Self-adaptive intelligent control method and system for highway drainage in hot and humid environment

Abstract

The application provides a kind of hot and humid environment highway drainage self-adaptive intelligent control method and system, it is related to drainage control technical field, including by obtaining hot and humid environment parameters and precipitation dynamic parameters, constructing evapotranspiration intensity field and calculating confluence process, identifying saturated pipe section and redundant pipe section, optimizing peak overlap degree index to obtain start time sequence and construct flow transfer path, while sequence fusion is carried out by recursive state estimation method and iteration update environmental response coefficient, drainage control strategy is generated and implemented. Realize the self-adaptive intelligent control of highway drainage system under hot and humid environment, improve drainage efficiency.

Description

Technical Field

[0001] This invention relates to the field of drainage control technology, and in particular to an adaptive intelligent control method and system for highway drainage in humid and hot environments. Background Technology

[0002] Highway drainage control methods typically rely on static hydrological models or empirical rules. In existing technologies, designers often pre-determine the size and slope of drainage pipes and channels based on historical precipitation data or regional climate zones, and use fixed thresholds to control the opening and closing of pump stations or gates. The operation and management of drainage systems are mainly based on manual inspections or simple liquid level feedback, lacking real-time response to dynamic changes in the flow generation process.

[0003] This conventional approach overlooks a crucial factor: intense evapotranspiration in hot and humid environments significantly alters surface runoff. When temperatures are high and humidity is high, the rates of water evaporation from soil and roads, as well as transpiration from plants, are far higher than in arid regions. This results in the actual amount of water flowing into drainage networks differing from calculations based solely on precipitation intensity. Hydrological models based on fixed empirical parameters cannot reflect the transient changes in evapotranspiration intensity with meteorological conditions, thus leading to inaccuracies in net runoff estimation.

[0004] Furthermore, existing drainage control strategies typically employ centralized scheduling logic, relying on water level sensor signals to uniformly control each drainage unit. This method struggles to effectively address the non-uniform spatiotemporal distribution of flow within the pipe network during rainfall. When multiple upstream areas simultaneously generate peak runoff, downstream pipe sections are prone to overload, leading to localized water accumulation risks. Meanwhile, other pipe sections in the system may be underloaded and underutilized. Due to the lack of real-time identification and flow transfer mechanisms for pipe section status (saturation or redundancy), existing technologies cannot achieve optimized flow allocation at the pipe network level, resulting in low drainage efficiency. Summary of the Invention

[0005] The present invention provides an adaptive intelligent control method and system for highway drainage in humid and hot environments, which can solve the problems in the prior art.

[0006] A first aspect of the present invention provides an adaptive intelligent control method for highway drainage in humid and hot environments, comprising:

[0007] Acquire the humid and thermal environmental parameters and precipitation dynamic parameters of the highway drainage area;

[0008] An evapotranspiration intensity field is constructed based on the parameters of the humid and hot environment. The net runoff of each drainage unit is obtained by calculating the evapotranspiration intensity field and precipitation dynamic parameters according to the boundary of the drainage unit. The runoff process of each drainage unit is calculated by the runoff transfer matrix.

[0009] Identify saturated and redundant pipe sections in the drainage network based on the confluence process;

[0010] Using the start-up time of each drainage unit as the decision variable, the start-up time sequence is obtained by optimizing the peak overlap metric index so that the peak values ​​of multiple upstream confluences flowing to the saturated pipe section are staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, a flow transfer path from the saturated pipe section to the redundant pipe section is constructed to generate a drainage control strategy.

[0011] The measured drainage flow and the calculated flow are sequentially fused by the recursive state estimation method. The deviation feature of the fused sequence is extracted and used as the correction value to iteratively update the environmental response coefficient in the evapotranspiration intensity field.

[0012] The drainage control strategy is translated into execution instructions for each drainage unit and implemented.

[0013] In one optional embodiment, an evapotranspiration intensity field is constructed based on humid and hot environmental parameters. The evapotranspiration intensity field and precipitation dynamic parameters are then calculated along the boundaries of the drainage units to obtain the net runoff of each drainage unit. The runoff process of each drainage unit is then calculated using a runoff transfer matrix, including:

[0014] Temperature and humidity field tensors are extracted from humid and hot environmental parameters. An evapotranspiration potential field is constructed through the mapping relationship of saturated water vapor pressure difference. The evapotranspiration potential field and wind speed field are coupled by tensor to generate an evapotranspiration intensity field with spatial variation characteristics.

[0015] The evapotranspiration intensity field is spatially divided according to the boundary of the drainage unit, the evapotranspiration intensity distribution matrix in each drainage unit is extracted, and the cumulative evapotranspiration of each drainage unit is obtained by time-domain integration of the evapotranspiration intensity distribution matrix.

[0016] The precipitation dynamic parameters are spatially matched according to the boundary of the same drainage unit, the cumulative precipitation of each drainage unit is extracted, and the difference between the cumulative precipitation and the cumulative evapotranspiration of the corresponding drainage unit is calculated to obtain the net runoff of each drainage unit.

[0017] A flow transfer matrix is ​​constructed based on the topological connection relationship of each drainage unit. The net flow of each drainage unit is iteratively calculated through the flow transfer matrix to obtain the flow time history curve of the outlet of each drainage unit as the flow transfer process.

[0018] In one optional embodiment, a flow confluence matrix is ​​constructed based on the topological connection relationship of each drainage unit. The net flow rate of each drainage unit is iteratively calculated through the flow confluence matrix to obtain the flow time history curve of the outlet of each drainage unit as the flow confluence process, which includes:

[0019] Traverse the outlet nodes of each drainage unit and the inlet nodes of adjacent drainage units, identify drainage unit pairs with water flow connectivity, and record the connection relationship of drainage unit pairs as a topology connection index.

[0020] The water flow propagation time is calculated based on the confluence path length and terrain slope of each drainage unit, and then converted into a time delay step.

[0021] Using the topology connection index as the matrix coordinate and the time delay step as the matrix element value, a flow transfer matrix is ​​constructed, and the water retention coefficient of each drainage unit is set at the diagonal position of the matrix.

[0022] The net flow of each drainage unit is organized into an initial flow vector. The flow transfer matrix is ​​multiplied with the initial flow vector to obtain the flow distribution vector at the first time step.

[0023] The flow distribution vector is used as input and iteratively calculated with the confluence transfer matrix until the difference between the flow distribution vectors of adjacent time steps satisfies the convergence condition.

[0024] Extract the flow components of each drainage unit outlet from the flow distribution vector at each time step, and organize them into a flow time history curve in chronological order.

[0025] In one optional embodiment, the start-up time of each drainage unit is used as the decision variable. By optimizing the peak overlap metric, the start-up time sequence is obtained, so that the peak flows of multiple upstream confluences to the saturated pipe section are staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, a flow transfer path from the saturated pipe section to the redundant pipe section is constructed, and a drainage control strategy is generated, including:

[0026] Extract the set of upstream drainage units flowing to each saturated pipe section, and identify the peak flow time and peak duration from the flow time history curves of each upstream drainage unit;

[0027] The start-up time of each drainage unit is used as a decision variable, and multiple candidate start-up time sequences are initialized in the solution space.

[0028] For each candidate start-up time sequence, the flow time history curves of each upstream drainage unit are shifted along the time axis according to the corresponding start-up time and superimposed moment by moment to obtain the inflow time history curve of the saturated pipe section.

[0029] From the inflow time history curve, the peak overlap metric is statistically analyzed. The distribution pattern of the start time corresponding to the minimum value of the peak overlap metric is extracted to determine the search direction guidance information. Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner until convergence, and the start time sequence is output.

[0030] Identify the upstream node location of each saturated pipe segment and the inlet node location of each redundant pipe segment in the drainage pipe network;

[0031] Search for the set of pipe segment transmission paths connecting the upstream node and the inlet node in the topological connection structure of the drainage pipe network, calculate the remaining water carrying capacity of each pipe segment transmission path, allocate the overload flow of each saturated pipe segment according to the remaining water carrying capacity, and construct the flow transfer path.

[0032] The startup time sequence is combined with the flow transfer path to generate a drainage control strategy.

[0033] In one optional embodiment, a peak overlap metric is statistically analyzed from the inflow time history curve. The distribution pattern of the start time corresponding to the minimum value of the peak overlap metric is extracted to determine the search direction guidance information. Based on the search direction guidance information, other candidate start time sequences are directionally adjusted until convergence. The output start time sequence includes:

[0034] Scan the inflow time history curve every moment to identify the time period when the flow rate exceeds the threshold of the saturated pipe section's water supply capacity.

[0035] The cumulative flow during the period exceeding the threshold is obtained by performing time-domain integration on the flow value and serving as a peak overlap metric.

[0036] The minimum value is identified from the peak overlap metric index corresponding to each candidate start time sequence, and the distribution pattern of the start time of each drainage unit in the candidate start time sequence corresponding to the minimum value is extracted to obtain search direction guidance information;

[0037] Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner so that the adjusted start time distribution pattern is closer to the search direction guidance information.

[0038] For the adjusted candidate start time sequence, re-execute the time axis translation and overlay operation, and update the peak overlap metric.

[0039] The iteration terminates when the change in the minimum value of the peak overlap metric is less than the preset convergence threshold, and the candidate start time sequence with the minimum peak overlap metric is output as the start time sequence.

[0040] In one optional embodiment, the measured drainage flow rate and the calculated flow rate are sequentially fused using a recursive state estimation method. The deviation characteristics of the fused sequence are extracted, and the deviation characteristics are used as correction values ​​to iteratively update the environmental response coefficients in the evapotranspiration intensity field.

[0041] The measured and calculated drainage flows of each drainage unit are collected at time steps to form observation sequence pairs;

[0042] Read the measured and calculated flow rates of the previous time step, calculate the predicted flow rate of the current time step based on the recursive propagation law, and convert the deviation between the predicted flow rate and the observed sequence pair into a correction amount and add it to the predicted flow rate to obtain the fused sequence.

[0043] The bias features are extracted from the fusion sequence by segmentation according to time windows;

[0044] The environmental response coefficient of each drainage unit is subjected to a disturbance increment and recalculated. A sensitivity correlation matrix is ​​constructed based on the calculated flow change.

[0045] The correction amount for the environmental response coefficient is obtained by calculating the deviation characteristic quantity and the sensitivity correlation matrix;

[0046] The correction is added to the environmental response coefficient to complete the iterative update, and the net production flow for the next time step is recalculated.

[0047] In one optional embodiment, the environmental response coefficient of each drainage unit is subjected to a disturbance increment and recalculated. A sensitivity correlation matrix is ​​constructed based on the calculated flow change, including:

[0048] Apply a perturbation increment to the environmental response coefficient of the target drainage unit;

[0049] The evapotranspiration intensity field is reconstructed using the perturbed environmental response coefficients, and the net flow rate and confluence process of the target drainage unit are recalculated. At the same time, the calculated flow rate changes of the downstream drainage units associated with the confluence transfer matrix are extracted.

[0050] The ratio of the calculated flow change of the target drainage unit to the disturbance increment is used as the direct sensitivity, and the ratio of the calculated flow change of the downstream drainage unit to the disturbance increment is used as the transmission sensitivity.

[0051] Fill the direct sensitivity and the transmitted sensitivity into the sensitivity correlation matrix.

[0052] A second aspect of the present invention provides an adaptive intelligent control system for highway drainage in humid and hot environments, comprising:

[0053] The parameter acquisition unit is used to acquire the humid and thermal environment parameters and precipitation dynamic parameters of the highway drainage area.

[0054] The evapotranspiration unit is used to construct an evapotranspiration intensity field based on the parameters of the humid and hot environment. The evapotranspiration intensity field and precipitation dynamic parameters are calculated according to the boundary of the drainage unit to obtain the net flow of each drainage unit. The flow confluence process of each drainage unit is calculated through the flow confluence transfer matrix.

[0055] The pipe segment identification unit is used to identify saturated and redundant pipe segments in the drainage network based on the confluence process.

[0056] The regulation and optimization unit is used to obtain the start-up time sequence by optimizing the peak overlap metric index of each drainage unit, so that the peak flow of multiple upstream confluence peaks flowing to the saturated pipe section is staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, the flow transfer path from the saturated pipe section to the redundant pipe section is constructed to generate the drainage regulation strategy.

[0057] The deviation correction unit is used to fuse the measured drainage flow and the calculated flow through a recursive state estimation method, extract the deviation feature of the fused sequence, and use the deviation feature as a correction quantity to iteratively update the environmental response coefficient in the evapotranspiration intensity field.

[0058] The instruction implementation unit is used to convert drainage control strategies into execution instructions for each drainage unit and implement them.

[0059] A third aspect of the present invention provides an electronic device, comprising:

[0060] processor;

[0061] Memory used to store processor-executable instructions;

[0062] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0063] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0064] In this embodiment of the invention, by constructing an evapotranspiration intensity field and integrating precipitation dynamic parameters, the net flow generated by each drainage unit under hot and humid climates can be accurately quantified. Combined with the confluence transfer matrix to deduce the confluence process, the flow generation estimation deviation caused by neglecting the influence of evapotranspiration in traditional methods is effectively overcome, making drainage prediction more consistent with actual hot and humid conditions. Based on the identification results of saturated and redundant pipe sections, the startup time sequence is optimized to achieve staggered distribution of upstream confluence peaks, significantly reducing the risk of pipe section overload operation. The remaining water supply capacity is used to construct flow transfer paths, fully tapping the potential of redundant pipe sections, and the overall drainage efficiency is significantly improved. Based on the recursive state estimation method, the measured flow and calculated flow are sequentially fused, and the deviation feature quantity is extracted to iteratively update the environmental response coefficient in the evapotranspiration intensity field in real time. This enables the model to automatically adapt to the dynamic changes of parameters such as soil moisture and air temperature in hot and humid environments, avoiding control deviations caused by parameter solidification and ensuring the accuracy and stability of the system's long-term operation. In the face of sudden rainstorms or continuous hot and humid weather, the system automatically coordinates the actions of equipment such as gates and pumping stations to ensure that the pipeline network is always in the optimal working state, effectively reducing the probability of road surface water accumulation and waterlogging, while extending the service life of the pipeline network. Attached Figure Description

[0065] Figure 1 A flowchart illustrating an adaptive intelligent control method for highway drainage in humid and hot environments;

[0066] Figure 2 Generate a logic flowchart for the drainage control strategy. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0069] Figure 1 This is a flowchart illustrating the adaptive intelligent control method for highway drainage in humid and hot environments according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0070] Acquire the humid and thermal environmental parameters and precipitation dynamic parameters of the highway drainage area;

[0071] An evapotranspiration intensity field is constructed based on the parameters of the humid and hot environment. The net runoff of each drainage unit is obtained by calculating the evapotranspiration intensity field and precipitation dynamic parameters according to the boundary of the drainage unit. The runoff process of each drainage unit is calculated by the runoff transfer matrix.

[0072] Identify saturated and redundant pipe sections in the drainage network based on the confluence process;

[0073] Using the start-up time of each drainage unit as the decision variable, the start-up time sequence is obtained by optimizing the peak overlap metric index so that the peak values ​​of multiple upstream confluences flowing to the saturated pipe section are staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, a flow transfer path from the saturated pipe section to the redundant pipe section is constructed to generate a drainage control strategy.

[0074] The measured drainage flow and the calculated flow are sequentially fused by the recursive state estimation method. The deviation feature of the fused sequence is extracted and used as the correction value to iteratively update the environmental response coefficient in the evapotranspiration intensity field.

[0075] The drainage control strategy is translated into execution instructions for each drainage unit and implemented.

[0076] In one optional embodiment, an evapotranspiration intensity field is constructed based on humid and hot environmental parameters. The evapotranspiration intensity field and precipitation dynamic parameters are then calculated along the boundaries of the drainage units to obtain the net runoff of each drainage unit. The runoff process of each drainage unit is then calculated using a runoff transfer matrix, including:

[0077] Temperature and humidity field tensors are extracted from humid and hot environmental parameters. An evapotranspiration potential field is constructed through the mapping relationship of saturated water vapor pressure difference. The evapotranspiration potential field and wind speed field are coupled by tensor to generate an evapotranspiration intensity field with spatial variation characteristics.

[0078] The evapotranspiration intensity field is spatially divided according to the boundary of the drainage unit, the evapotranspiration intensity distribution matrix in each drainage unit is extracted, and the cumulative evapotranspiration of each drainage unit is obtained by time-domain integration of the evapotranspiration intensity distribution matrix.

[0079] The precipitation dynamic parameters are spatially matched according to the boundary of the same drainage unit, the cumulative precipitation of each drainage unit is extracted, and the difference between the cumulative precipitation and the cumulative evapotranspiration of the corresponding drainage unit is calculated to obtain the net runoff of each drainage unit.

[0080] A flow transfer matrix is ​​constructed based on the topological connection relationship of each drainage unit. The net flow of each drainage unit is iteratively calculated through the flow transfer matrix to obtain the flow time history curve of the outlet of each drainage unit as the flow transfer process.

[0081] In one specific implementation, extracting the temperature and humidity field tensors from the humid and thermal environment parameters is the starting point for constructing the evapotranspiration intensity field. The temperature field tensor consists of air temperature observations from various spatial sampling points within the highway drainage area. Discrete station observations are extended into a continuous spatial distribution tensor using Kriging interpolation or inverse distance weighting methods, denoted as... Each element corresponds to the temperature value of a spatial grid cell. Humidity field tensor The data is generated using relative humidity or specific humidity as the basic variable and the same spatial interpolation strategy. In hot and humid environments, the spatial heterogeneity of temperature and humidity is significant. For example, the local warming effect near asphalt pavement can cause evapotranspiration intensity to be significantly higher than that of the surrounding vegetation cover area. Therefore, it is necessary to retain the spatial variation characteristics rather than using the regional average.

[0082] The saturated vapor pressure difference is the core physical quantity driving the evapotranspiration process. This is achieved using the temperature field tensor. Calculate the saturated vapor pressure at each grid point using the Magnus formula. ,Right now ,in Temperature in Celsius The unit is kPa. Actual water vapor pressure. From the humidity field tensor The product of the product and the saturated vapor pressure is obtained, i.e. (Element-by-element operation). Saturated water vapor pressure difference tensor Defined as This tensor directly reflects the driving force of atmospheric absorption of moisture; the larger the value, the stronger the evapotranspiration potential. Based on Furthermore, by incorporating solar radiation intensity parameters and employing a simplified mapping relationship within the Penman-Monteith framework, the saturated water vapor pressure difference is mapped into an evapotranspiration potential field. This mapping relationship takes into account the modulation effect of net radiative flux and aerodynamic impedance on evapotranspiration rate.

[0083] Evaporation potential energy field Tensor coupling with the wind speed field is a crucial step in generating the evapotranspiration intensity field. Wind speed field tensor Generated by interpolation of wind speed observations from various anemometer stations within the region, in m / s. The effect of wind speed on evapotranspiration is expressed through the aerodynamic transport coefficient. reflect, It is a monotonically increasing function of wind speed, with a larger increase at low wind speeds and tending to saturate at high wind speeds. It can be fitted using a logarithmic law or power function form. Evapotranspiration intensity field It is obtained through element-wise tensor coupling operations, i.e. ,in This is the environmental response coefficient tensor, reflecting the comprehensive correction of evapotranspiration by local environmental factors such as land cover type and soil moisture content. This indicates element-wise multiplication. In subsequent iterative update steps, dynamic corrections will be made based on the measured flow deviation, thereby continuously improving the estimation accuracy of the evapotranspiration intensity field over time.

[0084] Evaporation intensity field Spatial segmentation based on drainage unit boundaries is a transformation process from a continuous spatial field to discrete drainage unit quantities. The drainage unit boundaries are determined by hydrological analysis using a digital elevation model (DEM), and each drainage unit corresponds to an independent catchment area. For each drainage unit... Extract the evapotranspiration intensity values ​​of all grid points within the unit boundary to construct an evapotranspiration intensity distribution matrix. The row and column dimensions of the matrix correspond to the number of spatial grid cells within a single cell. Perform time-domain integration, that is, integrate the evapotranspiration intensity values ​​at each moment within the observation period according to the time step. Summing yields the drainage unit. Cumulative evaporation The calculation formula is ,in For the first Drainage unit within each time step The average surface evaporation intensity, This represents the total number of time steps. The time step is the average surface evapotranspiration intensity. Through the The arithmetic mean or area-weighted mean is obtained in the spatial dimension. The area-weighted mean method can better reflect the influence of topographic relief within the unit on the distribution of evapotranspiration.

[0085] When spatially matching precipitation dynamic parameters according to the boundaries of the same drainage unit, the precipitation intensity data of rain gauges or radar precipitation products are interpolated to a spatial grid consistent with the evapotranspiration intensity field, and then the precipitation values ​​of the corresponding grids are extracted according to the boundaries of each drainage unit. For drainage units... The cumulative precipitation is obtained by integrating the precipitation intensity over the same period of time. The calculation method is similar to that of cumulative evapotranspiration, that is... ,in For the first Drainage unit within each time step The average precipitation intensity. The cumulative precipitation. Cumulative evaporation of the corresponding drainage unit Perform the difference calculation to obtain the net production flow. .when When this occurs, it indicates that the drainage unit is generating positive flow and needs to proceed to subsequent flow convergence calculations; when When the evapotranspiration rate is not less than the precipitation recharge rate, the unit will not generate surface runoff during the current period. In actual hot and humid environments, when evapotranspiration is large but precipitation intensity is even greater, the net runoff may still be high. This difference calculation can accurately reflect the dynamic balance between the two.

[0086] Constructing the flow transfer matrix based on the topological connections of each drainage unit is the core step in calculating the flow convergence process. The topological connections are determined by the node-pipe segment connection diagram of the drainage network, describing the flow convergence direction and weight allocation between the outlet of each drainage unit and the inlet of the downstream pipe segment. Let there be a total... Each drainage unit, flow transfer matrix for A sparse matrix whose elements Indicates drainage unit The outflow flows to the drainage unit The ratio ensures that the sum of each column's elements does not exceed 1 (part of the traffic flows directly out of the system boundary). In the construction... At the same time, a convergence time lag parameter also needs to be introduced. This indicates from the drainage unit to drainage unit The flow transmission delay is estimated using the Manning formula, which takes into account the pipe length, slope, and roughness coefficient. The net flow rate of each drainage unit is then calculated using a time-series calculation. As input, iterative calculations are performed using the bus transfer matrix. In each iteration, the flow from the upstream unit is convolved with a time delay and then superimposed onto the flow from the downstream unit. After each iteration, the flow time-history curves of each drainage unit outlet are obtained, i.e., the confluence process. The peak time and peak flow of the flow time-history curves are important bases for subsequent identification of saturated and redundant pipe sections and the formulation of drainage control strategies. In hot and humid environments, short-duration heavy rainfall occurs frequently. It is recommended that the time resolution of the confluence process calculation be 5 to 15 minutes to accurately capture the spatiotemporal distribution characteristics of the flow peaks and provide refined data support for subsequent peak-shaving control decisions.

[0087] In one optional embodiment, a flow confluence matrix is ​​constructed based on the topological connection relationship of each drainage unit. The net flow rate of each drainage unit is iteratively calculated through the flow confluence matrix to obtain the flow time history curve of the outlet of each drainage unit as the flow confluence process, which includes:

[0088] Traverse the outlet nodes of each drainage unit and the inlet nodes of adjacent drainage units, identify drainage unit pairs with water flow connectivity, and record the connection relationship of drainage unit pairs as a topology connection index.

[0089] The water flow propagation time is calculated based on the confluence path length and terrain slope of each drainage unit, and then converted into a time delay step.

[0090] Using the topology connection index as the matrix coordinate and the time delay step as the matrix element value, a flow transfer matrix is ​​constructed, and the water retention coefficient of each drainage unit is set at the diagonal position of the matrix.

[0091] The net flow of each drainage unit is organized into an initial flow vector. The flow transfer matrix is ​​multiplied with the initial flow vector to obtain the flow distribution vector at the first time step.

[0092] The flow distribution vector is used as input and iteratively calculated with the confluence transfer matrix until the difference between the flow distribution vectors of adjacent time steps satisfies the convergence condition.

[0093] Extract the flow components of each drainage unit outlet from the flow distribution vector at each time step, and organize them into a flow time history curve in chronological order.

[0094] In one specific implementation, based on obtaining the net flow rate of each drainage unit, it is necessary to aggregate the flow generation information dispersed among the drainage units to the drainage network nodes to form a complete description of the confluence process. The outlet nodes of each drainage unit and the inlet nodes of adjacent drainage units are traversed, and the existence of actual water flow connectivity between adjacent units is checked one by one. The criteria for determining connectivity include: whether there is a direct connection between the drainage pipes and channels between the two drainage units, whether the direction of the terrain elevation difference is consistent (i.e., the outlet elevation of the upstream unit is higher than the inlet elevation of the downstream unit), and whether there are corresponding directed edges in the drainage network topology graph. For drainage unit pairs that meet the above conditions, their upstream unit numbers and downstream unit numbers are paired into an ordered pair and recorded as a topology connection index, forming a directed connectivity graph that fully describes the confluence direction relationship of the entire drainage area. This topology connection index table provides coordinate basis for the subsequent construction of the confluence transfer matrix, ensuring that the position of each non-zero element in the matrix strictly corresponds to the actual water flow path.

[0095] The propagation time of water flow from the upstream drainage unit to the downstream drainage unit directly affects the timing of the peak flow rate at the outlet of each drainage unit. This is based on the length of the confluence path between the drainage units. (Unit: m) and gradient of the terrain along the route (Dimensionless) Estimate the average velocity of the water flow along this path using Manning's formula or empirical velocity formula. (Unit: m / s), thus obtaining the water propagation time. (Unit: s). [The following appears to be a separate, unrelated sentence:] Will Divide by time step And rounded down to get the time delay step. ,Right now Time delay step The physical meaning is: upstream drainage unit The traffic generated at the current moment, after It will take several time steps to reach the downstream drainage unit. The entry node. When the confluence path is short or the slope is large. When the value is small, the flow transmission is almost synchronous; when the confluence path is long or the gradient is gentle, When the value is large, there is a significant lag in traffic transmission.

[0096] Indexed by topology connection Using matrix coordinates with time delay steps Construct a flow transfer matrix using the matrix element values. Its dimensions are ,in This represents the total number of drainage units. The nth unit in the matrix... Line 1 Column elements Indicates drainage unit Flow to drainage unit For data transmission delay, for pairs of units without direct connectivity, the corresponding matrix element is set to zero. At the diagonal position of the matrix, i.e., the [missing element]... Line 1 Columns are set to determine the water retention coefficient for each drainage unit. (Dimensionless, ranging from 0 to 1). The water retention coefficient reflects the flow attenuation effect within a drainage unit caused by factors such as water storage in pipes and channels, and interception by surface depressions: The closer it is to 1, the weaker the flow retention effect of the drainage unit, and the more the generated flow is transmitted downstream. The smaller the value, the stronger the water storage capacity inside the unit, and the more obvious the attenuation of the flow transmission. The value can be calibrated based on physical parameters such as the proportion of impermeable area of ​​the drainage unit and the ratio of water storage volume to water collection area in the pipe and channel.

[0097] The net flow rate of each drainage unit at the initial moment Organization as the initial flow vector , of which Each component corresponds to a drainage unit. Net flow. The flow transfer matrix. With the initial flow vector Perform matrix multiplication to obtain the flow distribution vector at the outlet of each drainage unit at the end of the first time step. ,Right now The essence of this operation is to redistribute the flow generated by each upstream drainage unit according to the topological connection relationship and time delay step, so that each downstream drainage unit receives the sum of the flow contributions from all its directly upstream units in the first time step. Due to the time delay step... The existence of this means that only the traffic components that meet the time delay condition are actually involved in the superposition of the matrix multiplication operation, and the remaining traffic components will arrive successively in subsequent time steps.

[0098] Traffic distribution vector As input to the next time step, it continues to be used with the flow transfer matrix. Perform iterative calculations to obtain the first... Flow distribution vector at time step Each iteration calculates the flow distribution from the previous time step, adding the new net flow from each drainage unit in the current time step. This ensures that the runoff generated continuously during precipitation is fully accumulated. Iterative calculations continue until the norm of the difference between the runoff distribution vectors of two adjacent time steps satisfies a preset convergence condition. ,in The convergence threshold is determined based on the design flow rate of the drainage network, typically between 0.1% and 1% of the peak design flow rate. Even after the precipitation process ends and the net flow rate of each drainage unit returns to zero, the iterative calculation continues until the residual flow in the network is completely dissipated and the flow distribution vector approaches zero, at which point the iteration can be terminated.

[0099] After completing the matrix multiplication operation at each time step, from the flow distribution vector Extract the flow components corresponding to the outlet of each drainage unit. This component is the drainage unit. In the The instantaneous flow rate at the outlet of each time step. Arrange the flow rate components of all time steps in chronological order to form a drainage unit. Flow time history curve ,in This represents the total number of time steps. The flow time history curve fully describes the change in flow rate at the drainage unit outlet over time, including three stages: the initial flow stage, the peak flow stage, and the receding flow stage. The time and peak flow rate corresponding to the peak flow stage are key inputs for subsequent identification of saturated and redundant pipe sections: if the flow rate carried by a pipe section at the peak time exceeds its design capacity, it is marked as a saturated pipe section; if the maximum flow rate of a pipe section during the entire confluence process is much lower than its design capacity, it is marked as a redundant pipe section, providing usable capacity basis for the subsequent construction of flow transfer paths. Through the above matrix iterative calculation process, a good balance can be achieved between computational efficiency and simulation accuracy, supporting the rapid generation of subsequent drainage control strategies.

[0100] like Figure 2 The diagram shown illustrates the logic flowchart for generating the drainage control strategy.

[0101] In one optional embodiment, the start-up time of each drainage unit is used as the decision variable. By optimizing the peak overlap metric, the start-up time sequence is obtained, so that the peak flows of multiple upstream confluences to the saturated pipe section are staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, a flow transfer path from the saturated pipe section to the redundant pipe section is constructed, and a drainage control strategy is generated, including:

[0102] Extract the set of upstream drainage units flowing to each saturated pipe section, and identify the peak flow time and peak duration from the flow time history curves of each upstream drainage unit;

[0103] The start-up time of each drainage unit is used as a decision variable, and multiple candidate start-up time sequences are initialized in the solution space.

[0104] For each candidate start-up time sequence, the flow time history curves of each upstream drainage unit are shifted along the time axis according to the corresponding start-up time and superimposed moment by moment to obtain the inflow time history curve of the saturated pipe section.

[0105] From the inflow time history curve, the peak overlap metric is statistically analyzed. The distribution pattern of the start time corresponding to the minimum value of the peak overlap metric is extracted to determine the search direction guidance information. Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner until convergence, and the start time sequence is output.

[0106] Identify the upstream node location of each saturated pipe segment and the inlet node location of each redundant pipe segment in the drainage pipe network;

[0107] Search for the set of pipe segment transmission paths connecting the upstream node and the inlet node in the topological connection structure of the drainage pipe network, calculate the remaining water carrying capacity of each pipe segment transmission path, allocate the overload flow of each saturated pipe segment according to the remaining water carrying capacity, and construct the flow transfer path.

[0108] The startup time sequence is combined with the flow transfer path to generate a drainage control strategy.

[0109] In one specific implementation, after obtaining the confluence process of each drainage unit and the identification results of saturated and redundant pipe sections, the generation of the drainage control strategy revolves around two core tasks: first, to achieve staggered distribution of peak flows from multiple upstream confluences by optimizing the start-up time sequence; and second, to construct a flow transfer path from saturated pipe sections to redundant pipe sections. The results of the two tasks are ultimately combined into an executable drainage control strategy.

[0110] For each saturated pipe segment, all upstream drainage units flowing into that segment are extracted to form an upstream drainage unit set. Peak values ​​are identified in the flow time-history curve of each drainage unit in the set, and the peak flow time and peak duration are recorded. The peak duration is defined as the period of flow exceeding the peak flow in the flow time-history curve. of times ( The value ranges from 0.5 to 0.8, with the specific value determined based on a combination of the pipe section's design flow capacity and the upstream catchment area, over a continuous period of time. For drainage unit Peak traffic, A threshold coefficient is identified for the peak duration period. The identified peak duration period provides a time window for subsequent statistics on peak overlap metrics.

[0111] Using the start-up time of each drainage unit as the decision variable, initialization is performed in the solution space. There are 10 candidate start-up time sequences, among which This represents the number of candidate solutions, typically between 20 and 100 to ensure search diversity. Each candidate start-up time sequence is defined by... The offset is composed of several start-up time offsets. The range of the offset value is determined based on the precipitation duration and the pipeline response time, and is generally set between 0 and... between, This represents the maximum allowable startup delay (in time steps). During initialization, a uniform random sampling method is used to cover the solution space, preventing candidate solutions from concentrating in local regions.

[0112] For each candidate start-up time sequence, the flow time history curves of each upstream drainage unit are shifted on the time axis according to the corresponding start-up time offset. The shifted flow time history curves are then superimposed element by element at each time point to obtain the inflow time history curve of the saturated pipe section. ,in This is a time index. The inflow time history curve reflects the change in the actual inflow load on the saturated pipe section over time under the current candidate start-up time sequence.

[0113] From the inflow time history curve Chinese statistical peak overlap metric . The calculation method is as follows: Calculate the design flow capacity of the pipe section exceeding saturation in the statistical inflow time history curve. The sum of the products of the total time period and the overload amount, i.e. ,in The design flow capacity of the saturated pipe section, For time step. The smaller the value, the better the peak-shaving effect of multiple upstream confluence peaks on the time axis, and the lower the overload degree of the saturated pipe section. This is based on the corresponding values ​​from all candidate start-up time sequences. The minimum value is extracted from the values, and the candidate start-up time sequence corresponding to the minimum value is taken as the current optimal solution. The distribution pattern of its start-up time is extracted as search direction guidance information. The search direction guidance information is specifically manifested as the distribution pattern of the start-up time offset of each drainage unit in the optimal solution, including the central tendency and dispersion of the offset.

[0114] Based on the search direction guidance information, other candidate start-up time sequences are adjusted in a directional manner: the start-up time offset of each drainage unit in each candidate sequence is moved towards the offset corresponding to the optimal solution by a certain step size. The step size decreases with each iteration to achieve a transition from global exploration to local fine-grained search. After each adjustment, the start-up time offset of each candidate sequence is recalculated. The value is then updated to the optimal solution. The iteration terminates when the optimal solution in two consecutive iterations is [value missing]. The change in value is less than the convergence threshold. ,in To optimize the convergence threshold (compared to the convergence threshold of the merging iteration) (These are independent of each other and have different meanings), or the number of iterations reaches a preset maximum value. After the iteration converges, the final start-up time sequence is output, which makes the peak values ​​of multiple upstream confluences flowing to the saturated pipe section staggered on the time axis.

[0115] After optimizing the startup time sequence, the flow transfer path construction phase begins. This involves identifying the upstream node locations of each saturated pipe segment in the drainage network, i.e., the network node where the inlet of the saturated pipe segment is located; simultaneously, it also identifies the inlet node locations of each redundant pipe segment, i.e., the network node where the starting end of the redundant pipe segment is located. Redundant pipe segments are identified based on the ratio of the current flow load to the design water supply capacity of the segment being lower than the redundancy threshold. , A value of 0.5 is typically used, indicating that the pipe section has sufficient remaining water flow capacity available for use.

[0116] In the topological connection structure of the drainage pipe network, starting from the upstream node of the saturated pipe segment and ending at the inlet node of each redundant pipe segment, a set of pipe segment transmission paths connecting the two is searched. The search process is implemented based on a breadth-first traversal of the pipe network topology graph, retaining only the current remaining water carrying capacity of each pipe segment on the path during the traversal. Paths where all values ​​are greater than zero, among which For pipe section The remaining water carrying capacity is defined as the difference between the designed water carrying capacity of a pipe segment and the current flow rate. For each pipe segment transmission path obtained from the search, the combined remaining water carrying capacity of the path is calculated. Take the minimum remaining water flow capacity of each pipe segment along the path, that is... ,in For path numbering, For path The pipe segment number is used. The overall remaining water capacity reflects the maximum transfer flow that the route can handle without overloading any intermediate pipe segments.

[0117] The overload flow rate of each saturated pipe section is allocated proportionally based on the combined remaining water supply capacity of each available transfer path. Let the overload flow rate of the saturated pipe section be... The set of available transfer paths is Then it will be assigned to the path Transfer flow satisfy ,in For the overload flow of the saturated pipe section, To assign to path The transfer flow, This is the set of available transfer paths. Let `<path>` be the path enumeration variable in the path set. Based on the above proportional allocation results, construct the flow transfer path from the saturated pipe segment to the redundant pipe segment, and clarify the flow allocation ratio and corresponding pipe segment sequence of each transfer path.

[0118] The optimized start-up time sequence is combined with the constructed flow transfer path to form a complete drainage control strategy. The start-up time sequence specifies when each drainage unit initiates drainage actions to achieve peak shaving, while the flow transfer path specifies which pipe segments overloaded flow is routed to redundant pipe segments to reduce pressure in saturated pipe segments. Together, they constitute a drainage control strategy that coordinates regulation in both time and space dimensions, providing a complete decision-making basis for subsequently generating execution instructions for each drainage unit.

[0119] In one optional embodiment, a peak overlap metric is statistically analyzed from the inflow time history curve. The distribution pattern of the start time corresponding to the minimum value of the peak overlap metric is extracted to determine the search direction guidance information. Based on the search direction guidance information, other candidate start time sequences are directionally adjusted until convergence. The output start time sequence includes:

[0120] Scan the inflow time history curve every moment to identify the time period when the flow rate exceeds the threshold of the saturated pipe section's water supply capacity.

[0121] The cumulative flow during the period exceeding the threshold is obtained by performing time-domain integration on the flow value and serving as a peak overlap metric.

[0122] The minimum value is identified from the peak overlap metric index corresponding to each candidate start time sequence, and the distribution pattern of the start time of each drainage unit in the candidate start time sequence corresponding to the minimum value is extracted to obtain search direction guidance information;

[0123] Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner so that the adjusted start time distribution pattern is closer to the search direction guidance information.

[0124] For the adjusted candidate start time sequence, re-execute the time axis translation and overlay operation, and update the peak overlap metric.

[0125] The iteration terminates when the change in the minimum value of the peak overlap metric is less than the preset convergence threshold, and the candidate start time sequence with the minimum peak overlap metric is output as the start time sequence.

[0126] In one specific implementation, after obtaining the inflow time history curves corresponding to each candidate start-up time sequence, it is necessary to quantitatively evaluate the peak overlap of each curve. The inflow time history curve is scanned moment by moment, and the inflow value at each moment is compared with the design flow capacity threshold of the saturated pipe section. Any consecutive period where the inflow value continuously exceeds the threshold is marked as an over-threshold period. The identification of over-threshold periods uses a sliding window method, progressing step-by-step from the start to the end of the time history curve. Once the inflow value first exceeds the threshold at a certain moment, that moment is recorded as the start of the over-threshold period, and so on until the inflow value drops back below the threshold, which is recorded as the end of the over-threshold period. This allows multiple independent over-threshold periods to be identified on a single inflow time history curve.

[0127] For each time period exceeding the threshold, the inflow value is integrated in the time domain. The integration interval is the start and end time of that time period. The integration result is the cumulative flow contribution of the portion exceeding the water supply capacity threshold within that time period. If... Indicates the saturated pipe section at time [time]. The inflow time history value, Indicates the design flow capacity of the saturated pipe section, then for the first... Periods exceeding the threshold The cumulative flow exceeding the threshold is calculated as follows: The peak overlap metric corresponding to the candidate start-up time sequence is obtained by summing the cumulative traffic of all periods exceeding the threshold. . The larger the value, the higher the degree of overlap of multiple upstream confluence peaks on the time axis under the current start-up time arrangement, and the more severe the overload pressure on the saturated pipe section. The smaller the value, the better the staggered distribution of the peak flows over time, and the more effectively the overload risk of the saturated pipe section is suppressed.

[0128] After completing all After calculating the peak overlap metric of the candidate start-up time sequences, the following was identified: The candidate sequence that yields the minimum value is used to extract the specific distribution pattern of the start-up times of each drainage unit within the sequence, which serves as guidance information for the search direction. Essentially, this guidance information is a correspondence between drainage unit numbers and their start-up times, describing the staggered scheduling scheme that minimizes overload pressure in the current iteration. When extracting this pattern, the start-up offset of each drainage unit relative to a reference time is recorded. The overall distribution pattern of these offsets forms the basis for guiding the adjustment of the direction of other candidate sequences.

[0129] Based on the search direction guidance information, for the rest The candidate start-up time sequences are adjusted in a directional manner. The core logic of the adjustment is: for each drainage unit in each candidate sequence, its current start-up time is moved closer to the start-up time of the corresponding drainage unit in the search direction guidance information. The step size for moving closer can be dynamically scaled according to the current iteration round. A larger step size is used in the early stage of the iteration to speed up the convergence speed, and a smaller step size is used in the later stage of the iteration for fine-tuning. During the adjustment process, it must be ensured that the start-up time of each drainage unit never exceeds the maximum allowable start-up delay time. If an adjustment causes the start time to exceed the specified range, it will be truncated to the boundary value. In addition, a certain degree of difference is maintained among the candidate sequences to avoid all sequences converging to the same point too early and losing the ability to explore the solution space. This can be achieved by superimposing a small random perturbation on the adjusted start time to maintain population diversity. The perturbation amplitude gradually decreases with the increase of the iteration rounds.

[0130] After completing the directional adjustment, the time axis shift and overlay operation is re-executed for each adjusted candidate start-up time sequence. Specifically, the confluence process time history curves of each drainage unit are shifted along the time axis according to the adjusted start-up time, and then the flow contributions from all upstream drainage units to the saturated pipe section are overlaid on the time axis to obtain the updated inflow time history curves. Based on this, the threshold period identification and temporal integration operations are re-executed to update the peak overlap metric for each candidate sequence. The updated minimum The minimum value compared to the previous iteration The absolute value of the difference between the values ​​is the change in this iteration, denoted as . .

[0131] when Less than the preset convergence threshold When the iteration process is considered to have converged, the loop terminates. At this point, a sequence of candidate start times is selected. The smallest sequence is used as the starting time sequence for the final output. If the preset maximum number of iterations is reached before... Never lower than Similarly, it will be forcibly terminated when the maximum number of iterations is reached, and the current iteration will be output. The smallest candidate sequence. The output start-up time sequence gives the specific timing arrangement of each drainage unit's start-up drainage, so that the multiple upstream confluence peaks flowing to the saturated pipe section are staggered on the time axis, thereby reducing the overload risk of the saturated pipe section to an acceptable range.

[0132] In actual highway drainage scenarios in hot and humid environments, due to differences in the runoff path length, terrain slope, and water retention characteristics of different drainage units, the timing of the peak flow reaching the saturated pipe section already exhibits a certain degree of natural peak-shaving effect. The aforementioned optimization iteration process further explores the space for artificially controllable start-up times based on this natural peak-shaving. By systematically searching for the optimal peak-shaving scheme, it fully utilizes the existing water-carrying capacity of the drainage network, avoiding local overload problems caused by the simultaneous superposition of multiple peaks. The entire iteration process does not require modification of the network topology; it only requires adjusting the start-up time of each drainage unit to achieve a balanced distribution of flow over time, offering advantages such as low engineering implementation cost and fast response speed.

[0133] In one optional embodiment, the measured drainage flow rate and the calculated flow rate are sequentially fused using a recursive state estimation method. The deviation characteristics of the fused sequence are extracted, and the deviation characteristics are used as correction values ​​to iteratively update the environmental response coefficients in the evapotranspiration intensity field.

[0134] The measured and calculated drainage flows of each drainage unit are collected at time steps to form observation sequence pairs;

[0135] Read the measured and calculated flow rates of the previous time step, calculate the predicted flow rate of the current time step based on the recursive propagation law, and convert the deviation between the predicted flow rate and the observed sequence pair into a correction amount and add it to the predicted flow rate to obtain the fused sequence.

[0136] The bias features are extracted from the fusion sequence by segmentation according to time windows;

[0137] The environmental response coefficient of each drainage unit is subjected to a disturbance increment and recalculated. A sensitivity correlation matrix is ​​constructed based on the calculated flow change.

[0138] The correction amount for the environmental response coefficient is obtained by calculating the deviation characteristic quantity and the sensitivity correlation matrix;

[0139] The correction is added to the environmental response coefficient to complete the iterative update, and the net production flow for the next time step is recalculated.

[0140] In one specific implementation, measured and calculated drainage flow rates are synchronously collected for each drainage unit at time steps. The values ​​of both at the same time step are paired to form observation sequence pairs. The measured drainage flow rate comes from flow sensors deployed at the outlet of each drainage unit, while the calculated flow rate is obtained by combining the flow transfer matrix and net production flow rate. Observation sequence pairs are gradually accumulated along the time axis, providing the input data basis for subsequent recursive state estimation. During the acquisition process, it is necessary to ensure time alignment between the measured signal and the calculation time to avoid introducing systematic biases due to sampling delays.

[0141] After reading the measured and calculated drainage flow rates from the previous time step, the flow state at the current time step is predicted based on the recursive transfer law. The recursive transfer law describes the dynamic evolution of the outlet flow rate of a drainage unit between adjacent time steps, and its core assumption is that the flow state at the current moment can be derived from the state at the previous moment through a linear transfer equation. Let the drainage unit... In the The predicted flow for each time step is The estimated fusion traffic value of the previous time step is The state transitivity is Then the predicted flow satisfies:

[0142] ;

[0143] in This represents the driving force contributed by net runoff inputs, reflecting the incremental contribution of precipitation runoff at the current time step to the outflow. Forecast Flow Measured drainage flow rate at the current time step If there is a deviation, multiply the deviation by the gain factor. The correction amount is then added to the predicted flow to obtain the fused flow estimate for the current time step.

[0144] ;

[0145] Gain coefficient The calculation is based on the variance of the prediction error. With observation noise variance The ratio relationship is as follows:

[0146] ;

[0147] The prediction error variance is updated recursively at each time step, while the observation noise variance is given by the sensor calibration parameters. Through the above recursive process, the fusion sequence is gradually generated on the time axis, and the fusion value at each time step simultaneously takes into account the dynamic constraints of the computational model and the information correction of the measured data.

[0148] Bias features are extracted from the fused sequence segmented by time window. Time window length. Based on the typical duration of precipitation events, the timeframe typically covers a complete runoff generation and confluence cycle. Within each time window, the mean deviation between the fused sequence and the calculated sequence is calculated. and deviation variance subscript The time window is numbered. The mean deviation reflects systematic shifts, while the variance deviation reflects the degree of random fluctuations. The mean deviation and variance deviation are combined to form a vector of deviation characteristics. This serves as the driving input for subsequent environmental response coefficient correction. When the absolute value of the mean deviation exceeds a preset threshold... If a systematic deviation is found in the environmental response coefficient of the current evapotranspiration intensity field, an iterative update process is triggered.

[0149] A perturbation increment is applied to the environmental response coefficient of each drainage unit to construct a sensitivity correlation matrix. Let the drainage unit... The current estimated value of the environmental response coefficient is superscript This is the iteration round number. For Apply positive disturbance The evapotranspiration intensity field and net flow rate of the drainage unit are recalculated, and then the perturbed calculated flow rate sequence is obtained through the flow transfer matrix. The change in calculated flow rate before and after the perturbation is divided by the perturbation amount. This yields an approximate partial derivative of the flow rate with respect to the environmental response coefficient, i.e., the sensitivity coefficient. Perturbations are applied sequentially to all drainage units, and the above process is repeated. The sensitivity coefficients of each unit are then arranged according to the drainage unit dimension and the time window dimension to form a sensitivity correlation matrix. The sensitivity correlation matrix of the first Line 1 Column elements Indicates drainage unit The effect of environmental response coefficient change on the first The degree of influence of the deviation characteristic quantity within a time window, with the dimension being the ratio of flow rate units to coefficient units.

[0150] The deviation characteristic vector and the sensitivity correlation matrix are used to calculate the correction amount of the environmental response coefficient. A least squares problem is constructed, using the correction amount vector of the environmental response coefficient of each drainage unit. For unknown quantities, use the deviation characteristic vector To establish a system of linear equations for the target residual:

[0151] ;

[0152] When the sensitivity correlation matrix is ​​for an overdetermined or underdetermined system, the regularized least squares method is used to solve it, introducing a regularization coefficient. Control the magnitude of the correction to avoid over-correction that could lead to parameter oscillations. The regularized least squares solution is:

[0153] ;

[0154] in It is the identity matrix. The value of is adaptively adjusted based on the condition number of the sensitivity correlation matrix; it increases when the condition number is large. To enhance numerical stability.

[0155] Correction amount This value is superimposed onto the current environmental response coefficient estimate to complete the first step. Round of iterative updates:

[0156] ;

[0157] The updated environmental response coefficients are substituted into the evapotranspiration intensity field calculation to recalculate the net flow rate for the next time step. The iteration termination condition is the norm of the correction vector. Below the preset convergence threshold Or the number of iterations reaches the maximum allowed number. After the termination condition is met, the converged environmental response coefficients are written back to the evapotranspiration intensity field tensor as the reference parameters for subsequent time step calculations.

[0158] In practical applications, due to differences in underlying surface type, vegetation cover, and road surface materials among different drainage units, the environmental response coefficients of each unit have varying effects on evapotranspiration intensity regulation. By employing the aforementioned mechanism of joint solution of recursive state estimation and sensitivity matrix, the deviations in the evapotranspiration intensity fields of each drainage unit can be dynamically tracked and corrected as the precipitation process continues. This gradually narrows the systematic deviation between calculated and measured flow rates, thereby improving the accuracy of net runoff prediction and providing a more accurate hydrological basis for the generation of subsequent drainage regulation strategies.

[0159] In one optional embodiment, the environmental response coefficient of each drainage unit is subjected to a disturbance increment and recalculated. A sensitivity correlation matrix is ​​constructed based on the calculated flow change, including:

[0160] Apply a perturbation increment to the environmental response coefficient of the target drainage unit;

[0161] The evapotranspiration intensity field is reconstructed using the perturbed environmental response coefficients, and the net flow rate and confluence process of the target drainage unit are recalculated. At the same time, the calculated flow rate changes of the downstream drainage units associated with the confluence transfer matrix are extracted.

[0162] The ratio of the calculated flow change of the target drainage unit to the disturbance increment is used as the direct sensitivity, and the ratio of the calculated flow change of the downstream drainage unit to the disturbance increment is used as the transmission sensitivity.

[0163] Fill the direct sensitivity and the transmitted sensitivity into the sensitivity correlation matrix.

[0164] In one specific implementation, during the construction of the sensitivity correlation matrix, it is necessary to apply a disturbance increment to the environmental response coefficient of each target drainage unit to quantify the impact of parameter changes of that unit on the calculated flow distribution of the entire pipe network. Specifically, for the unit numbered... The target drainage unit, in its current iteration round, has an estimated environmental response coefficient. Based on this, a preset disturbance is superimposed. The environmental response coefficients after disturbance are obtained. Disturbance quantity The value of needs to balance numerical stability and sensitivity calculation accuracy: if the disturbance is too large, the nonlinear characteristics of the evapotranspiration intensity field will lead to a deviation in sensitivity estimation; if the disturbance is too small, the calculated flow rate change will be submerged in numerical noise due to the limitations of floating-point arithmetic precision. In practical applications, Typically, 1% to 5% of the current coefficient estimate is taken as a reference range and kept consistent across each drainage unit to ensure comparability of sensitivity between different units.

[0165] After the perturbation is completed, the target drainage unit is reconstructed using the environmental response coefficients after the perturbation. The evapotranspiration intensity field. Due to... Tensor only the first While the components of a single element change, the evapotranspiration intensity field of the remaining elements remains unchanged. Therefore, the reconstruction process only requires local updates to the target drainage element. This eliminates the need for a full recalculation of the evapotranspiration intensity field for the entire region, effectively reducing computational overhead. The updated evapotranspiration intensity field is then incorporated into the net production flow calculation process to re-obtain the target drainage unit. The net production flow time sequence at each time step. The difference between the net production flow time sequence before and after the disturbance is used as the production flow disturbance signal and input into the flow transfer matrix. Perform convergence calculations to trace the entire process of the disturbance signal propagating downstream along the convergence path. Because... Middle elements Describes the unit Flowing unit The proportional relationship and the corresponding traffic transmission delay After being transmitted through multiple stages of merging, the disturbance signal will reach the target unit. Observable computational flow changes are generated at the outlets of each downstream unit with upstream and downstream topological connections.

[0166] For the target drainage unit It extracts the difference in the program sequence when calculating the flow rate before and after the disturbance at its outlet, within the time window. The average of the differences within the corresponding time period is recorded as the direct calculated flow change of that unit. For those who pass With target unit There are downstream associated drainage units ( Similarly, extract its exit point within the time window. The average value of the calculated flow rate change within the time frame is denoted as the transfer calculated flow rate change. Direct sensitivity is defined as the target cell. The ratio of the direct calculated flow change to the disturbance increment, i.e. Transmission sensitivity is defined as the downstream unit. The ratio of the change in flow rate to the increment of the disturbance, i.e. The above ratios, from the perspective of numerical differentiation, approximately describe the gradient of the impact of changes in the environmental response coefficient on the flow rate at each node of the pipeline network, and are the core basis for subsequently constructing the correction vector.

[0167] Sensitivity Correlation Matrix The row dimension corresponds to the number of each drainage unit, and the column dimension corresponds to the time window number. For the target drainage unit... Its direct sensitivity Fill in The Line 1 List, i.e., order For the target unit Units with downstream topological connections , will transmit sensitivity Fill in cumulatively using an additive method. The Line 1 The reason for using a superposition method is because the unit... It may be located downstream of the confluence path of multiple upstream units at the same time, and its calculated flow change includes the superimposed contribution of disturbances from multiple target units; only after applying disturbances to each target unit one by one and completing the sensitivity calculation can the cumulative results of each row and column fully reflect the comprehensive sensitivity of the unit's flow to the response coefficients of all upstream environments.

[0168] At the engineering implementation level, The computational complexity of applying perturbations to each drainage unit and completing the flow calculation is related to the total number of drainage units. Proportional. To improve computational efficiency, it can be based on... The sparse topology structure groups drainage units, classifying units without upstream or downstream connections into the same batch. Perturbations are applied in parallel and calculated synchronously within each batch, ensuring that perturbation signals do not interfere with each other, thus transforming serial computation into batch parallel computation. Furthermore, for drainage units located at the end of the pipe network with very few downstream connected units, their transmission sensitivity has a limited impact on the entire pipe network. A transmission depth cutoff threshold can be set to track the impact of perturbation signals only within a finite number of confluence transmission levels. Transmission sensitivity beyond the cutoff depth is approximately zero, further reducing the computational load.

[0169] Regarding numerical stability guarantees, when a certain time window When the change in calculated flow rate is extremely small (close to the lower limit of numerical precision), the corresponding sensitivity calculation results may exhibit abnormal amplification or sign instability. To address this, each element can be processed before the sensitivity is entered into the matrix. Perform a validity check: If the absolute value of the calculated flow rate change is lower than the preset lower limit for numerical validity, the corresponding sensitivity element is set to zero to avoid invalid values ​​contaminating subsequent correction calculations. After completing the disturbance calculations for all drainage units, Once constructed, it can be directly used to solve for the environmental response coefficient correction vector using a least-squares regularization framework. This drives the environmental response coefficient to gradually converge to the optimal estimate that matches the measured drainage flow rate during iterative updates.

[0170] A second aspect of the present invention provides an adaptive intelligent control system for highway drainage in humid and hot environments, comprising:

[0171] The parameter acquisition unit is used to acquire the humid and thermal environment parameters and precipitation dynamic parameters of the highway drainage area.

[0172] The evapotranspiration unit is used to construct an evapotranspiration intensity field based on the parameters of the humid and hot environment. The evapotranspiration intensity field and precipitation dynamic parameters are calculated according to the boundary of the drainage unit to obtain the net flow of each drainage unit. The flow confluence process of each drainage unit is calculated through the flow confluence transfer matrix.

[0173] The pipe segment identification unit is used to identify saturated and redundant pipe segments in the drainage network based on the confluence process.

[0174] The regulation and optimization unit is used to obtain the start-up time sequence by optimizing the peak overlap metric index of each drainage unit, so that the peak flow of multiple upstream confluence peaks flowing to the saturated pipe section is staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, the flow transfer path from the saturated pipe section to the redundant pipe section is constructed to generate the drainage regulation strategy.

[0175] The deviation correction unit is used to fuse the measured drainage flow and the calculated flow through a recursive state estimation method, extract the deviation feature of the fused sequence, and use the deviation feature as a correction quantity to iteratively update the environmental response coefficient in the evapotranspiration intensity field.

[0176] The instruction implementation unit is used to convert drainage control strategies into execution instructions for each drainage unit and implement them.

[0177] A third aspect of the present invention provides an electronic device, comprising:

[0178] processor;

[0179] Memory used to store processor-executable instructions;

[0180] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0181] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0182] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive intelligent control method for highway drainage in humid and hot environments, characterized in that, include: Acquire the humid and thermal environmental parameters and precipitation dynamic parameters of the highway drainage area; An evapotranspiration intensity field is constructed based on the parameters of the humid and hot environment. The net runoff of each drainage unit is obtained by calculating the evapotranspiration intensity field and precipitation dynamic parameters according to the boundary of the drainage unit. The runoff process of each drainage unit is calculated by the runoff transfer matrix. Identify saturated and redundant pipe sections in the drainage network based on the confluence process; Using the start-up time of each drainage unit as the decision variable, the start-up time sequence is obtained by optimizing the peak overlap metric index so that the peak values ​​of multiple upstream confluences flowing to the saturated pipe section are staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, a flow transfer path from the saturated pipe section to the redundant pipe section is constructed to generate a drainage control strategy. The measured drainage flow and the calculated flow are sequentially fused by the recursive state estimation method. The deviation feature of the fused sequence is extracted and used as the correction value to iteratively update the environmental response coefficient in the evapotranspiration intensity field. The drainage control strategy is translated into execution instructions for each drainage unit and implemented.

2. The method according to claim 1, characterized in that, An evapotranspiration intensity field is constructed based on the humid and thermal environment parameters. The net runoff of each drainage unit is obtained by calculating the evapotranspiration intensity field and precipitation dynamic parameters according to the boundary of the drainage unit. The runoff process of each drainage unit is calculated through the runoff transfer matrix, including: Temperature and humidity field tensors are extracted from humid and hot environmental parameters. An evapotranspiration potential field is constructed through the mapping relationship of saturated water vapor pressure difference. The evapotranspiration potential field and wind speed field are coupled by tensor to generate an evapotranspiration intensity field with spatial variation characteristics. The evapotranspiration intensity field is spatially divided according to the boundary of the drainage unit, the evapotranspiration intensity distribution matrix in each drainage unit is extracted, and the cumulative evapotranspiration of each drainage unit is obtained by time-domain integration of the evapotranspiration intensity distribution matrix. The precipitation dynamic parameters are spatially matched according to the boundary of the same drainage unit, the cumulative precipitation of each drainage unit is extracted, and the difference between the cumulative precipitation and the cumulative evapotranspiration of the corresponding drainage unit is calculated to obtain the net runoff of each drainage unit. A flow transfer matrix is ​​constructed based on the topological connection relationship of each drainage unit. The net flow of each drainage unit is iteratively calculated through the flow transfer matrix to obtain the flow time history curve of the outlet of each drainage unit as the flow transfer process.

3. The method according to claim 2, characterized in that, A flow confluence matrix is ​​constructed based on the topological connections of each drainage unit. The net flow rate of each drainage unit is iteratively calculated using the flow confluence matrix to obtain the flow time history curves of the outlets of each drainage unit. The flow confluence process includes: Traverse the outlet nodes of each drainage unit and the inlet nodes of adjacent drainage units, identify drainage unit pairs with water flow connectivity, and record the connection relationship of drainage unit pairs as a topology connection index. The water flow propagation time is calculated based on the confluence path length and terrain slope of each drainage unit, and then converted into a time delay step. Using the topology connection index as the matrix coordinate and the time delay step as the matrix element value, a flow transfer matrix is ​​constructed, and the water retention coefficient of each drainage unit is set at the diagonal position of the matrix. The net flow of each drainage unit is organized into an initial flow vector. The flow transfer matrix is ​​multiplied with the initial flow vector to obtain the flow distribution vector at the first time step. The flow distribution vector is used as input and iteratively calculated with the confluence transfer matrix until the difference between the flow distribution vectors of adjacent time steps satisfies the convergence condition. Extract the flow components of each drainage unit outlet from the flow distribution vector at each time step, and organize them into a flow time history curve in chronological order.

4. The method according to claim 2, characterized in that, Using the start-up time of each drainage unit as the decision variable, the start-up time sequence is obtained by optimizing the peak overlap metric index to ensure that the peak flows from multiple upstream confluences to the saturated pipe section are staggered on the time axis. Simultaneously, based on the remaining water flow capacity of the pipe section transmission path, a flow transfer path from the saturated pipe section to the redundant pipe section is constructed, generating drainage control strategies including: Extract the set of upstream drainage units flowing to each saturated pipe section, and identify the peak flow time and peak duration from the flow time history curves of each upstream drainage unit; The start-up time of each drainage unit is used as a decision variable, and multiple candidate start-up time sequences are initialized in the solution space. For each candidate start-up time sequence, the flow time history curves of each upstream drainage unit are shifted along the time axis according to the corresponding start-up time and superimposed moment by moment to obtain the inflow time history curve of the saturated pipe section. From the inflow time history curve, the peak overlap metric is statistically analyzed. The distribution pattern of the start time corresponding to the minimum value of the peak overlap metric is extracted to determine the search direction guidance information. Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner until convergence, and the start time sequence is output. Identify the upstream node location of each saturated pipe segment and the inlet node location of each redundant pipe segment in the drainage pipe network; Search for the set of pipe segment transmission paths connecting the upstream node and the inlet node in the topological connection structure of the drainage pipe network, calculate the remaining water carrying capacity of each pipe segment transmission path, allocate the overload flow of each saturated pipe segment according to the remaining water carrying capacity, and construct the flow transfer path. The startup time sequence is combined with the flow transfer path to generate a drainage control strategy.

5. The method according to claim 4, characterized in that, Peak overlap metrics are statistically analyzed from the inflow time history curve. The distribution pattern of the start time corresponding to the minimum value of the peak overlap metric is extracted to determine the search direction guidance information. Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner until convergence. The output start time sequence includes: Scan the inflow time history curve every moment to identify the time period when the flow rate exceeds the threshold of the saturated pipe section's water supply capacity. The cumulative flow during the period exceeding the threshold is obtained by performing time-domain integration on the flow value and serving as a peak overlap metric. The minimum value is identified from the peak overlap metric index corresponding to each candidate start time sequence, and the distribution pattern of the start time of each drainage unit in the candidate start time sequence corresponding to the minimum value is extracted to obtain search direction guidance information; Based on the search direction guidance information, other candidate start time sequences are adjusted in a directional manner so that the adjusted start time distribution pattern is closer to the search direction guidance information. For the adjusted candidate start time sequence, re-execute the time axis translation and overlay operation, and update the peak overlap metric. The iteration terminates when the change in the minimum value of the peak overlap metric is less than the preset convergence threshold, and the candidate start time sequence with the minimum peak overlap metric is output as the start time sequence.

6. The method according to claim 1, characterized in that, The measured drainage flow and the calculated flow are sequentially fused using a recursive state estimation method. The deviation characteristics of the fused sequence are extracted, and these deviation characteristics are used as correction factors to iteratively update the environmental response coefficients in the evapotranspiration intensity field. The measured and calculated drainage flows of each drainage unit are collected at time steps to form observation sequence pairs; Read the measured and calculated flow rates of the previous time step, calculate the predicted flow rate of the current time step based on the recursive propagation law, and convert the deviation between the predicted flow rate and the observed sequence pair into a correction amount and add it to the predicted flow rate to obtain the fused sequence. The bias features are extracted from the fusion sequence by segmentation according to time windows; The environmental response coefficient of each drainage unit is subjected to a disturbance increment and recalculated. A sensitivity correlation matrix is ​​constructed based on the calculated flow change. The correction amount for the environmental response coefficient is obtained by calculating the deviation characteristic quantity and the sensitivity correlation matrix; The correction is added to the environmental response coefficient to complete the iterative update, and the net production flow for the next time step is recalculated.

7. The method according to claim 6, characterized in that, The environmental response coefficients of each drainage unit are subjected to disturbance increments and recalculated. A sensitivity correlation matrix is ​​constructed based on the calculated flow change, including: Apply a perturbation increment to the environmental response coefficient of the target drainage unit; The evapotranspiration intensity field is reconstructed using the perturbed environmental response coefficients, and the net flow rate and confluence process of the target drainage unit are recalculated. At the same time, the calculated flow rate changes of the downstream drainage units associated with the confluence transfer matrix are extracted. The ratio of the calculated flow change of the target drainage unit to the disturbance increment is used as the direct sensitivity, and the ratio of the calculated flow change of the downstream drainage unit to the disturbance increment is used as the transmission sensitivity. Fill the direct sensitivity and the transmitted sensitivity into the sensitivity correlation matrix.

8. A highway drainage adaptive intelligent control system for hot and humid environments, used to implement the method as described in any one of claims 1-7, characterized in that, include: The parameter acquisition unit is used to acquire the humid and thermal environment parameters and precipitation dynamic parameters of the highway drainage area. The evapotranspiration unit is used to construct an evapotranspiration intensity field based on the parameters of the humid and hot environment. The evapotranspiration intensity field and precipitation dynamic parameters are calculated according to the boundary of the drainage unit to obtain the net flow of each drainage unit. The flow confluence process of each drainage unit is calculated through the flow confluence transfer matrix. The pipe segment identification unit is used to identify saturated and redundant pipe segments in the drainage network based on the confluence process. The regulation and optimization unit is used to obtain the start-up time sequence by optimizing the peak overlap metric index of each drainage unit, so that the peak flow of multiple upstream confluence peaks flowing to the saturated pipe section is staggered on the time axis. At the same time, based on the remaining water carrying capacity of the pipe section transmission path, the flow transfer path from the saturated pipe section to the redundant pipe section is constructed to generate the drainage regulation strategy. The deviation correction unit is used to fuse the measured drainage flow and the calculated flow through a recursive state estimation method, extract the deviation feature of the fused sequence, and use the deviation feature as a correction quantity to iteratively update the environmental response coefficient in the evapotranspiration intensity field. The instruction implementation unit is used to convert drainage control strategies into execution instructions for each drainage unit and implement them.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

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

Images