Dynamic numerical simulation and contour prediction method for overhead line system icing

Abstract

The invention provides a dynamic numerical simulation and contour prediction method for overhead line system icing, and relates to the field of high-speed railway overhead line system icing research in an atmospheric environment, and the method comprises the steps: analyzing overhead line system icing influence factors; on the basis of a heat balance equation, calculating icing key parameters of each collision particle, and constructing an overhead line system icing growth model; carrying out dynamic numerical simulation on the icing contour of the overhead line system on the basis of the dynamic grid and flow field analysis; pre-processing the icing contour of the overhead line system; determining an input variable, after dynamic numerical simulation of the overhead line system icing contour, constructing a training sample set and a test set by using the preprocessed overhead line system icing contour, and establishing an overhead line system icing contour prediction model by using a CNN-GRU network; and the icing contour of the overhead line system is adjusted in a self-adaptive manner. According to the method, the problem that the icing contour state of the contact network is difficult to monitor and predict is solved.

Description

Technical Field

[0001] The present invention relates to the field of research on icing of high-speed railway contact networks under atmospheric environment, and in particular to a dynamic numerical simulation and contour prediction method of icing of contact networks. Background Art

[0002] The special groove structure on the surface of the contact network leads to a high degree of asymmetry in the growth of its ice profile. On the other hand, due to the high stiffness of the high-speed railway contact network, the ice quality has little effect on its sag, so the ice profile becomes a direct factor affecting the sliding electrical contact. The pantograph current collection is sensitive to early contact network ice, and the limit of ice tolerance is small, but the contact network ice profile data faces the problems of difficult online monitoring, insufficient samples, and difficult prediction.

[0003] The prior art proposes an electrified railway overhead wire icing monitoring method based on atmospheric parameters of a freezing environment. This method optimizes the calculation method of the contact wire icing mass and density by using atmospheric icing parameters, and builds a single contact wire icing shape model in combination with the Comsol software. However, this method has the following disadvantages:

[0004] This technology treats the overhead contact network as a circular conductor and does not consider the impact of the heterogeneous structure of the overhead contact network on the icing process. In view of the unsteady change process of overhead contact icing, the use of inherent atmospheric environmental factors and functional models will inevitably lead to a decrease in the accuracy of icing monitoring.

[0005] A method for early warning of catenary icing is proposed in the prior art. This technology uses a neural network to construct an early warning model for catenary icing, in which the input layer mainly includes historical icing data, which solves the problems of poor icing early warning accuracy and weak anti-interference ability. However, it has the following disadvantages:

[0006] This technology is not sensitive to the characteristics of early icing, and the contact network collects current through sliding electrical contact. The icing contour of the contact network causes further deterioration of the bow-net current collection, but this icing warning method cannot accurately predict the icing contour.

[0007] In summary, the sliding electrical contact of the catenary is the only way for high-speed trains to obtain energy. Catenary icing further leads to the intensification of catenary impact vibration and frequent arcing, which seriously threatens the safety of high-speed rail energy supply. Therefore, based on the above background, there are currently problems such as unclear growth process of catenary icing profile, unclear rules, and difficulty in characterization. Summary of the invention

[0008] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for dynamic numerical simulation and contour prediction of contact network icing. Through the present invention, the evolution law of the contact network icing contour is explored, the problem of difficult monitoring and prediction of the contact network icing contour status is solved, and the safe and stable operation of the high-speed railway under icing conditions is ensured.

[0009] In order to achieve the above objectives, the technical solution adopted by the present invention is: a method for dynamic numerical simulation and profile prediction of catenary icing, comprising the following steps:

[0010] S1. Analyze the factors affecting the icing of the contact network;

[0011] S2. According to the analysis results, based on the heat balance equation, the key parameters of ice coating of each collision particle are calculated, and a catenary ice coating growth model is constructed;

[0012] S3. According to the catenary ice growth model, dynamic numerical simulation of the catenary ice profile is carried out based on dynamic grid and flow field analysis;

[0013] S4. Preprocessing the ice-covered profile of the contact network;

[0014] S5. Taking ambient temperature, wind speed, median diameter of water droplets and liquid water content as input variables, after dynamic numerical simulation of catenary icing profile, the training sample set and test set are constructed with the pre-processed catenary icing profile, and the CNN-GRU network is used to establish a catenary icing profile prediction model;

[0015] S6. Using the catenary icing profile prediction model, the catenary icing profile is adjusted in an adaptive manner to complete the prediction of the catenary icing profile.

[0016] Furthermore, the S2 is specifically:

[0017] According to the analysis results, the catenary icing is related to the collision, capture and freezing of water droplets in the air on the icing surface;

[0018] Based on the heat balance equation, the freezing rate, a key parameter of catenary icing, is obtained;

[0019] At wind speed v p and the surface length L of the micro-element control body on the ice-covered wire c Under the influence of , the characteristics of ice formation by water droplet content (LWC) in the air are obtained using the following formula:

[0020]

[0021] Where m represents the ice mass, t represents the ice time, α1, α2 and α3 represent the collision rate, capture rate and freezing rate of water droplets in the air respectively, and L represents the unit wire length;

[0022] According to the characteristics of the ice layer formed under different icing time steps, the icing quality is iteratively updated and a contact network icing growth model is constructed.

[0023] Furthermore, the S3 is specifically:

[0024] The script files of Gambit software were redeveloped using VB language, and the boundary and region were meshed using the adaptive advancing wave method.

[0025] In the flow field analysis, the co-located grid algorithm is used to solve the incompressible Reynolds average equation RANS and SST-kw turbulence model to obtain the air flow field;

[0026] According to the catenary ice growth model and flow field analysis results, the dynamic numerical simulation of the catenary ice profile is carried out in combination with the freezing coefficient, collision coefficient and water droplet velocity.

[0027] Furthermore, the S4 is specifically:

[0028] Taking the center point of the contact network as the circle, the ice covering the contact network is divided into sectors and the sector interval is selected;

[0029] Transform the ice contour points in different sector-shaped partitions into two-dimensional coordinates and polar coordinates;

[0030] An extreme point in different sector partitions is selected as the prediction sample object to complete the preprocessing of the catenary ice cover contour.

[0031] Furthermore, the S5 is specifically:

[0032] Taking ambient temperature, wind speed, median diameter of water droplets and liquid water content as input variables, the input variables are brought into the dynamic numerical simulation of catenary icing to obtain the test data of icing profile under different working conditions.

[0033] The ice profile test data under different working conditions are subjected to standard normalization processing;

[0034] The ice profile test data after standard normalization is divided into a training sample set and a test set;

[0035] The CNN-GRU network is used to establish a catenary icing contour prediction model, and the training sample set and test set are used to train and test the catenary icing contour prediction model to complete the construction of the catenary icing contour prediction model.

[0036] Furthermore, the S6 is specifically:

[0037] The contact network icing contour prediction model is used to take the icing environment parameters under different icing times as input, and the coordinate points of the icing contour as output. The contact network icing contour is predicted in an adaptive manner to complete the prediction of the contact network icing contour. The icing environment parameters include temperature, wind speed, median diameter of water droplets and liquid water content.

[0038] Beneficial effects of the present invention:

[0039] (1) The present invention proposes a method for dynamic simulation and contour prediction of icing on high-speed railway contact networks, which simulates the dynamic evolution of supercooled water droplets colliding, capturing and freezing on the surface of the contact network's special-shaped groove structure, reveals the evolution mechanism of ice growth on the contact network conductors, solves the problem that the mechanism of icing on the contact network is unclear and difficult to monitor, and realizes dynamic simulation and contour prediction of icing on the contact network under four environmental factors: wind speed, temperature, and liquid water content of the median diameter of water droplets.

[0040] (2) The present invention establishes a mechanism-data dual-driven contact network icing prediction model, realizes the early prediction of icing contours under multiple working conditions, solves the problem of difficult monitoring and prediction of contact network icing status, and lays a theoretical foundation for the safe and stable operation of high-speed railways under icing conditions.

[0041] (3) In view of the problems of unclear growth process, unclear rules and difficult characterization of the catenary icing profile, the present invention proposes a method for dynamic simulation of catenary icing, and builds a catenary icing profile prediction model in combination with a deep learning algorithm. The present invention explores the evolution law of the catenary icing profile, solves the problem of difficult monitoring and prediction of the catenary icing profile state, and ensures the safe and stable operation of high-speed railways under icing conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The figure is a flow chart of the method of the present invention.

[0043] Figure 2 Schematic diagram for calculation of collision coefficient of ice-covered surface of contact network.

[0044] Figure 3 Schematic diagram of dynamic numerical simulation of catenary icing.

[0045] Figure 4 Schematic diagram of catenary ice profile preprocessing. DETAILED DESCRIPTION

[0046] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0047] Example

[0048] like Figure 1 As shown, the present invention provides a method for dynamic numerical simulation and profile prediction of catenary icing, and the implementation method thereof is as follows:

[0049] S1. Analyze the factors affecting the icing of the contact network;

[0050] In this embodiment, the operating environment of the high-speed railway contact network is complex. On the one hand, with the development of icing, the icing environment has the characteristics of dynamic change; on the other hand, with the change of icing contour, the distribution of airflow characteristics in the micro-region near the icing conductor changes, and then the collision coefficient of the icing surface also changes, affecting the subsequent icing. According to the differences in the four types of icing, frost, rime, mixed rime, and sleet, the icing influencing factors are obtained: ambient temperature, wind speed, median diameter of water droplets, and liquid water content.

[0051] S2. According to the analysis results, based on the heat balance equation, the key parameters of ice coating of each collision particle are calculated, and the contact network ice coating growth model is constructed, which is as follows:

[0052] According to the analysis results, it is found that the icing of the contact network is related to the collision, capture, and freezing process of water droplets in the air on the ice-covered surface. The local collision rate of water droplets on the contact network surface is calculated using the following formula:

[0053]

[0054] Among them, Figure 2 As shown, dl represents the difference in the initial ordinate before the water droplet hits the wire, and dL represents the surface length between the upper and lower limits after the water droplet hits the wire;

[0055] Based on the heat balance equation, the freezing rate, a key parameter of catenary icing, is obtained;

[0056] At wind speed v p and the surface length L of the micro-element control body on the ice-covered wire c Under the influence of , the characteristics of ice formation by water droplet content LWC in the air are obtained using the following formula:

[0057]

[0058] Where m represents the ice mass, t represents the ice time, α1, α2 and α3 represent the collision rate, capture rate and freezing rate of water droplets in the air respectively, and L represents the unit wire length;

[0059] According to the characteristics of ice layers formed at different time steps, the ice quality is iteratively updated to build an ice growth model for the contact network.

[0060] In this embodiment, the heat exchange of the ice-covered contact network mainly includes convection, conduction, evaporation, and sublimation, and the heat caused by the damage to the ice-covered surface and the heat absorbed are balanced. Therefore, the freezing rate of the ice-covered contact network is obtained in combination with the thermodynamic characteristics.

[0061] S3. According to the catenary ice growth model, dynamic numerical simulation of the catenary ice profile is carried out based on dynamic grid and flow field analysis;

[0062] In this embodiment, Figure 3 As shown, the present invention uses grid generation and flow field analysis as the calculation basis, and according to the catenary icing growth model, realizes the dynamic numerical simulation of catenary icing under time-varying meteorological parameters, and the realization process is as follows Figure 3 As shown, the specific ones are:

[0063] The script files of Gambit software were redeveloped using VB language, and the boundary and region were meshed using the adaptive advancing wave method.

[0064] In the flow field analysis, the co-located grid algorithm is used to solve the incompressible Reynolds average equation RANS and SST-kw turbulence model to obtain the air flow field;

[0065] According to the catenary ice growth model and flow field analysis results, the dynamic numerical simulation of the catenary ice profile is carried out in combination with the freezing coefficient, collision coefficient and water droplet velocity.

[0066] In this embodiment, the present invention uses VB language to conduct secondary development of the script file of the mesh generation software Gambit, and adopts the adaptive propulsion wave method to mesh the boundary and the region respectively, wherein the contact network diameter is 14.4 mm, and the incoming flow parameters define the left side as the input boundary and the right side as the output boundary. In the flow field analysis, the same-grid SIMPLE algorithm is used to solve the Reynolds average equation of incompressible flow and the SST-kw turbulence model to obtain the air flow field. In the icing calculation, the freezing coefficient, collision coefficient, and water droplet velocity are combined to realize the dynamic simulation of the ice layer. In this embodiment, the turbulence model is an existing model and is not described here.

[0067] S4. Preprocessing the ice-covered profile of the contact network, specifically:

[0068] Taking the center point of the contact network as the circle, the ice covering the contact network is divided into sectors and the sector interval is selected;

[0069] Transform the ice contour points in different sector-shaped partitions into two-dimensional coordinates and polar coordinates;

[0070] An extreme point in different sector partitions is selected as the prediction sample object to complete the preprocessing of the catenary ice cover contour.

[0071] In this embodiment, in view of the differences in ice cover contours under different working conditions, a sector partitioning method is proposed in ice cover contour prediction preprocessing in order to mine the logical data features of ice cover contours. The method is shown as follows: Figure 4 As shown, the implementation is as follows:

[0072] The catenary ice is divided into sectors with the center point of the catenary as the center of the circle. The present invention selects 7.5° as the sector interval, and the number of polar coordinates is 24. When the output of the catenary ice profile prediction model is the ice profile, after sample training, the output data is the extreme value points in different sector areas, and the sampling points are curve fitted using the least squares method.

[0073] S5. According to the preprocessing results and the dynamic numerical simulation results, the catenary ice profile is predicted, which is specifically as follows:

[0074] Taking ambient temperature, wind speed, median diameter of water droplets and liquid water content as input variables, the input variables are brought into the dynamic numerical simulation of catenary icing to obtain the test data of icing profile under different working conditions.

[0075] The ice profile test data under different working conditions are subjected to standard normalization processing;

[0076] The ice profile test data after standard normalization is divided into a training sample set and a test set;

[0077] The CNN-GRU network is used to establish a catenary ice profile prediction model, and the training sample set and test set are used to train and test the catenary ice profile prediction model to complete the construction of the catenary ice profile prediction model.

[0078] S6. Using the catenary icing profile prediction model, the catenary icing profile is adjusted in an adaptive manner to complete the prediction of the catenary icing profile, which is specifically as follows:

[0079] The contact network icing contour prediction model is used to take the icing environment parameters under different icing times as input, and the coordinate points of the icing contour as output. The contact network icing contour is predicted in an adaptive manner to complete the prediction of the contact network icing contour. The icing environment parameters include temperature, wind speed, median diameter of water droplets and liquid water content.

[0080] In this embodiment, the catenary icing state has the characteristics of dynamic change according to the adjustment of the icing environment. In order to realize the adaptive prediction of the icing contour under different icing environments, the CNN-GRU network combined machine learning method is used to establish the catenary icing contour prediction model, which improves the generalization ability and adaptability of icing prediction.

[0081] In this embodiment, different icing environments are used as test objects and brought into the dynamic numerical simulation of catenary icing to obtain the icing profile test data under different working conditions. The output icing profile is preprocessed by standard normalization. The ratio of the training set to the test set in the CNN-GRU network parameters is 15:1. The catenary icing prediction model is used to adaptively adjust the catenary icing profile, realizing the rapid prediction of the catenary icing profile.

Claims

1. A method for dynamic numerical simulation and profile prediction of catenary icing, characterized in that: The following steps are involved: S1. Analyze the factors affecting the icing of the contact network; S2. According to the analysis results, based on the heat balance equation, the key parameters of ice coating of each collision particle are calculated, and a catenary ice coating growth model is constructed; S3. According to the catenary ice growth model, dynamic numerical simulation of the catenary ice profile is carried out based on dynamic grid and flow field analysis; S4. Preprocessing the ice-covered profile of the contact network; S5. Taking ambient temperature, wind speed, median diameter of water droplets and liquid water content as input variables, after dynamic numerical simulation of catenary icing profile, the training sample set and test set are constructed with the pre-processed catenary icing profile, and the CNN-GRU network is used to establish a catenary icing profile prediction model; S6. Using the catenary icing profile prediction model, the catenary icing profile is adjusted in an adaptive manner to complete the prediction of the catenary icing profile.

2. The method for dynamic numerical simulation and profile prediction of catenary icing according to claim 1, characterized in that: The S2 is specifically: According to the analysis results, it is found that the icing of the contact network is related to the collision, capture and freezing process of water droplets in the air on the icing surface. Based on the heat balance equation, the freezing rate of the overhead wire ice coating parameter is obtained; At wind speed v p and the surface length L of the micro-element control body on the ice-covered wire c Under the influence of , the characteristics of ice formation by water droplet content LWC in the air are obtained using the following formula: Where m represents the ice mass, t represents the ice time, α1, α2 and α3 represent the collision rate, capture rate and freezing rate of water droplets in the air respectively, and L represents the unit wire length; According to the characteristics of the ice layer formed under different icing time steps, the icing quality is iteratively updated and a contact network icing growth model is constructed.

3. The method for dynamic numerical simulation and profile prediction of catenary icing according to claim 1, characterized in that: The S3 is specifically: The script files of Gambit software were redeveloped using VB language, and the boundary and region were meshed using the adaptive advancing wave method. In the flow field analysis, the co-located grid algorithm is used to solve the incompressible Reynolds average equation RANS and SST-kw turbulence model to obtain the air flow field; According to the catenary ice growth model and flow field analysis results, the dynamic numerical simulation of the catenary ice profile is carried out in combination with the freezing coefficient, collision coefficient and water droplet velocity.

4. The method for dynamic numerical simulation and profile prediction of catenary icing according to claim 1, characterized in that: The S4 is specifically: Taking the center point of the contact network as the circle, the ice covering the contact network is divided into sectors and the sector interval is selected; Transform the ice contour points in different sector-shaped partitions into two-dimensional coordinates and polar coordinates; An extreme point in different sector partitions is selected as the prediction sample object to complete the preprocessing of the catenary ice cover contour.

5. The method for dynamic numerical simulation and profile prediction of catenary icing according to claim 1, characterized in that: The S5 is specifically: Taking ambient temperature, wind speed, median diameter of water droplets and liquid water content as input variables, the input variables are brought into the dynamic numerical simulation of catenary icing to obtain the test data of icing profile under different working conditions. The ice profile test data under different working conditions are subjected to standard normalization processing; The ice profile test data after standard normalization is divided into a training sample set and a test set; The CNN-GRU network is used to establish a catenary icing contour prediction model, and the training sample set and test set are used to train and test the catenary icing contour prediction model to complete the construction of the catenary icing contour prediction model.

6. The method for dynamic numerical simulation and profile prediction of catenary icing according to claim 1, characterized in that: The S6 is specifically: The contact network icing contour prediction model is used to take the icing environment parameters under different icing times as input, and the coordinate points of the icing contour as output. The contact network icing contour is predicted in an adaptive manner to complete the prediction of the contact network icing contour. The icing environment parameters include temperature, wind speed, median diameter of water droplets and liquid water content.

Images