Method and device for reducing order of bulk grain pneumatic conveying system model

Abstract

The invention relates to the field of bulk grain pneumatic conveying, and provides a bulk grain pneumatic conveying system model order reduction method and device in order to improve the accuracy of an order reduction model, when a function library is constructed, a traditional function is introduced, and a function for reflecting nonlinear and chaotic characteristics generated by interaction of particles and airflow is also introduced; during data processing, a dynamic sliding window is adopted for data collection, and when data changes drastically, the window is enlarged properly to obtain more comprehensive information; when the data change is small, the window is correspondingly reduced, and the real-time response capability is improved; model optimization is carried out based on chaos indexes, so that the model accuracy is improved; by regularly verifying the accuracy of the model, the model can continuously and accurately describe the dynamic characteristics of the system.

Description

Technical Field

[0001] The invention relates to the field of bulk grain pneumatic conveying, and in particular to a bulk grain pneumatic conveying system model order reduction method and device. Background Art

[0002] The pneumatic conveying system for bulk grains uses high-speed airflow to transport solid particles in pipelines, which has the advantages of non-contact, low dust and high efficiency. However, in actual operation, due to the complex interaction between particles and gas, collision between particles and the geometric structure of the pipeline, the system exhibits significant nonlinear, time-varying and chaotic characteristics, such as pressure fluctuations, local flow velocity mutations and uneven particle distribution. These phenomena not only lead to increased energy consumption, but may also cause problems such as particle breakage and pipeline blockage. Therefore, how to scientifically and accurately describe and predict the main dynamic behaviors of the pneumatic conveying system for bulk grains has become the key to improving the safety and efficiency of the system.

[0003] Although traditional high-order physical models can describe the system dynamics to a certain extent, they are complex in structure, computationally intensive, and difficult to meet real-time control requirements. Existing model reduction methods (such as eigenvalue decomposition and equilibrium truncation) mainly rely on linear assumptions, which do not adequately describe the inherent nonlinear and chaotic characteristics of the system, resulting in low model accuracy. In recent years, data-driven model identification technology, especially the use of sparse regression technology and its extended methods, has provided a new way to extract low-order models that describe the main dynamic behaviors of the system from a large number of candidate functions. However, when directly applied to bulk grain pneumatic conveying systems, it still faces problems such as candidate function library design, data noise, and chaotic characteristic capture. Summary of the invention

[0004] In order to improve the accuracy of a reduced-order model of a bulk grain pneumatic conveying system, the present invention provides a method and device for reducing the order of a bulk grain pneumatic conveying system model.

[0005] The technical solution adopted by the present invention to solve the above problems is: The model reduction method for bulk grain pneumatic conveying system includes: Step 1: Real-time collection of air flow pressure, pipeline pressure difference, flow velocity and particle charge in the bulk grain pneumatic conveying system; Step 2: Preprocess the real-time collected data, including filtering, normalization and time synchronization; Step 3: Create a function library, which contains traditional functions and functions for capturing nonlinear and chaotic behaviors; Step 4: Create a bulk grain pneumatic conveying system model based on the function library and preprocessed data.

[0006] Furthermore, traditional functions include polynomial functions, sine functions and cosine functions; functions used to capture nonlinear and chaotic behaviors include delayed embedding functions and exponential decay functions.

[0007] Furthermore, step 2 also includes: selecting the preprocessed data using a dynamic sliding window according to data changes.

[0008] Furthermore, step 3 also includes removing functions from the function library based on the preprocessed data.

[0009] Furthermore, step 4 also includes optimizing the bulk grain pneumatic conveying system model based on chaos indicators, where the chaos indicators are maximum Lyapunov exponent, approximate entropy and recurrence rate.

[0010] Furthermore, the method further includes step 5: regularly verifying the accuracy of the bulk grain pneumatic conveying system model, and if the accuracy is lower than the accuracy threshold, re-acquiring the model or re-acquiring the model after updating the function library.

[0011] The model reduction device for bulk grain pneumatic conveying system includes: Data acquisition unit: used to collect air flow pressure, pipeline pressure difference, flow rate and particle charge in bulk grain pneumatic conveying system in real time; Data processing unit: used to pre-process the data collected in real time; Function library unit: used to create a function library, which contains traditional functions and functions for capturing nonlinear and chaotic behaviors; Model building unit: Create a bulk grain pneumatic conveying system model based on the function library and preprocessed data.

[0012] Furthermore, it also includes a dynamic sliding window unit, which is used to select the pre-processed data using a dynamic sliding window according to the data change situation.

[0013] Furthermore, it also includes a function screening unit for eliminating functions in the function library based on the preprocessed data.

[0014] Furthermore, it also includes a model optimization unit for regularly verifying the accuracy of the bulk grain pneumatic conveying system model. If the accuracy is lower than the accuracy threshold, the model is retrained or the model is retrained after the function library is updated.

[0015] Compared with the prior art, the present invention has the following beneficial effects: when constructing a function library, in addition to introducing traditional functions, the present invention also introduces functions for reflecting the nonlinear and chaotic characteristics generated by the interaction between particles and airflow; when processing data, a dynamic sliding window is used for data collection, and when the data changes drastically, the window is appropriately enlarged to obtain more comprehensive information; when the data changes slightly, the window is correspondingly reduced to improve the real-time response capability; model optimization based on chaos indicators improves model accuracy; and by regularly verifying the model accuracy, it is ensured that the model can continue to accurately describe the dynamic characteristics of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is the flow chart of the order reduction method for the bulk grain pneumatic conveying system model; Figure 2 This is a schematic diagram of the structure of the reduced-order device for the bulk grain pneumatic conveying system model. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] like Figure 1 As shown in the figure, the bulk grain pneumatic conveying system model reduction method includes: Step 1: Real-time collection of air flow pressure, pipeline pressure difference, flow velocity and particle charge in the bulk grain pneumatic conveying system; Step 2: Preprocess the real-time collected data, including filtering, normalization, and time synchronization; Step 3: Create a function library, which contains traditional functions and functions for capturing nonlinear and chaotic behaviors; Step 4: Create a bulk grain pneumatic conveying system model based on the function library and preprocessed data.

[0019] In the actual operation of the bulk grain pneumatic conveying system, due to the influence of multiple factors such as the complex interaction between particles and gas, collisions between particles and the geometric structure of the pipeline, the system exhibits significant nonlinear, time-varying and chaotic characteristics. In order to effectively capture these characteristics and improve the accuracy of the bulk grain pneumatic conveying system model, the present invention, when creating a function library, not only sets traditional functions, such as polynomial functions, sine functions and cosine functions, but also sets functions for capturing nonlinear and chaotic behaviors, such as delayed embedding functions and exponential decay functions.

[0020] Since many candidate functions are set in the function library, in order to improve processing efficiency, the candidate functions can be preliminarily screened based on the collected data to retain the candidate functions with higher relevance.

[0021] Since there are often sensor noise and outliers in the data collection process of bulk grain pneumatic conveying systems, in order to improve the robustness of model parameter estimation, an anti-noise strategy can be adopted in the sparse regression process when creating a bulk grain pneumatic conveying system model: Based on the objective function, the Huber loss function is introduced, and the loss function is as follows: , in , is the prediction residual, , are the observed value of the i-th sample and the input vector of the i-th sample, respectively. is the parameter vector of the regression model (or prediction model), To create a bulk grain pneumatic conveying system model, is the preset threshold.

[0022] In addition, the phase space reconstruction method can be used to calculate chaos indicators such as the maximum Lyapunov exponent, approximate entropy and recurrence rate, and the model can be optimized based on the chaos indicators.

[0023] The data changes in the bulk grain pneumatic conveying system are not smooth, but vary significantly with the working conditions. If a unified time window is used for data collection, the dynamic characteristics may not be fully captured. Therefore, the present invention adaptively adjusts the window length according to the data changes (such as variance or spectral distribution).

[0024] Set the initial window length , and dynamically adjust the window size through the following formula : , in, represents the variance of the data in the window before time t, Indicates the time interval The data segments within the dynamic sliding window strategy are as follows: k is a tuning parameter (this parameter is given a reasonable initial value through offline tests or simulations, usually between 0.1 and 5.0, and then continuously observed and fine-tuned in real operation). When the data fluctuates greatly, the dynamic sliding window strategy increases the window length to capture more information; when the data is stable, the window is reduced to improve real-time performance; it ensures that the selected data is both representative and reduces the interference of redundant information.

[0025] Furthermore, it also includes step 5: regularly verifying the accuracy of the bulk grain pneumatic conveying system model, and if the accuracy is lower than the accuracy threshold, re-acquiring the model or re-acquiring the model after updating the function library; ensuring that the reduced-order model always accurately reflects the main dynamic characteristics of the bulk grain pneumatic conveying system.

[0026] Correspondingly, the present invention also provides a device for reducing the order of the model of a bulk grain pneumatic conveying system, such as Figure 2 As shown, including: Data acquisition unit: used to collect air flow pressure, pipeline pressure difference, flow rate and particle charge in bulk grain pneumatic conveying system in real time; Data processing unit: used to pre-process the data collected in real time; Function library unit: used to create a function library, which contains traditional functions and functions for capturing nonlinear and chaotic behaviors; Model building unit: Create a bulk grain pneumatic conveying system model based on the function library and preprocessed data.

[0027] Furthermore, it also includes a dynamic sliding window unit, which is used to select the preprocessed data using a dynamic sliding window according to the data changes; a function screening unit, which is used to eliminate the functions in the function library based on the preprocessed data; and a model optimization unit, which is used to regularly verify the accuracy of the bulk grain pneumatic conveying system model. If the accuracy is lower than the accuracy threshold, the model is retrained or the function library is updated and then the model is retrained.

Claims

1. The bulk grain pneumatic conveying system model reduction method is characterized by: include: Step 1: Real-time collection of air flow pressure, pipeline pressure difference, flow velocity and particle charge in the bulk grain pneumatic conveying system; Step 2: Preprocess the real-time collected data, including filtering, normalization and time synchronization; Step 3: Create a function library, which contains traditional functions and functions for capturing nonlinear and chaotic behaviors; Step 4: Create a bulk grain pneumatic conveying system model based on the function library and preprocessed data.

2. The bulk grain pneumatic conveying system model reduction method according to claim 1 is characterized in that: Traditional functions include polynomial functions, sine functions, and cosine functions; functions used to capture nonlinear and chaotic behaviors include time-delay embedded functions and exponential decay functions.

3. The bulk grain pneumatic conveying system model reduction method according to claim 1 is characterized in that: Step 2 also includes: selecting the pre-processed data using a dynamic sliding window according to data changes.

4. The bulk grain pneumatic conveying system model reduction method according to claim 1 is characterized in that: Step 3 also includes eliminating functions in the function library based on the preprocessed data.

5. The bulk grain pneumatic conveying system model reduction method according to claim 1 is characterized in that: Step 4 also includes optimizing the bulk grain pneumatic conveying system model based on chaos indicators, where the chaos indicators are maximum Lyapunov exponent, approximate entropy and recurrence rate.

6. The bulk grain pneumatic conveying system model reduction method according to any one of claims 1 to 5, characterized in that: The method also includes step 5: periodically verifying the accuracy of the bulk grain pneumatic conveying system model. If the accuracy is lower than the accuracy threshold, the model is re-acquired or the function library is updated and the model is re-acquired.

7. A device for reducing the order of the model of bulk grain pneumatic conveying system, characterized in that: include: Data acquisition unit: used to collect air flow pressure, pipeline pressure difference, flow rate and particle charge in bulk grain pneumatic conveying system in real time; Data processing unit: used to pre-process the data collected in real time; Function library unit: used to create a function library, which contains traditional functions and functions for capturing nonlinear and chaotic behaviors; Model building unit: Create a bulk grain pneumatic conveying system model based on the function library and preprocessed data.

8. The device for reducing the model of bulk grain pneumatic conveying system according to claim 7 is characterized in that: It also includes a dynamic sliding window unit, which is used to select the pre-processed data using a dynamic sliding window according to the data change situation.

9. The device for reducing the model of bulk grain pneumatic conveying system according to claim 7, characterized in that: It also includes a function screening unit for eliminating functions in the function library based on the preprocessed data.

10. The device for reducing the model of bulk grain pneumatic conveying system according to any one of claims 7 to 9, characterized in that: A model optimization unit is also included for periodically verifying the accuracy of the bulk grain pneumatic conveying system model. If the accuracy is lower than an accuracy threshold, the model is retrained or the model is retrained after the function library is updated.

Images