monitoring system based on an FPGA-integrated reduced-order model includes a sensing parameter measurement module 1, which is connected to an FPGA development board 2. The FPGA development board 2 is connected to a visualization display module 3. The FPGA development board 2 is equipped with a data receiving and filtering module 201, a reduced-order algorithm integration module 202, and an output signal conversion module 203. The sensing parameter measurement module 1 is used to acquire flow state physical parameter signals in real time.

[0042] The data receiving and filtering module 201 is used to receive the collected flow state physical parameter signals and perform filtering processing;

[0043] The order reduction algorithm integration module 202 is used to combine the physical parameters of the flow state with the order reduction model, and output the physical field parameters of the flow state in the full three-dimensional space by solving the order reduction model.

[0044] The output signal conversion module 203 is used to convert the physical field parameters of the full three-dimensional spatial flow state into corresponding visual data and transmit them to the visualization display module;

[0045] The visualization module 3 is used to display the monitoring results of the full three-dimensional spatial flow status online.

[0046] In practical applications, the sensing parameter measurement module 1 includes, but is not limited to, temperature sensors, speed sensors, pressure sensors, and flow sensors, while the visualization display module 3 can be implemented using a remote host computer.

[0047] Applying the aforementioned online flow state monitoring system, a method for online flow state monitoring based on an FPGA-integrated reduced-order model is implemented, such as... Figure 2 As shown, it includes the following steps:

[0048] S1. Real-time acquisition of physical parameters of the flow state (including but not limited to temperature, velocity, pressure, and flow rate);

[0049] S2. Filter the collected flow state physical parameter signals;

[0050] S3. Based on the filtered physical parameters of the flow state, a reduced-order model is constructed and solved to obtain the physical field parameters of the flow state in the full three-dimensional space.

[0051] S4. After converting the physical field parameters of the full three-dimensional spatial flow state into corresponding visual data, the data is transmitted to the visualization display module for display.

[0052] This embodiment applies the above technical solution, and its main contents are as follows:

[0053] I. Establish a high-speed data sensing and acquisition system: Different physical parameters are acquired and amplified through various conventional sensors, and the processed signals are input into the FPGA for parameter signal conversion and storage.

[0054] 2. Filter the sensor signals acquired in step one: In order to eliminate various noise interferences in the sensor signals, signal filtering methods are needed to eliminate periodic and random interferences mixed in the sensor signals.

[0055] 3. Establish a numerical model with sensor data as boundary parameters: Use the physical parameters collected by the sensor as the boundary conditions of the fluid dynamics physical equations, and use computational fluid dynamics to establish a numerical model to simulate the flow state.

[0056] IV. FPGA Programming Integration of Dimensional Reduction Model: Establish basic algorithms for matrix and vector operations in FPGA, adopt a linear array structure for parallel design of matrix and vector multiplication calculations in dimensionality reduction projection, and apply dimensionality reduction projection to realize the dimensionality reduction of discrete variables, significantly reducing the complexity of equation system calculations in FPGA.

[0057] V. Establishing the basic algorithm for calculating equations in FPGA: The conjugate gradient algorithm is used to calculate and solve the equations in FPGA, and the result is various parameter data of the flow state in the full three-dimensional space.

[0058] 6. The various physical parameters calculated in step 5 are output to the host computer via the communication bus for display, thereby realizing full three-dimensional monitoring of each state during the flow process.

[0059] This embodiment focuses on the flow state in a high-temperature steam pipeline, conducting online monitoring of the full three-dimensional flow state of high-temperature steam inside the pipeline. The hardware includes a sensor parameter measurement module, a data receiving and filtering module, a reduction-order algorithm integration module, an output signal conversion module, and a remote visualization display module. The sensor parameter measurement module contains sensors for collecting various physical parameters at the pipeline's inlet and outlet locations, such as temperature, pressure, and flow rate sensors for collecting parameters like temperature, pressure, and flow rate of the high-temperature steam at these locations. In this embodiment, a high-temperature resistant thermocouple temperature sensor is selected, with a measurement range of 0-500℃, a measurement accuracy of ±2℃, a power supply voltage of 24DCV, and an output voltage of 0-5V. The pressure sensor used is a high-temperature resistant steam pressure transmitter, with a measurement range of 0-100MPa, a measurement accuracy of 1%, a power supply voltage of 24DCV, and an output voltage signal of 0-5V. The flow sensor used is a turbine flow meter, with a measurement accuracy of 1% and a measurement range of 0-200m³. 3 / h, the power supply voltage is 24DCV, and the output signal is 4-20mA.

[0060] The data receiving and filtering module receives sensor signals and performs data filtering on the data signals collected by various sensors. The data filtering algorithm is pre-coded in the FPGA board. In this embodiment, the data filtering algorithm uses the Kalman filter method to filter the temperature, pressure, and flow signals collected by various sensors to eliminate noise signals. Taking the temperature signal as an example, the filtering equation is:

[0061] X(k)=AX9k-1)+TW(k-1)+ω(k-1)

[0062] Z(k)=HX(k)+V(k)

[0063] Where X(k) is the temperature at the current time t=k, and X(k-1) is the temperature at the previous time t=k-1; A is the state transition matrix; W is the system input; T is the input matrix; ω is the system noise. Z(k) is the parameter measured by the sensor at the current time; H is the output matrix; and V is the measurement noise.

[0064] The order reduction algorithm integration module integrates the order reduction algorithm of computational fluid dynamics for the full three-dimensional physical field into the FPGA board, and establishes a full three-dimensional spatial computational simulation of the pipe flow state using sensor data parameters as boundary inputs. In this embodiment, the computational fluid dynamics control equations for the pipe flow state include the continuity equation, the momentum conservation equation, and the energy conservation equation, as shown below:

[0065]

[0066]

[0067]

[0068]

[0069]

[0070] where ρ, c p and k are the fluid density, specific heat capacity, and thermal conductivity, respectively; v is the fluid velocity; τ is the shear stress; g is the acceleration due to gravity; T is the temperature; G k represents the turbulent kinetic energy generation term; G ω represents the ω generation term; Γ ω and Γ k represent the effective diffusion coefficients of k and ω; Y k and Y ω represent the dissipation terms of k and ω; E represents the total energy; k eff represents the effective thermal conductivity; (τ ij ) eff represents viscous heating.

[0071] The boundary conditions of the physical equations are obtained by measuring the sensors at the inlet and outlet of the pipeline. The measured steam temperature and pressure are used as the inlet conditions, and the measured flow parameters are used as the outlet boundary conditions. After introducing the spatial discretization format and coupling the pressure-velocity algorithm, the parameter solution equation can be written in the following form:

[0072] Jx = R

[0073] where J (m×m) is the system matrix assembled by the CFD discretization format, m is the degree of freedom of the spatial discretization format; x (m×1) is the physical parameter in the full three-dimensional space; R (m×1) is the residual.

[0074] The integrated reduced-order model performs dimensionality reduction projection based on the above equation, that is, converting the large-scale matrix calculation in the above equation into a small-matrix calculation. As Figure 3 and Figure 4 shown, the dimensionality reduction projection of the matrix and vector is implemented in the FPGA by adopting array calculations, as shown below:

[0075]

[0076]

[0077] where, and are the matrix and vector after dimensionality reduction, respectively, and r << m realizes the reduction of the degree of freedom of the global calculation of physical parameters. Φ (m×r) is the projection matrix, which can be obtained by the proper orthogonal decomposition method. As Figure 5As shown, the solution obtained by solving the above low-dimensional equations using the conjugate gradient method can be transformed into various parameters of the flow state in full three-dimensional space through spatial back projection, as shown below:

[0078]

[0079] The output signal conversion module converts the physical field parameters calculated by the FPGA into visual data signals that can be transmitted quickly, so as to send them to the host computer at the remote end. Finally, the full three-dimensional physical state is displayed online in the visualization module at the remote end.

[0080] In practical applications, FPGA boards in electronic devices integrate read-only memory (ROM) and random access memory (RAM), which can perform various appropriate actions and processes according to program instructions pre-stored in the ROM. The RAM stores the physical data transmitted by the sensors and various calculation formats and model algorithms required for model building. The FPGA, ROM, and RAM are interconnected via a bus, and the input / output (I / O) interfaces on the board are also connected to the bus.

[0081] Other devices in this solution are connected to the FPGA's I / O interface, including: input units such as temperature, pressure, and flow sensors; output units such as displays, speakers, and host computers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication units allow devices to exchange information / data with other devices via computer networks such as the Internet and / or various communication lines.

[0082] The processing unit is used to execute the various methods and data processing steps described above. For example, in some embodiments, steps two through five may be pre-integrated into the FPGA, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the FPGA, one or more steps two through five described above may be performed. Alternatively, in other embodiments, the FPGA may be configured to perform steps two through five described above by any other suitable means (e.g., by means of firmware).

[0083] In summary, this technical solution considers that the reduced-order model is a low-dimensional approximation model proposed relative to the traditional numerical model. It relies on data analysis and physical feature extraction to construct a low-order model that approximates the behavior of the original high-order model, capturing the dominant modes and physical essence of the flow state and describing the flow state in full three-dimensional space. However, most current reduced-order models are integrated on PCs and run on CPUs, whose computational efficiency is far lower than that of FPGAs, hindering the integration of sensor signals and the realization of online flow state monitoring. Therefore, this invention proposes an online flow state monitoring scheme based on an FPGA-integrated reduced-order model. First, the analog and digital signals monitored by traditional sensors are transmitted to the FPGA for parameter acquisition and processing. Then, a physical reduced-order model is designed for the FPGA and integrated into the FPGA chip. Combining the sensor parameters and the reduced-order model, rapid simulation of the full physical field is achieved. Finally, the monitoring parameters at key locations are transmitted to a host computer for remote display of the physical state, thereby effectively improving the timeliness and accuracy of full three-dimensional flow state monitoring.

Claims

1. A method for online monitoring of flow state, applied to an online monitoring system for flow state based on an FPGA-integrated reduced-order model, characterized in that, The online flow state monitoring system includes a sensor parameter measurement module (1), which is connected to an FPGA development board (2). The FPGA development board (2) is connected to a visualization display module (3). The FPGA development board (2) is equipped with a data receiving and filtering module (201), a reduction algorithm integration module (202), and an output signal conversion module (203). The sensor parameter measurement module (1) is used to collect flow state physical parameter signals in real time. The data receiving and filtering module (201) is used to receive the collected flow state physical parameter signals and perform filtering processing; The order reduction algorithm integration module (202) is used to combine the physical parameters of the flow state with the order reduction model, and output the physical field parameters of the flow state in the full three-dimensional space by solving the order reduction model; The output signal conversion module (203) is used to convert the physical field parameters of the full three-dimensional spatial flow state into corresponding visualization data and transmit them to the visualization display module (3). The visualization module (3) is used to display the monitoring results of the full three-dimensional spatial flow state online; The above-mentioned online monitoring method for flow status includes the following steps: S1. Real-time acquisition of physical parameter signals of flow state; S2. Filter the collected flow state physical parameter signals; S3. Based on the filtered physical parameters of the flow state, a reduced-order model is constructed and solved to obtain the physical field parameters of the flow state in the full three-dimensional space. S4. After converting the physical field parameters of the full three-dimensional spatial flow state into corresponding visualization data, the data is transmitted to the visualization display module for display. Step S3 specifically includes the following steps: S31. Based on the physical parameters of the flow state after filtering, a numerical model simulating the flow state is established using computational fluid dynamics. S32. Using a reduction algorithm, construct a corresponding reduced-order model based on the numerical model established in step S31; S33. Calculate and solve the reduced-order model to obtain the physical field parameters of the flow state in the full three-dimensional space. Step S31 specifically involves using the filtered flow state physical parameters as boundary conditions for the fluid dynamics physical equations, and establishing a numerical model simulating the flow state using computational fluid dynamics. The specific process of step S32 is as follows: a linear array structure is used to design the matrix multiplication and vector multiplication calculations in the dimension reduction projection in parallel, and the dimension reduction projection is applied to realize the dimension reduction processing of discrete variables, so as to carry out dimension reduction projection on the basis of numerical model and construct a reduced-order model.

2. The method for online monitoring of flow state according to claim 1, characterized in that, The sensing parameter measurement module (1) includes a temperature sensor, a speed sensor, a pressure sensor, and a flow sensor.

3. The method for online monitoring of flow state according to claim 1, characterized in that, The visualization display module (3) is specifically a host computer.

4. The method for online monitoring of flow state according to claim 1, characterized in that, The physical parameter signals of the flow state in step S1 include temperature, velocity, pressure, and flow rate.

5. The method for online monitoring of flow state according to claim 1, characterized in that, Specifically, step S2 involves using a Kalman filter to filter the acquired flow state physical parameter signals in order to eliminate noise signals.

6. The method for online monitoring of flow state according to claim 1, characterized in that, The specific process of calculating and solving the reduced-order model in step S33 is as follows: the conjugate gradient algorithm is used to calculate and solve the reduced-order model.

Images