Method for coordinated dispatching and dynamic load distribution of cascade pumping stations based on multi-source water regime feedback

Abstract

This invention relates to the field of pump station scheduling and control technology, and discloses a method for coordinated scheduling and dynamic load allocation of cascade pump station groups based on multi-source hydrological feedback. The method includes: acquiring the rate of change of forebay water level, flow rate, rainfall intensity, and inter-stage transmission delay matrix of each pump station; constructing a feature vector characterizing the inflow pressure of each node; using the delay matrix to obtain the fluctuation feature components of the regulation quantity migrating along the channel; smoothly reconstructing the fluctuation feature components by calling a discrete convolution kernel whose distribution width is inversely proportional to the average water level; determining the load regulation weight based on the overlap integral value of the reconstructed component and the water level change feature; driving unit frequency fine-tuning and online correction of delay parameters. This invention simulates physical dissipation through a command kinetic energy diffusion mechanism, eliminates regulation phase misalignment, suppresses periodic water level oscillations, and ensures the regulation stability of the system under dynamic evolution of the physical environment.

Description

Technical Field

[0001] This invention relates to a method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback, belonging to the field of pumping station scheduling and control technology. Background Technology

[0002] Current cascade pumping station control systems employ local regulation logic based on forebay water level feedback to maintain inter-stage flow balance. However, the inter-stage channels exhibit hydraulic time lag effects, with upstream outflows undergoing long-distance migration to the downstream forebay. This results in phase deviations between regulation commands and flow peaks in the time dimension, leading to periodic water level oscillations and inducing water hammer effects that damage the channel's physical structure. In addition to physical environmental limitations on regulation stability, existing control methods also have shortcomings in handling large-lag nonlinear disturbances. For example, Chinese invention patent CN113422365B discloses a hierarchical distributed collaborative scheduling optimization method for smart pumping station groups in small watersheds. This scheme optimizes the day-ahead scheduling cost of the pumping station group and reduces communication exchange volume through a hierarchical distributed architecture. The core scheduling logic is based on a power supply and demand balance and static flow prediction model, abstracting the pumping stations as discrete power or flow balance nodes. However, it does not investigate the momentum dissipation waveform distortion and dynamic time delay characteristics affected by seasonal roughness evolution during the long-distance migration of flow fluctuations in inter-stage channels.

[0003] Existing solutions that introduce neural network prediction or hydraulic models to perform feedforward compensation are difficult to deploy stably in low-power control units in industrial settings and cannot adapt to the nonlinear characteristics of channel physical properties that change with the seasons. Existing technologies have the following shortcomings: 1. The interstage hydraulic transmission delay exhibits dynamic characteristics with channel siltation, weed growth, and temperature differences, and preset parameters cannot achieve time axis translation and alignment; 2. Sensors are affected by environmental interference, resulting in non-physical water level jumps, and the feedback logic is at risk of misadjustment; 3. The flow fluctuation migration process involves kinetic energy dissipation and waveform distortion, and linear translation schemes are unable to correct command phase mismatch.

[0004] Therefore, how to overcome the large lag nonlinear interference in long-distance channels and achieve physical consistency verification of feedback signals, and achieve smooth transition of cross-level flow under the constraints of conventional control hardware computing power, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback, the method comprising the following steps: Step S1: Obtain the forebay water level change rate, outflow rate, rainfall intensity data, and inter-stage transmission delay matrix of each pumping station in the cascade pumping station group; Step S2: Based on rainfall intensity data and the rate of change of water level in the forebay, construct an inflow feature vector that characterizes the inflow pressure at each node. Step S3: Using the inter-stage transmission delay matrix, the inflow feature vector is converted into a fluctuation feature component that characterizes the migration and evolution of the upstream regulation quantity along the inter-stage channel. Step S4: Call the discrete convolution kernel to perform smooth reconstruction processing on the wave feature components to simulate the physical dissipation of the wave feature components during the migration process of the inter-stage channel. The distribution width of the discrete convolution kernel is inversely proportional to the real-time average water level of the inter-stage channel. Step S5: Calculate the integral value of the overlap between the smoothed and reconstructed fluctuation characteristic component and the rate of change of the forebay water level of this pumping station on the time axis. Step S6: When the overlap integral value is within the preset phase position information interval, determine the load adjustment weight according to the overlap integral value, and generate a frequency conversion adjustment command according to the load adjustment weight and the outflow flow to control the units in each level of the pump station to perform fine adjustment of the operating frequency. Step S7: When the overlap integral value is lower than the preset phase correlation threshold, it is determined to be a phase misalignment state, the output value of the frequency conversion adjustment command is locked, and the current inflow fluctuation is absorbed by the storage weight generated by the remaining storage capacity of the adjacent channel. Step S8: Calculate the time delay residual between the predicted time when the fluctuation characteristic component arrives at the next-level pump station and the actual time when the liquid level signal of the next-level pump station changes trend, and perform incremental correction on the inter-stage transmission time delay matrix based on the time delay residual.

[0006] Preferably, step S4 includes: determining the initial parameter set of the discrete convolution kernel based on the Manning roughness coefficient and cross-sectional geometric characteristics of the inter-stage channel; dynamically scaling the initial parameter set according to the collected real-time average water level during each control cycle, so that the diffusion width of the discrete convolution kernel increases as the water depth decreases; and performing discrete convolution operation on the wave characteristic components using the dynamically scaled discrete convolution kernel to transform the discrete momentum of the upstream regulation into a continuous kinetic energy distribution feature with a waveform envelope.

[0007] Preferably, the process of determining the load regulation weight includes: using the overlap integral value as the coupling ratio factor between the regulation amount and the inflow load; calculating the remaining inflow impact after the inflow characteristic vector is regulated by the inter-stage channel; and combining the coupling ratio factor and the remaining inflow impact, using a preset nonlinear proportional function to generate the frequency conversion regulation weight for the next-stage pumping station.

[0008] Preferably, in step S1, the forebay water level change rate is checked for water balance before participating in the construction of the inflow feature vector. The water balance check includes: obtaining the real-time outflow of the upstream pumping station, the predicted inflow of the local pumping station, and the liquid level integral increment of the interstage channel; calculating the water balance deviation between the cumulative flow difference between adjacent pumping stations and the liquid level integral increment; when the flow increment corresponding to the forebay water level change rate and the water balance deviation are mismatched in physical magnitude, the water level signal is determined to be abnormal, and the change range of the load regulation weight is limited.

[0009] Preferably, the process of correcting the inter-stage transmission delay matrix further includes: controlling the first stage... The pumping station outputs a micro-disturbance characteristic wave signal with specific frequency characteristics; in the first stage... The arrival time of the micro-disturbance characteristic wave signal is extracted from the pump station; the real-time transmission delay is calculated using the emission and arrival times of the micro-disturbance characteristic wave signal, and the corresponding elements in the inter-stage transmission delay matrix are updated.

[0010] Preferably, the process of determining the load adjustment weight introduces a verification step based on mass conservation: when the rate of change of the forebay water level exceeds the preset safety threshold range, the load adjustment weight is locked to a fixed value based on the local forebay water level feedback.

[0011] Preferably, in step S6, the load allocation process of the units in each level of pumping station includes: acquiring the output current, voltage and torque characteristics of the frequency converter of each pumping station unit, and evaluating the input power fluctuation rate of each unit; establishing the correspondence between the input power fluctuation rate and the operating efficiency, and identifying the high-efficiency operating range boundary of each unit under the current operating conditions; adjusting the adaptive allocation factor to preferentially allocate the load increment to the units within the high-efficiency operating range boundary.

[0012] Preferably, the water balance deviation The calculation follows these rules: ,in, This is due to water balance deviation; This refers to the real-time outflow rate of the upstream pumping station; This is the predicted inflow rate for this level of pumping station; This is the integration time step; The bottom area of ​​the inter-stage channel; This represents the integral increment of the liquid level.

[0013] Preferably, the process of fine-tuning the operating frequency of the unit includes: converting the load regulation weight into the target regulation rate of the frequency converter; and fine-tuning the operating frequency of the unit step by step according to the preset time step so that the pumping flow of the pumping station matches the peak value of the inflow pressure at the time when the fluctuation characteristic component arrives.

[0014] Preferably, the method further includes: using the shift register of the controller to store the fluctuation feature components; during the translation of the fluctuation feature components along the time axis, calling the discrete convolution kernel corresponding to the real-time average water level of the inter-channel in real time, so as to realize the dynamic matching of the adjustment command to the evolution of the channel physical environment.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In a cascade pumping station group with multi-source water information feedback, an inflow feature vector containing rainfall intensity and water level change rate is constructed. Combined with a preset hydraulic transmission time delay matrix, the upstream feedback signal is shifted and aligned on the time axis, so that the load regulation command and the flow peak after long-distance migration coincide in the time dimension. This eliminates the phase mismatch between the regulation command and the actual flow field caused by hydraulic time delay, suppresses periodic water level oscillations, and maintains the pumping unit operating in a stable operating range.

[0016] 2. Using discrete convolution kernels to spread the logical momentum feature packet representing the fluid kinetic energy characteristics, simulate the kinetic energy dissipation and waveform distortion characteristics of the pressure wave generated by the control command during the channel migration process; calculate the overlap integral value between the spread feature packet and the real-time water level change characteristics of the pump station, dynamically scale the load adjustment weight, adjust the physical coupling of energy and inflow impact force, and improve the dynamic adjustment accuracy of the large lag nonlinear system.

[0017] 3. Based on the feedback source audit mechanism of spatial flux conservation, compare the flow meter signals of adjacent pumping stations between stages, the integral of liquid level changes in inter-stage channels and the predicted inflow increment; perform physical consistency verification before calculating load adjustment weights, identify and intercept non-physical water level jump signals caused by sensor hardware failures or environmental noise, and ensure the determinism and security of control command output under abnormal operating conditions of the sensing layer. Attached Figure Description

[0018] Figure 1 This is the main control flowchart for the coordinated scheduling and dynamic load allocation of the cascade pumping station group in this invention; Figure 2 This is a time-domain response and overlapping integral curve of the wave characteristic components and water level change characteristics of the present invention; Figure 3 This is a schematic diagram of the system logic functional architecture and module interaction based on multi-source hydrological feedback of the present invention. Detailed Implementation

[0019] This section is intended to describe the present invention in detail. The following embodiments are used to explain and illustrate the present invention, and are not intended to limit the scope of protection of the present invention.

[0020] This invention provides a method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback. It comprises a multi-source information sensing module, a spatiotemporal coordinate mapping logic, a command kinetic energy dissipation reconstruction unit, and a load weight allocation closed loop. The multi-source information sensing module acquires parameters such as water level change rate, outflow rate, and regional rainfall intensity. The spatiotemporal coordinate mapping logic performs time axis translation processing on the feedback signal using the inter-stage transmission delay matrix. The command kinetic energy dissipation reconstruction unit uses a discrete convolution kernel with a distribution width inversely proportional to the average water level to simulate the waveform distortion of the regulation amount migrating along the channel. The load weight allocation closed loop determines the unit frequency fine-tuning command based on the overlap integral value of the fluctuation characteristic components and the water level characteristics, and absorbs the current inflow fluctuation using the storage weight generated from the remaining storage capacity of adjacent inter-stage channels. Based on the construction procedure of the inflow characteristic vector, the system acquires the forebay water level change rate. and real-time rainfall intensity Then, the real-time rainfall intensity is calculated using a preset catchment area coefficient. Converted into predicted runoff components, and the rate of change of forebay water level based on the cross-sectional area of ​​the inter-stage channels. The predicted runoff flow component is converted into the measured inflow fluctuation component. By performing a weighted summation operation on the predicted runoff flow component and the measured inflow fluctuation component, an initial feature vector representing the current node's overall inflow pressure is constructed. This construction procedure corrects the perception and recognition bias of a single hydraulic sensor under sudden runoff conditions by fusing rainfall feedforward information and water level feedback information. The initial feature vector, Real-time rainfall intensity, in mm / h. This represents the rate of change of water level in the forebay, expressed in m / s.

[0021] The system verifies the authenticity of the signal before constructing the feature vector, that is, by calculating the water balance deviation between the cumulative flow difference between adjacent pumping stations and the change in water storage capacity within the canal section. Water balance deviation The calculation follows these rules: ,in, This represents the water balance deviation, in meters. 3 ; This represents the real-time outflow rate of the upstream pumping station, in cubic meters per second (m³). 3 / s; The predicted inflow rate for this level of pumping station is expressed in m³. 3 / s; This is the integration time step, in seconds. The bottom area of ​​the inter-stage channel is expressed in m². 2 ; This represents the integral increment of the liquid level, in meters; when the water balance deviation... When the water level change rate is within the preset tolerance range, determine the rate of change. The signal is valid; if the measured flow rate increment deviates from the water balance... If a mismatch occurs in the quantity, the system will lock the frequency converter control commands of this pumping station until the signal regains physical and logical consistency; during long-term operation, when a water balance deviation is detected in real time... When the mean square error of the signal exhibits a unidirectional increasing trend over time, it is determined that the Manning roughness coefficient of the physical channel has undergone environmental evolution. At this time, the system uses a preset inter-stage transmission delay matrix to shift the time axis of the upstream feedback signal and introduces an execution residual reverse mapping mechanism to calculate the time delay residual value between the predicted time when the predicted inflow characteristic quantity arrives at the next-stage pumping station and the actual time when the liquid level signal of the next-stage pumping station changes trend. Based on this time delay residual value, the inter-stage transmission delay matrix is ​​incrementally corrected. In addition, the control unit periodically drives the first stage... The first-stage pump station outputs a micro-disturbance characteristic wave signal with specific frequency characteristics, which is generated by the first... The pumping station identifies the arrival time of this characteristic wave and uses this information to calculate the real-time transmission delay under the current operating conditions. This eliminates adjustment phase errors caused by channel siltation or seasonal roughness evolution, maintaining the stability of the system under dynamic physical environment conditions.

[0022] To address the nonlinear dissipation of the control pressure wave during its migration, the system executes waveform elastic mapping logic based on the conservation of command kinetic energy before calculating the load adjustment weights. The control unit then adjusts the frequency regulation increment of the upstream pumping station. Convert to logical momentum value It follows the rules ,in, For logical momentum value; The momentum conversion coefficient is preset based on the cross-sectional geometric features; The frequency adjustment rate is expressed in Hz / s; the system calls the discrete convolution kernel. The wave characteristic components are smoothly reconstructed to simulate the physical dissipation characteristic that waveform diffusion is more severe as water depth decreases. This step utilizes a shift register within the control unit to store the wave characteristic components and calls them in real time during the translation process, taking into account the current average water level. The corresponding discrete convolution parameter set converts discrete momentum into continuous kinetic energy distribution characteristics with waveform envelopes. The initial distribution width of the discrete convolution kernel is identified based on the Manning roughness coefficient of the inter-stage channel and the hydraulic radius of the cross-section. Establish the scaling law of kernel function ,in To design a normal water level, real-time average water level is retrieved within the sampling period. Dynamically update discrete convolution kernels Parameters, using this kernel function to process the logical momentum sequence Perform temporal convolution operations The waveform spread spectrum and kinetic energy dissipation distortion generated by the simulated energy under the condition of low water level and strong friction are used to reconstruct the upstream step regulation command into a kinetic energy distribution envelope that conforms to the wave dynamics characteristics, and to compensate for the command perception distortion caused by the channel geometric nonlinearity.

[0023] When determining the phase misalignment state, the system extracts the smoothed and reconstructed fluctuation feature component sequence and the real-time water level change feature sequence of the local pumping station. By executing a normalized cross-correlation algorithm, the similarity coefficient between the two time series within the current sampling window is calculated, and this similarity coefficient is used as the overlap integral value. When participating in logical decision-making, the overlapping integral value When the phase correlation is below a preset threshold, the system determines that there is a phase mismatch between the current adjustment command and the physical flow peak. It then locks the frequency converter control command and switches to the regulation weight generated by the remaining storage capacity of the inter-stage channels to absorb the current fluctuation. This determination procedure uses the quantitative offset of the similarity coefficient to replace the subjective judgment of waveform alignment. The value of the overlap integral. To determine the real-time average water level, the lower limit of the phase position information is calibrated based on the gradient disturbance experiment of the inter-stage channel cross-section. The value is 0.65. A shift register is used to extract the smoothed and reconstructed fluctuation feature sequence within the current sampling window. Water level change characteristic sequence Perform normalized cross-correlation operation ,like In the range The determination of phase matching between the adjustment command and the physical flow field is mapped. To adjust the weighting factor, if This triggers the phase misalignment protection logic, locks the current inverter frequency output, calls the inter-stage channel storage margin Crem to intervene in flow fluctuation absorption, and calculates the predicted time through the residual inverse mapping mechanism. With actual time deviation Perform incremental calibration of the delay matrix.

[0024] The load distribution process obtains the output current of each unit's frequency converter. ,Voltage And torque characteristics, by establishing input power fluctuation rate and operating efficiency The system identifies the boundary of the efficient operating range of the units under the current operating conditions through mapping relationships. It adjusts the adaptive allocation factor to prioritize the allocation of load increments to units within the boundary of the efficient operating range, and guides the pumping station cluster to the optimal system performance state by fine-tuning the inter-stage water level target value. Through deep coupling of the water balance verification layer, time delay self-calibration layer, kinetic energy dissipation reconstruction layer, and energy efficiency optimization allocation layer, the system constructs a control system covering physical consistency at the perception layer, waveform consistency at the logic layer, and phase consistency at the execution layer. This enables smooth flow transition of the cascade pumping station group under complex operating conditions. This refers to the output current, measured in amperes (A). Output voltage, in volts (V). For operational efficiency.

[0025] Example 1: In a system containing three pumping stations with an inter-stage channel length of... In the coordinated operation of a cascade pumping station group spanning km, when the upstream No. 1 pumping station is subject to regulation demand, the outflow rate will be adjusted accordingly. Depend on m 3 / s adjustment to m 3 At a rate of / s, the water level in the forebay of the downstream No. 2 pumping station is stable before the fluctuation signal migrates to this location via long-distance water transport. The system uses a multi-source information sensing module to obtain the outflow rate of the No. 1 pumping station. Changes and real-time rainfall intensity The system constructs an inflow feature vector characterizing the inflow pressure at the nodes and calls the transmission delay determined in the inter-stage transmission delay matrix. The feedback signal is shifted along the time axis to predict the time when the peak flow of pump station 1 will reach pump station 2. To address waveform distortion of the regulation command during long-distance channel migration, the command kinetic energy dissipation reconstruction unit modulates the frequency regulation increment of pump station 1. Convert to logical momentum value The logical momentum value Follow the formula ,in, For logical momentum value; The momentum conversion coefficient is a pre-set value based on the geometric characteristics of the inter-channel section. Frequency adjustment rate, in Hz / s; system call distribution width and real-time average water level. Discrete convolution kernels with inverse proportional relationship For logical momentum values Perform smooth reconstruction to simulate the physical dissipation phenomenon of fluid kinetic energy spreading as water depth decreases, and convert the translated discrete commands into continuous kinetic energy distribution characteristics with physical envelope features.

[0026] When the processed fluctuation characteristic component overlaps with the measured water level change characteristic of the forebay of Pump Station No. 2 on the time axis, the load weight allocation closed-loop calculation of the overlap integral value is performed, and the output current of the frequency converter of the unit in Pump Station No. 2 is obtained. ,Voltage In addition to torque characteristics, the input power fluctuation rate was evaluated to determine the boundary of the unit's efficient operating range. By adjusting the adaptive allocation factor, the system fine-tuned the operating frequency of the No. 2 pumping station unit at the arrival time of the flow peak, so that the increase in pumping capacity matched the upstream inflow peak in phase and amplitude, and the oscillation amplitude of the forebay water level of the No. 2 pumping station was maintained at... Within m, the system utilizes the remaining storage capacity of the inter-stage channels to absorb flow fluctuations, eliminates the phase mismatch between the regulation command and the flow peak, and the entire system operates within the preset energy efficiency range.

[0027] Example 2: At an equivalent length of In a physical test rig for a cascade channel spanning km and including three controlled pumping station units, to address hydraulic waveform distortion caused by long-distance migration and sensor electronic noise interference in the field environment, the test group used submersible hydrostatic level gauges installed in the forebays of each pumping station to acquire water level data. The measurement range of the submersible hydrostatic level gauges was [missing information]. to m, with a precision of %, sampling period To balance the real-time updating of data and the load calculation of the controller, the sampling period is determined by detecting the pressure wave transmission characteristics generated by the flow step of Pump Station No. 1. for s; The test platform superimposes the amplitude value onto the original sensor signal as the measurement range. % random impulse interference, and construct a signal-to-noise ratio of %. dB Gaussian background noise, by controlling the output flow of the upstream pumping station exist s from m 3 / s adjustment to m 3 The system response is triggered by a / s interval. The experiment compares the response accuracy of a control group implementing local water level feedback, a control group implementing linear time delay compensation, and an experimental group implementing the complete procedure of this method under different disturbance intensities. The water balance deviation is also considered. Follow the formula ,in, This represents the water balance deviation, in meters. 3 ; This represents the real-time outflow rate of the upstream pumping station, in cubic meters per second (m³). / s; The predicted inflow rate for this level of pumping station is expressed in m³. 3 / s; This is the integration time step, in seconds. The bottom area of ​​the inter-stage channel is expressed in m². 2 ; The value represents the integral increment of the liquid level, in meters (m). The experimental group used water balance verification logic to identify and eliminate injected non-physical jump signals, even with increased background noise power. Under conditions where the water level is doubled, the effective judgment rate remains at [percentage missing]. The percentage is above 90%. Table 1 shows the comparison data of regulation performance under different control methods.

[0028] Table 1: Comparison of Pump Station Group Dispatch Response Characteristics Data

[0029] During the test, the system executed the command kinetic energy dissipation reconstruction logic, adjusting the frequency of pump station No. 1 by increment. Convert to logical momentum value It follows the rules and call discrete convolution kernels Perform smooth reconstruction on it, where discrete convolution kernels The distribution width varies with the real-time average water level Decrease and increase; at the real-time average water level from m drops to In the nonlinear gradient test of m, when the water level Below At time m, the frictional dissipation at the bottom of the channel increases, leading to an increase in the overlap integral value. Descending to If the correlation threshold is below the threshold, the system determines that it is a phase misalignment and locks the current frequency output. The remaining inflow impact is then absorbed by the remaining storage capacity of the inter-stage channel to avoid premature triggering of the adjustment command caused by waveform diffusion under the linear compensation framework. For logical momentum value; The momentum conversion coefficient is preset based on the cross-sectional geometric features; This is the frequency regulation rate, expressed in Hz / s. It is a discrete convolution kernel; This is the real-time average water level, in meters (m). The value is the overlap integral; the test group monitors the output current of the frequency converters of each pump station unit. ,Voltage In addition to torque characteristics, the power input fluctuation rate of each unit is calculated and the operating efficiency is identified. High-efficiency operating range boundary; unit operating point entry efficiency is lower than When the vibration sensitivity zone is within a certain percentage, the power fluctuation characteristic is changed by... % surged to %, the adaptive allocation factor will be based on this feature offset % of the load increment is redistributed to parallel units operating within their high-efficiency range, resulting in an overall average unit efficiency of [missing information]. Improved compared to the control group % This refers to the output current, measured in amperes (A). Output voltage, in volts (V). For operational efficiency, gradient data obtained from the physical simulation platform confirms the stability of the regulation logic in handling large time-delay nonlinear flow fields. The synergistic effect of water balance verification and electrical parameter allocation mechanism enables the system to achieve smooth transition of cross-level flow rates even in the presence of measurement noise and changes in the physical environment.

[0030] Example 3: This example combines Figures 1 to 3 This section explains the method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source water information feedback, such as... Figure 1 As shown, in step S1, the forebay water level change rate, outflow rate, and rainfall intensity are collected through the multi-source data acquisition module, and preset parameters or dynamically updated interstage transmission delay matrices are also acquired. In step S2, an inflow feature vector representing the node inflow pressure is constructed based on the rainfall intensity and water level change rate. In step S3, the inflow feature vector is converted into a fluctuation feature component representing the upstream regulation and its migration and evolution along the interstage channel using the interstage transmission delay matrix. In step S4, a discrete convolution kernel with a distribution width inversely proportional to the average water level is used to perform smoothing and reconstruction processing on the fluctuation feature component. In step S5, the smoothed fluctuation feature component is calculated and compared with the water level in the forebay of this pumping station. The system determines whether the overlapping integral value of the phase change rate on the time axis is within the preset phase position information interval. If it is, step S6 is executed to determine the load adjustment weight and generate a frequency conversion adjustment command to drive the unit frequency fine adjustment. Otherwise, step S7 is executed to process the phase misalignment, lock the command, and use the storage weight generated by the remaining storage capacity of the adjacent channel to absorb the current inflow fluctuation. In addition, the process uses the residual data generated in step S5 and subsequent steps to calculate the time delay residual value in step S8 and perform incremental correction on the inter-stage transmission time delay matrix. The updated parameters are fed back to step S1 for matrix configuration in the next control cycle.

[0031] like Figure 2As shown, the left vertical axis represents the magnitude of the characteristic value, ranging from 0 to 1.0, and the right vertical axis represents the magnitude of the overlap integral value, ranging from 0 to 0.8. The solid line in the figure represents the wave characteristic component after smoothing and reconstruction, and its waveform shows a continuous trend of first increasing and then decreasing. The dashed line represents the water level change characteristic, and its trajectory closely follows the wave characteristic component on the time axis, exhibiting a similar envelope shape. The dotted line represents the overlap integral value, which increases with the synchronicity between the wave characteristic component and the water level change characteristic, reaching a peak near the 5th control cycle, reflecting the degree of matching between the regulation command and the flow field response in the time dimension. Figure 3 As shown, the overall functional architecture of the system consists of five logical modules: data perception, core algorithm, control strategy, feedback optimization, and target performance. The data perception module is responsible for collecting data on the rate of change of water level in the forebay, multi-source outflow, and rainfall intensity, and generating nodal inflow pressure characteristics. The core algorithm module performs fluctuation feature component conversion, discrete convolution smoothing reconstruction, and overlapping integral value calculation. The control strategy module performs phase matching judgment based on the algorithm results, determines the load adjustment weight, and generates frequency conversion commands. The feedback optimization module performs incremental correction of the time delay matrix based on the time delay residual calculation results. The target performance module aims to reduce the cumulative effect of time delay and improve the scheduling response speed. All modules work together to ultimately support the main functions of coordinated scheduling and dynamic load allocation of cascade pumping station groups, so as to achieve efficient, stable, and low-latency operation of the system.

[0032] Example 4: When the system faces a situation where the channel roughness coefficient shifts due to seasonal siltation, the inter-stage transmission delay matrix... With physical flow field wave velocity generation The phase deviation of s; to correct the phase deviation, the control unit activates the momentum conversion coefficient. The online calibration procedure involves calculating the current real-time average water level by reading the geometric parameters of the channel cross-section. The cross-sectional area of ​​the water flow below The momentum conversion coefficient Determined as the cross-sectional area of ​​the water passage With gravitational acceleration The square root of the product, the momentum conversion coefficient As logical momentum value The scaling factor, the system executes discrete convolution kernels Dynamic scaling based on real-time average water level With the design normal water level The ratio determines the convolution distribution width. Convolution distribution width The calculation follows the formula: ,in, The width of the convolution distribution; The baseline width coefficient; The design water level is in meters (m). This is the real-time average water level, in meters (m). When the real-time average water level... When reduced, the convolution distribution width To increase the time delay, the system uses a residual reverse mapping mechanism to update the inter-stage transmission delay matrix. The control unit captures the actual moment when the extreme point of the liquid level signal occurs. And the predicted time of arrival of the fluctuation characteristic component. Compare and calculate the time delay residual value. Updated latency parameters Follow the formula below: ,in, This is the updated transmission latency; The transmission delay at the current moment; The convergence damping factor is set to a value of [value to be filled in]. ; The values ​​are time delay residuals, all in seconds; Table 2 shows the system's time delay residuals. Time delay correction and water level deviation data within each control cycle.

[0033] Table 2: Example Table of Data for Parameter Adaptive Correction Process

[0034] Recursive correction After one cycle, the time delay residual value Reduced to s, delay parameter Dynamic updates eliminate phase shifts caused by changes in the physical environment, reducing the peak water level fluctuations at Pump Station No. 2 from... m decreased to m, the system maintains consistency between the adjustment command and the physical causal chain of the actual flow field.

[0035] Example 5: In a length of A standardized initial calibration process was deployed for the newly constructed inter-stage canal section (km long), by controlling the output flow gradient of the upstream pumping station units. m 3 / s、 m 3 / s and m 3 A stepped signal of / s is used to synchronously record the time span of each gradient flow wave propagating to the downstream forepool, thereby filling the interstage transmission delay matrix. The static reference data; the process uses level gauges to capture the arrival time of the wavefront, and combines this with the measured roughness coefficient of the channel section to verify the wave velocity transmission characteristics, limiting the phase deviation of the spatiotemporal coordinate mapping logic in the initial operation phase to within... Within s, and for discrete convolution kernels The initial parameters are identified by analyzing the fluctuation diffusion characteristics under the flow gradient, and the baseline width coefficient is determined. The calibration is a fixed proportional multiple of the hydraulic radius of the current canal section, where, For inter-stage transmission delay matrix, For discrete convolution kernels, This is the baseline width coefficient.

[0036] During the full-load joint commissioning process, the control unit presets the momentum conversion coefficient based on the cross-sectional geometry of the pumping station. Momentum conversion coefficient Set as the cross-sectional area of ​​the water passage With gravitational acceleration The square root of the product, combined with frequency adjustment increments. Verify logical momentum value To verify the mapping accuracy, and to validate the stability of the eigenvectors of the multi-source hydrological feedback flow, the system simulates rainfall confluence by setting up water level disturbance sources within the canal section, and observes the effect of load regulation weights on the rate of water level change. The response during mutations was compared by different convergence damping factors. The convergence rate at a given value will affect the convergence damping factor. Set as To balance adjustment accuracy and controller computational load, the system performs baseline alignment of channel physical dissipation characteristics and command distortion characteristics before entering cross-level collaborative scheduling. The momentum conversion coefficient, The cross-sectional area of ​​the water passage is expressed in m². 2 , This is the acceleration due to gravity, measured in m / s². 2 , This represents the frequency adjustment increment, measured in Hz. For logical momentum value, This represents the rate of change of water level, expressed in m / s. This is the convergence damping factor.

[0037] Example 6: When performing pre-deployment calibration in a cascade pumping station group containing four variable frequency pump sets, the system adjusts the frequency adjustment increment in the frequency step fine-tuning method. Perform physical response testing by controlling a single unit from the lower frequency limit. The frequency increases at a constant rate until it reaches the upper limit. Hz, while simultaneously collecting the vibration amplitude output from the unit's vibration sensor. To identify the peak range that generates mechanical resonance; the control unit adjusts the frequency increment based on the measured unit vibration characteristics. Set as Hz, frequency adjustment increment The magnitude of the regulation pressure wave satisfies the constraint condition that the pressure wave is within the mechanical bearing capacity of the channel sidewall. A corresponding time pulse width is set for each regulation step, and the energy output of the regulation command is maintained within the unit's mechanical safety boundary using calibration procedures. This represents the frequency adjustment increment, measured in Hz. The value represents the vibration amplitude, expressed in mm.

[0038] When determining the safe capacity boundary of the virtual buffer zone, the control unit obtains the highest safe operating water level of the canal section. Compared with the current real-time average water level The physical drop will reduce the remaining storage capacity. Converted into quantified water level redundancy height; the system sets locking logic for the illusory buffer, which is triggered when the predicted water level rise is caused by inflow load. Greater than the difference between the embankment crest elevation and the designed wave wall height in the canal section. When the load adjustment weight is %, the locking decision logic will adjust the load adjustment weight. The rate of change is limited to Within each sampling period, the rate of change is determined according to the law of conservation of fluid mass to reduce the probability of overflow. During the reset phase after the flow peak passes, the weight of old data is reduced using a time decay factor. Dynamic deviations in the logical calculation of the imaginary buffer are compensated by periodically identifying the inter-stage transmission delay matrix online. The system utilizes a boundary verification process to construct an online fault-tolerant mechanism to cope with extreme water condition disturbances. This is the highest safe operating water level, measured in meters (m). This is the real-time average water level, in meters (m). The remaining storage capacity is expressed in meters (m). 3 , To predict the water level rise, the unit is meters (m). This is the load adjustment weight.

[0039] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback, characterized in that, The method includes the following steps: Step S1: Obtain the forebay water level change rate, outflow rate, rainfall intensity data, and inter-stage transmission delay matrix of each pumping station in the cascade pumping station group; Step S2: Based on rainfall intensity data and the rate of change of forebay water level, construct an inflow feature vector that characterizes the inflow pressure at each node. Step S3: Using the inter-stage transmission delay matrix, the inflow feature vector is converted into a fluctuation feature component that characterizes the migration and evolution of the upstream regulation quantity along the inter-stage channel. Step S4: Call the discrete convolution kernel to perform smooth reconstruction processing on the wave feature components to simulate the physical dissipation of the wave feature components during the migration process of the inter-stage channel. The distribution width of the discrete convolution kernel is inversely proportional to the real-time average water level of the inter-stage channel. Step S5: Calculate the integral value of the overlap between the smoothed and reconstructed fluctuation characteristic component and the rate of change of the forebay water level of this pumping station on the time axis. Step S6: When the overlap integral value is within the preset phase position information interval, determine the load adjustment weight according to the overlap integral value, and generate a frequency conversion adjustment command according to the load adjustment weight and the outflow flow to control the units in each level of the pump station to perform fine adjustment of the operating frequency. Step S7: When the overlap integral value is lower than the preset phase correlation threshold, it is determined to be a phase misalignment state, the output value of the frequency conversion adjustment command is locked, and the current inflow fluctuation is absorbed by the storage weight generated by the remaining storage capacity of the adjacent channel. Step S8: Calculate the time delay residual between the predicted time when the fluctuation characteristic component arrives at the next-level pump station and the actual time when the liquid level signal of the next-level pump station changes trend, and perform incremental correction on the inter-stage transmission time delay matrix based on the time delay residual.

2. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, Step S4 includes: determining the initial parameter set of the discrete convolution kernel based on the Manning roughness coefficient and cross-sectional geometric characteristics of the inter-stage channel; dynamically scaling the initial parameter set according to the collected real-time average water level during each control cycle, so that the diffusion width of the discrete convolution kernel increases as the water depth decreases; and performing discrete convolution operation on the wave characteristic components using the dynamically scaled discrete convolution kernel to transform the discrete momentum of the upstream regulation into a continuous kinetic energy distribution feature with a waveform envelope.

3. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, The process of determining the load regulation weight includes: using the overlap integral value as the coupling ratio factor between the regulation amount and the inflow load; calculating the remaining inflow impact after the inflow characteristic vector is regulated by the inter-stage channel; and using a preset nonlinear proportional function to generate the frequency conversion regulation weight for the next-stage pumping station by combining the coupling ratio factor and the remaining inflow impact.

4. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, In step S1, the forebay water level change rate undergoes a water balance check before participating in the construction of the inflow feature vector. The water balance check includes: obtaining the real-time outflow of the upstream pumping station, the predicted inflow of the local pumping station, and the liquid level integral increment of the interstage channel; calculating the water balance deviation between the cumulative flow difference between adjacent pumping stations and the liquid level integral increment; and determining that the water level signal is abnormal when the flow increment corresponding to the forebay water level change rate and the water balance deviation are mismatched in terms of physical magnitude, and limiting the change range of the load regulation weight.

5. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, The process of correcting the inter-stage transmission delay matrix also includes: controlling the first stage... The pumping station outputs a micro-disturbance characteristic wave signal with specific frequency characteristics; in the first stage... The arrival time of the micro-disturbance characteristic wave signal is extracted from the pump station; the real-time transmission delay is calculated using the emission and arrival times of the micro-disturbance characteristic wave signal, and the corresponding elements in the inter-stage transmission delay matrix are updated.

6. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, The process of determining the load regulation weight introduces a verification step based on mass conservation: when the rate of change of the forebay water level exceeds the preset safety threshold range, the load regulation weight is locked to a fixed value based on the local forebay water level feedback.

7. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, In step S6, the load allocation process of the units in each level of pumping station includes: acquiring the output current, voltage and torque characteristics of the frequency converter of each pumping station unit, and evaluating the input power fluctuation rate of each unit; establishing the correspondence between the input power fluctuation rate and the operating efficiency, and identifying the high-efficiency operating range boundary of each unit under the current operating conditions; adjusting the adaptive allocation factor to preferentially allocate the load increment to the units within the high-efficiency operating range boundary.

8. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 4, characterized in that, Water balance deviation The calculation follows these rules: ,in, This is due to water balance deviation; This refers to the real-time outflow rate of the upstream pumping station; This is the predicted inflow rate for this level of pumping station; This is the integration time step; The bottom area of ​​the inter-stage channel; This represents the integral increment of the liquid level.

9. A method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, The process of fine-tuning the operating frequency of the unit includes: converting the load regulation weight into the target regulation rate of the frequency converter; and fine-tuning the operating frequency of the unit step by step according to the preset time step so that the pumping flow of the pumping station matches the peak value of the inflow pressure at the time when the fluctuation characteristic component arrives.

10. The method for coordinated scheduling and dynamic load allocation of cascade pumping station groups based on multi-source hydrological feedback as described in claim 1, characterized in that, The method also includes: using the controller's shift register to store the fluctuation feature components; and during the translation of the fluctuation feature components along the time axis, calling the discrete convolution kernel corresponding to the real-time average water level of the inter-channel in real time.

Images