Methods, equipment, and computer programs for simulating runoff in plain watersheds with strong artificial regulation

By coupling the PHNS hydrological model with various machine learning methods, and utilizing residual sequence correction and manual parameter adjustment, the problem of nonlinear simulation of hydrological processes in polder areas was solved, achieving high-precision runoff simulation and multi-scenario analysis, and improving the model's generalization ability and interpretability.

CN123069918APending Publication Date: 2026-11-10NANJING INST OF GEOGRAPHY & LIMNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611510947.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-28
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing technologies lack a complete mechanism-data coupling modeling framework adapted to plain river network polder areas, which cannot effectively solve the nonlinearity and abrupt changes of hydrological processes in polder areas. Furthermore, machine learning models suffer from insufficient generalization and poor black-box interpretability in runoff simulation.

Method used

A hydrological cycle simulation framework coupling the PHNS hydrological model with multiple machine learning methods is constructed. Through residual sequence correction, the machine learning model is trained using climate characteristics and artificially controlled parameters to correct the simulation results of the PHNS model, thereby achieving high-precision runoff simulation and conducting multi-scenario analysis.

Benefits of technology

It significantly improves the accuracy of rainfall-runoff simulation in plain watersheds, enabling quantitative assessment of the impact of human activities and climate change on hydrological processes in polder areas, and providing a scientific basis for water resource management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123069918A_ABST
    Figure CN123069918A_ABST
Patent Text Reader

Abstract

This invention relates to a method, equipment, and computer program for simulating runoff in heavily artificially regulated plain watersheds. It utilizes the PHNS hydrological model to simulate runoff processes in polder areas, obtaining a simulated flow time series. A residual sequence is constructed between the simulated and measured flow time series. This residual sequence is used as the output variable of a machine learning model, with climate characteristic parameters, simulated flow, and artificial regulation parameters representing the operational status of artificial water management facilities at each time point in the residual sequence as input variables. The machine learning model is then trained to obtain an error prediction model. The machine learning model establishes a nonlinear mapping relationship between the input and output variables. The hydrological model is used to simulate runoff processes in polder areas, and the error prediction model is used to correct the simulation results during the prediction process to obtain the final simulation results. This invention can be applied to heavily artificially regulated plain river network areas, significantly improving the accuracy of rainfall-runoff simulation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of hydrological simulation and water resources management technology, specifically involving a method, equipment, and computer program for simulating runoff in plain watersheds with strong artificial regulation. Background Technology

[0002] A polder is a low-lying, flat area protected by dikes and regulated by artificial irrigation and drainage systems. It is typically located in areas with dense rivers and lakes, flat terrain, and high susceptibility to flooding. The hydrological system of a polder is primarily controlled by artificial mechanisms, with natural processes and engineering measures highly coupled. Hydrological processes encompass precipitation infiltration, surface runoff, river network regulation, and irrigation and drainage scheduling. In addition to climatic factors such as precipitation and potential evapotranspiration, human activities such as pumping stations, irrigation, and water level management significantly alter the spatial and temporal distribution and transport paths of water, making them far more complex than natural watersheds.

[0003] Previous parametric hydrological models (HBV, GR, etc.) were mostly built based on slope runoff mechanisms, suitable for mountainous and hilly areas, but they face significant simulation problems in flat plains and polder areas with strong human intervention (Napiorkowski et al., 2022). To address this, Yan et al. (2021) localized and improved the WALRUS model, embedding artificial control mechanisms such as pump stations, to construct the polder hydrological and nitrogen simulation system PHNS. However, the polder scheduling behavior brings strong nonlinearity and abrupt changes to the hydrological process, making it difficult to obtain control parameters. PHNS still has simulation biases, and its accuracy needs improvement. Machine learning (RF, LSTM, SVR, XGBoost, GPR) has strong nonlinear fitting capabilities and can uncover complex correlations in multi-source and multi-process hydrological data, and is widely used in runoff simulation. However, it generally suffers from insufficient generalization and poor black-box interpretability. Coupled with hydrological models and machine learning, the interpretability of hydrological models and the flexibility and intelligence of machine learning models can be fully utilized to improve the level of refined representation of polder hydrological processes.

[0004] Existing technologies lack a complete mechanism-data coupling modeling framework adapted to plain river network polder areas, and have not formed an effective set of solutions for machine learning input feature selection, model optimization, and multi-scenario hydrological response analysis. Summary of the Invention

[0005] The purpose of this invention is to provide a method, equipment, and computer program for simulating runoff in plain watersheds with strong artificial regulation. By constructing a hydrological cycle simulation framework that couples the PHNS hydrological model with multiple machine learning methods, it achieves high-precision simulation of hydrological processes in polder areas and systematically analyzes hydrological response changes under human activities and climate change scenarios.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for simulating runoff in a plain watershed with strong artificial regulation, the method comprising:

[0008] The PHNS hydrological model was used to simulate the runoff process in the polder area, and the time series of the simulated flow was obtained.

[0009] Construct the residual sequence between the simulated flow time series and the measured flow time series of the hydrological model;

[0010] The residual sequence is used as the output variable of the machine learning model, and the climate characteristic parameters, simulated flow, and artificial control parameters characterizing the operation status of artificial water management facilities at each time corresponding to each residual in the residual sequence are used as input variables to train the machine learning model and obtain an error prediction model; wherein, the machine learning model establishes a nonlinear mapping relationship between the input variables and the output variables.

[0011] The hydrological model is used to simulate the runoff process in the polder area. During the prediction process, the error prediction model is used to correct the simulation results and obtain the final simulation results.

[0012] In some embodiments of the present invention, the climate characteristic parameters are precipitation, temperature and potential evapotranspiration; the simulated flow includes the simulated flow at the current moment and the simulated flow at the previous moment.

[0013] In some embodiments of the present invention, the artificially controlled parameters are selected from the lower limit of suitable water depth in paddy fields hQmin1, the upper limit of suitable water depth in paddy fields hQmax1, the upper limit of water depth tolerating flooding in rice hQflood1, and the starting water level for surface water drainage. and surface water drainage stop level One or more of them.

[0014] In some embodiments of the present invention, the artificially controlled parameters are the lower limit of suitable water depth in paddy fields hQmin1, the upper limit of water depth tolerating flooding in rice hQflood1, and the starting water level for surface water drainage. .

[0015] In some embodiments of the present invention, the machine learning model is selected as a Gaussian process regression model.

[0016] In some embodiments of the present invention, obtaining the final simulation results includes:

[0017] The predicted residuals are obtained by using a trained Gaussian process regression model to predict the residuals.

[0018] The corrected flow rate is obtained by adding the simulated flow rate of the PHNS hydrological model to the predicted residual.

[0019] In some embodiments of the present invention, the method further includes a scenario simulation analysis step:

[0020] By adjusting the values ​​of the artificial control parameters or removing the preset artificial control parameters, different human activity scenarios or climate change scenarios can be constructed.

[0021] Input data corresponding to different scenarios are input into the hydrological model for simulation, and the simulation results are corrected using the error correction model to obtain the final simulation results under different scenarios, so as to assess the impact of human activities or climate change on the runoff process of the polder area.

[0022] The present invention further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0023] The present invention further provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0024] The present invention further provides a computer program product, including a computer program / instruction that, when executed by a processor, implements the steps of the above-described method.

[0025] The present invention has the following beneficial effects:

[0026] (1) This invention constructs a hydrological simulation framework that couples the PHNS model with multiple machine learning methods. It uses machine learning to fit the deviation of PHNS under strong artificial regulation. While maintaining the interpretability of the physical model, it significantly improves the accuracy of rainfall-runoff simulation in plain watersheds under strong artificial regulation. Experiments show that compared with the original PHNS model, the model evaluation indicators during the coupling model period and the validation period are significantly improved to varying degrees.

[0027] (2) This invention introduces key artificial control parameters characterizing the operation status of artificial water management facilities into the machine learning residual correction process, and determines the optimal input combination by comparing multiple sets of input feature schemes. Experimental results show that after removing redundant parameters (upper limit of suitable water depth in paddy fields and surface water drainage stop level), the overall performance of the model during the calibration and validation periods is better than the scheme containing all parameters, indicating that screening key control parameters can reduce redundant information and improve the accuracy of runoff simulation under strong artificial control conditions.

[0028] (3) Based on the optimal coupling model, this invention conducts multi-scenario analysis, which can quantitatively assess the impact of human activities such as irrigation and pumping station operation, as well as future climate change on the hydrological processes of the polder area, and provide a scientific basis for water resource management such as flood control, drainage, and irrigation planning in the plain river network polder area. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention.

[0030] Figure 2 This is a comparison chart of the performance indicators of the PHNS model and the PHNS-ML coupled model during the calibration and validation periods.

[0031] Figure 3 This is a comparison chart of the simulation results of the PHNS model and five PHNS-ML coupled models (ML input parameter is scheme E) during the calibration period.

[0032] Figure 4 This is a comparison chart of the simulation results of the PHNS model and five PHNS-ML coupled models (ML input parameter is scheme E) during the validation period.

[0033] Figure 5 This is a comparison chart of simulation curves from a hydrological model-machine learning coupling framework under scenarios with and without irrigation, and with and without pumps. Detailed Implementation

[0034] To facilitate a clearer understanding of this technical solution by those skilled in the art, the method of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0035] Example 1

[0036] This embodiment takes Dadangwei, a typical agricultural polder area in the Taihu Lake Basin, as an example to specifically illustrate the implementation of the method of the present invention. The steps are as follows:

[0037] S100 polder area dataset construction

[0038] Data time range: 2020-2023: meteorological data are taken from regional national-level rainfall, temperature and evaporation stations; runoff data is the sum of measured flow rates of pump stations P1 and P2; land use is based on the interpretation of Gaofen-1 remote sensing images, unified WGS84 coordinate system, and spatial processing is completed by ArcGIS.

[0039] Future climate data: CMIP6 MRI-ESM2-0 model SSP2-4.5 scenario, extracting daily precipitation, temperature, and potential evapotranspiration for 2030-2035 and 2040-2045, for subsequent climate scenario simulations.

[0040] S200 PHNS hydrological model simulation

[0041] PHNS (Polder Hydrology and Nitrogen Modeling System) is a hydrological-water quality coupled simulation system developed for plain polder areas. It was proposed by Yan et al., based on the WALRUS model. The model, with the WALRUS-paddy hydrological module at its core, effectively characterizes the hydrological response of polder areas under the combined effects of natural rainfall and strong human activities such as pumping stations and irrigation by coupling water balance and artificial regulation processes. PHNS consists of a water balance module, a farmland nitrogen balance module, and a surface water nitrogen balance module. It achieves integrated simulation of hydrological and nitrogen processes by coupling the WALRUS-paddy, MUSLE, and INCA models. The PHNS model has been validated and applied in typical plain polder areas such as the Taihu Lake Basin. For its detailed structure and calculation formulas, please refer to the literature (Yan et al., 2017).

[0042] The basic form of the water balance module in the PHNS hydrological model is as follows:

[0043]

[0044] In the formula, S represents the total water storage capacity of the polder system; P is the precipitation input; ET is the actual evapotranspiration; and Q is the total water storage capacity of the polder system. out With Q in These represent the amount of water discharged from the polder area and the amount of artificial irrigation, respectively.

[0045] The PHNS model has been disclosed in the inventors' previous article (Yan, R., Brunner P., Gao*, J., 2021. Simulation of nitrogen dynamics in lowland polders using a new coupled modelling approach: Insights into management. Journal of Cleaner Production, 313: 127753), and will not be repeated here.

[0046] S300 Machine Learning Bias Correction Framework Construction and Parameter Tuning:

[0047] To address the residuals of the PHNS systematic simulation, five model residual correction frameworks (RF, LSTM, SVR, XGBoost, and GPR) are established. These frameworks uniformly use the residuals as output and climate driving factors, PHNS model simulated flow rates, and artificially controlled parameters as inputs to construct a machine learning residual prediction model. The climate driving factors include precipitation, air temperature, and potential evapotranspiration; the PHNS model simulated flow rates include the simulated flow rates at the current and previous times; and the artificially controlled parameters include the lower limit of suitable paddy field water depth (hQmin1), the upper limit of suitable paddy field water depth (hQmax1), the upper limit of paddy field flood tolerance water depth (hQflood1), and the surface water drainage start-up level (hQmin1). ) and surface water drainage stop level ( All models underwent standardized feature preprocessing and 5-fold cross-validation optimization. The dataset was divided into a 7:3 time series period and a validation period. The independent parameter settings for each model are as follows:

[0048] Random Forest (RF) parameter tuning: A nonparametric supervised learning method based on ensemble learning, constructed from multiple decision trees through a bootstrap sampling approach. Key parameter settings are as follows:

[0049] ntree (number of decision trees): 300

[0050] mtry (number of random features per tree): 2

[0051] nodesize (minimum number of samples per leaf node): 30

[0052] maxnodes (maximum number of nodes in the tree): 50

[0053] Long Short-Term Memory (LSTM) Network Parameter Tuning: An improved recurrent neural network, which achieves temporal information memorization through a gating mechanism, with key parameter settings as follows:

[0054] Units (number of neurons in the LSTM hidden layer): Set to 20;

[0055] Dropout (random dropout rate): Set to 0.2 to prevent overfitting;

[0056] batch_size (batch size): Set to 16;

[0057] epochs (number of training rounds): set to 200, using the Adam optimizer, and with the EarlyStopping (patience=15) early stopping strategy to prevent overfitting;

[0058] Support Vector Regression (SVR) Parameter Tuning: A kernel method based on statistical learning theory, with key parameter settings as follows:

[0059] type: Use ε-insensitive regression (eps-regression);

[0060] kernel: Uses radial basis function (radial / RBF);

[0061] cost (regularization parameter C): Set to 5 to control the balance between error tolerance and model complexity;

[0062] gamma (RBF kernel function coefficient): set to 0.05;

[0063] epsilon (width of ε-insensitive loss function): set to 0.01;

[0064] scale: Enables automatic standardization processing;

[0065] Extreme Gradient Boosting (XGBoost) Parameter Tuning: An ensemble learning method based on gradient boosting trees, with key parameter settings as follows:

[0066] booster: Uses a GBTree-based learner;

[0067] objective: Use squared error regression;

[0068] eta (learning rate): 0.01;

[0069] max_depth (maximum tree depth): 2;

[0070] nrounds (number of iterations): 500;

[0071] subsample: 0.7;

[0072] colsample_bytree (feature sampling): 0.7;

[0073] min_child_weight: 15;

[0074] early_stopping_rounds=20;

[0075] Gaussian Process Regression (GPR) Parameter Tuning: Based on a nonparametric regression model within a Bayesian statistical framework, the key parameter settings are as follows:

[0076] Kernel function: Radial basis function (RBF / rbfdot) is used.

[0077] Sigma (RBF kernel width parameter): Search range 0.001-10;

[0078] var (noise variance): Search range 0.1-0.2;

[0079] Using the validation period NSE as the objective function, the optimal parameter combination is selected from 25 sets of parameters.

[0080] Comparison of S400 Machine Learning Input Schemes and Selection of the Optimal Scheme

[0081] To clarify the impact of different types of input information on the residual correction effect of machine learning, five different input feature schemes were designed and compared, and residual prediction inputs were constructed from three aspects: climate driving factors, PHNS model simulated flow, and artificially controlled parameters.

[0082] Option A: Use only climate factors as input, including precipitation, temperature, and potential evapotranspiration;

[0083] Option B: Use only the PHNS model to simulate the flow state as input, including the simulated flow at the current moment and the simulated flow at the previous moment;

[0084] Option C: Use climate driving factors and PHNS model simulation flow state as input, including precipitation, temperature, potential evapotranspiration, current simulated flow and previous simulated flow;

[0085] Option D: Based on Option C, further add artificially controlled parameters (all of which are already present in the PHNS model, see the aforementioned cited article for details), including the lower limit of suitable water depth for paddy fields (hQmin1), the upper limit of suitable water depth for paddy fields (hQmax1), the upper limit of water depth for rice to withstand flooding (hQflood1), and the starting water level for surface water drainage (…). ), surface water drainage stop level ( ).

[0086] Option E: Based on Option D, remove the upper limit of the suitable water depth in paddy fields (hQmax1) and the surface water drainage stop level. The lower limit of suitable water depth for paddy fields (hQmin1), the upper limit of water depth for rice to withstand flooding (hQflood1), and the starting water level for surface water drainage are retained. ).

[0087] By calculating the NSE and R0 of the flow simulation results under different model combinations and input schemes, the NSE and R0 values ​​were obtained. 2 Evaluation metrics such as RMSE and PBIAS were used to quantitatively assess the residual correction effectiveness of different input feature schemes. Results are as follows: Figure 2 As shown in Table 1, the specific results of the NSE index are as follows:

[0088] Table 1 compares the NSE of the five machine learning models under the five input feature combination schemes.

[0089]

[0090] The results show that Scheme E outperforms Schemes A, B, C, and D in both the calibration and validation periods. Scheme D shows a smaller improvement compared to Scheme C, but all five models show improvement in both the calibration and validation periods, indicating that the improvement from adding manually adjusted parameters is limited. Further removal of hQmax1 and... Afterwards, scheme E can effectively improve the machine learning model's ability to predict PHNS simulation residuals.

[0091] Therefore, scheme E is determined as the optimal combination of input features, and scheme E is used as the unified input feature for subsequent machine learning model selection.

[0092] S500 Optimal Coupling Model Determination

[0093] Under the optimal scheme E, after correcting the PHNS simulation residuals using machine learning, the accuracy of the polder flow simulation was significantly improved during both the calibration and validation periods.

[0094] like Figure 2 , Figure 3 , Figure 4 As shown, the R-value of the periodic PHNS model 2 The R-value is 0.52, and the NSE is 0.46; under scheme E, the R-values ​​of each coupled model are... 2 The R² was improved to 0.68–0.85, and the NSE was improved to 0.75–0.90. Among them, GPR performed best, with an NSE of 0.90, an R² of 0.85, and an RMSE of 5.33. During the validation period, the R² of the original PHNS model was... 2 The NSE is 0.44, and the NSE is 0.40. Under scheme E, the NSE of each coupled model is higher than that of the original PHNS, ranging from 0.56 to 0.65. R 2 The performance was improved to 0.51~0.60, with GPR having an NSE of 0.65, R² of 0.60, and RMSE of 9.48. Its RMSE is close to that of the LSTM model, and its overall performance is the best.

[0095] S600 Scenario Analysis

[0096] Based on the optimal coupling model, we conduct hydrological response pattern analysis under human activities and climate change:

[0097] (1) Analysis of the impact of irrigation

[0098] By adjusting the model's optimal field maximum (h) Qmax1 ), minimum water level (h) Qmin1 Design two scenarios, "with irrigation" and "without irrigation," to ensure that the operation rules and other boundary conditions of the drainage pumping station are completely consistent.

[0099] The results show that the trends of the two runoff curves are basically the same under both irrigation and non-irrigation scenarios. However, during the period of heavy rainfall from late June to early July, the initial peak runoff was slightly higher under the non-irrigation scenario, while the peak flow in the middle and later stages was relatively higher and fluctuated more significantly under the irrigated scenario. During the low-flow phase, the flow rate under the irrigated scenario was generally higher than that under the non-irrigated scenario. Irrigation, by altering the underlying surface water state and combining this with the storage capacity of dikes and ponds in the polder area, replenishes the runoff formation process.

[0100] (2) Analysis of the impact of pump station operation

[0101] By adjusting the surface water drainage start-up water level ( ), surface water drainage stop level ( Two contrasting scenarios were designed: "with drainage pumping stations in operation" and "without drainage pumping stations in operation," while keeping the irrigation system and other meteorological and hydrological boundary conditions unchanged.

[0102] The results show that the runoff hydrograph under the pump-free scenario is relatively flat overall, with small fluctuations in peak and trough values. Under the pump-equipped scenario, the runoff hydrograph exhibits strong fluctuations, with a significant high flow peak occurring during periods of heavy rainfall; subsequently, the runoff rapidly declines, dropping to near zero during the rainfall subsidence and non-rainfall periods. Pumping stations can rapidly discharge high flow peaks during periods of heavy rainfall, while causing the runoff to quickly drop to near zero and remain at extremely low levels during the rainfall subsidence and non-rainfall periods, forming a "peak-trough" phenomenon.

[0103] (3) Analysis of the impacts of climate change

[0104] The MRI-ESM2-0 global climate model from the CMIP6 Sixth International Coupled Model Intercomparison Project was adopted, and meteorological forecast data under the SSP2-4.5 moderate emissions scenario were selected as the future driving force. Two time periods, 2030-2035 and 2040-2045, were selected respectively to extract core meteorological elements such as daily precipitation, maximum and minimum temperatures, and input them into the optimal coupled model to conduct future hydrological response simulations.

[0105] The results show that, on an annual average basis, runoff in all future periods will show a downward trend, decreasing by approximately 0.31 mm / d and 0.30 mm / d, respectively. On a seasonal scale, the average summer runoff for 2030-2035 and 2040-2045 will be lower than that for 2020-2025, decreasing by 1.25 mm / d and 1.15 mm / d, respectively. Autumn runoff also shows a significant decrease, decreasing by approximately 0.30 mm / d and 0.32 mm / d, respectively. Conversely, spring and winter runoff show slight increases. The future runoff pattern in the polder area will exhibit a significant decrease in summer and autumn, an increase in winter and spring, and an overall decline in annual runoff.

Claims

1. A method for simulating runoff in a plain watershed with strong artificial regulation, characterized in that, The method includes: The PHNS hydrological model was used to simulate the runoff process in the polder area, and the time series of the simulated flow was obtained. Construct the residual sequence between the simulated flow time series and the measured flow time series of the hydrological model; The residual sequence is used as the output variable of the machine learning model, and the climate characteristic parameters, simulated flow, and artificial control parameters characterizing the operation status of artificial water management facilities at each time corresponding to each residual in the residual sequence are used as input variables to train the machine learning model and obtain an error prediction model; wherein, the machine learning model establishes a nonlinear mapping relationship between the input variables and the output variables. The hydrological model is used to simulate the runoff process in the polder area. During the prediction process, the error prediction model is used to correct the simulation results and obtain the final simulation results.

2. The method according to claim 1, characterized in that, The climate characteristic parameters are precipitation, temperature and potential evapotranspiration; the simulated flow includes the simulated flow at the current moment and the simulated flow at the previous moment.

3. The method according to claim 1, characterized in that, The artificially controlled parameters are selected from the following parameters: lower limit of suitable water depth in paddy fields (hQmin1), upper limit of suitable water depth in paddy fields (hQmax1), upper limit of water depth tolerating flooding in rice (hQflood1), and starting water level for surface water drainage. and surface water drainage stop level One or more of them.

4. The method according to claim 3, characterized in that, The artificially controlled parameters are the lower limit of suitable water depth in paddy fields (hQmin1), the upper limit of water depth tolerating flooding in rice (hQflood1), and the starting water level for surface water drainage. .

5. The method according to claim 1, characterized in that, The machine learning model used is a Gaussian process regression model.

6. The method according to claim 1, characterized in that, The process of obtaining the final simulation results includes: The predicted residuals are obtained by using a trained Gaussian process regression model to predict the residuals. The corrected flow rate is obtained by adding the simulated flow rate of the PHNS hydrological model to the predicted residual.

7. The method according to claim 1, characterized in that, The method also includes a scenario simulation analysis step: By adjusting the values ​​of the artificial control parameters or removing the preset artificial control parameters, different human activity scenarios or climate change scenarios can be constructed. Input data corresponding to different scenarios are input into the hydrological model for simulation, and the simulation results are corrected using the error correction model to obtain the final simulation results under different scenarios, so as to assess the impact of human activities or climate change on the runoff process of the polder area.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-7.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-7.