An intelligent control system for deep well dewatering of hydraulic engineering

By using wellhead benchmark modeling and a graded pump start-up control module, combined with a common-mode calibration and verification module, the problem of inconsistent water level, flow rate, and pressure in deep well dewatering control was solved, achieving stability and calculability, reducing equipment wear risk, and meeting the requirements of construction safety and accountability.

CN122111110AActive Publication Date: 2026-05-29ANHUI SHUIAN CONSTR GRP CO LTD

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
ANHUI SHUIAN CONSTR GRP CO LTD
Filing Date
2026-04-27
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

In existing technologies, the control of deep well dewatering during foundation pit excavation lacks a unified wellhead elevation record, resulting in inconsistent sampling of water level, flow rate, and pressure. This leads to inconsistencies in pump start-up sequence and frequency allocation, increasing the risk of equipment wear. Furthermore, the lack of a traceable chain of evidence makes it difficult to meet the requirements for construction safety and accountability.

Method used

The system employs a wellhead benchmark modeling module, a staged pump start-up control module, a common mode correction verification module, and a rebound exit monitoring module. By unifying the water level elevation through the wellhead reference elevation and obtaining an effective groundwater level elevation through median filtering, an initial benchmark is established. This enables staged pump start-up, amplitude-limited frequency modulation, and well addition/stop control. Furthermore, by constructing a caliber consistency judgment value through inter-group common mode offset and operating well pressure flow evidence, a compensation time shift correction is triggered, forming a calculable evidence chain.

Benefits of technology

It improves the stability of deep well dewatering control, reduces the risk of water hammer and wear, realizes the calculability and auditability of control actions, and ensures construction safety and accountability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122111110A_ABST
    Figure CN122111110A_ABST
Patent Text Reader

Abstract

The application discloses a deep well dewatering intelligent control system for hydraulic engineering, relates to the technical field of deep well dewatering intelligent control of hydraulic engineering foundation pits, and is used for solving the problems of water level caliber common mode offset and regulation and control rearrangement and rebound difficult control caused by atmospheric pressure compensation time sequence misplacement in multi-well linkage; the well mouth reference elevation is taken as a benchmark to convert the underground water level elevation, the effective underground water level elevation is obtained through median filtering, the initial benchmark is established, the target water level and the upper and lower limit thresholds are generated; the pumping flow and the water outlet pressure are synchronously collected, and the graded pump starting and the amplitude limiting frequency modulation are implemented. For the common mode offset caused by the compensation link time sequence misplacement, the caliber consistency judgment value triggering time shift correction and perturbation verification are constructed, and the stable closed loop input is stabilized; the pump is stopped and batched out and the rebound abnormality is monitored.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for deep well dewatering in water conservancy engineering foundation pits, and more specifically, to an intelligent control system for deep well dewatering in water conservancy engineering. Background Technology

[0002] Deep well dewatering is commonly used to lower the groundwater level during foundation pit excavation. However, on-site operations often rely on experience for start-up and shutdown, resulting in scattered wellhead elevation records and inconsistent sampling clocks for water level, flow rate, and pressure. This leads to inconsistent closed-loop control standards and makes it difficult to audit operations. When ventilated and non-ventilated hydrostatic level gauges are used interchangeably, the non-ventilated type requires external atmospheric pressure compensation. When communication delays, sampling alignment deviations, or timestamp binding errors exist, the atmospheric pressure link may cause the water level conversion value to shift in the same direction across several sampling periods. While a single well may still meet the requirements for continuous arrival and reasonable range checks, the water level deviation may be amplified or reduced in multi-well linkage. This causes non-physical rearrangement of pump start-up sequencing and frequency allocation, resulting in decoupling phenomena such as frequent command changes and weak or inconsistent pressure-flow responses, increasing the risk of excessive dewatering, idling, and equipment wear. If there is a lack of phased rhythm and rebound monitoring during pump shutdown, sudden rises in water level or abnormal rebounds are likely to occur, and the lack of a traceable chain of evidence makes it difficult to meet construction safety and accountability requirements.

[0003] To address the above problems, this invention proposes a solution. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide an intelligent control system for deep well dewatering in water conservancy projects to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] In a preferred embodiment, it includes: a wellhead benchmark modeling module, a staged pump start-up control module, a common mode correction and verification module, a rebound exit monitoring module, and signal connections between the modules;

[0007] The wellhead benchmark modeling module is used to set up dewatering wells and external observation wells based on the pit area and excavation control elevation, and to register the well type characteristics table and the initial groundwater level benchmark.

[0008] The graded pump start-up control module is used to perform graded pump start-up after verifying the groundwater level elevation, pumping flow rate, and outlet pressure. It also combines the well type characteristic table to form a control dead zone based on the target water level and upper and lower limit thresholds, and performs frequency regulation, well addition, and well shutdown control.

[0009] The common-mode correction and verification module is used to divide wellpoints into a ventilated wellpoint set and a non-ventilated wellpoint set, and obtain the effective groundwater level elevation for each set. It calculates the inter-group common-mode offset and the common-mode dispersion of the non-ventilated wellpoint set. Simultaneously, it obtains the median pumping flow rate change and the median outlet pressure change for the operating wellpoint set within adjacent sampling periods, and calculates the pressure-flow evidence matching coefficient based on the correspondence between the median pumping flow rate change, the median outlet pressure change, and the inter-group common-mode offset. Finally, it integrates the inter-group common-mode offset, common-mode dispersion, and pressure-flow evidence... After normalization and weighted calculation based on the matching coefficients, a caliber consistency judgment value is generated. This caliber consistency judgment value is used to characterize the degree of consistency between the current water level caliber and the pressure-flow caliber. The smaller the inter-group common mode offset, the smaller the common mode dispersion, and the larger the pressure-flow evidence matching coefficient, the larger the caliber consistency judgment value. When the caliber consistency judgment value is continuously lower than a preset threshold, the compensation time shift is searched and the amplitude is limited for updating. Frequency perturbation is applied to the selected verification well points, and the verification residual is obtained by combining the well type characteristic table. The water level input is switched or frozen and archived based on the verification residual.

[0010] The rebound exit monitoring module is used to lock the reference water level for pump shutdown, reduce the frequency of pump shutdown in batches, close valves for isolation and compensate for pumping capacity, and restart the pump or trigger an alarm if the rebound amount or rebound rate exceeds the limit.

[0011] In a preferred embodiment, the layout of dewatering wells and external observation wells is determined based on the excavation pit range and excavation control elevation, and the reference elevation of each wellhead is measured; groundwater level sensors are installed in each well to collect water pressure and atmospheric pressure is collected synchronously when needed; the control device converts the pressure into groundwater level elevation according to a unified sampling cycle and binds the well number and time.

[0012] In a preferred embodiment, a flow meter and a pressure sensor are connected in series in the outlet pipeline of the dewatering well to synchronously collect water level, flow rate, and outlet pressure and synchronize the time. Under conditions of no pumping or extremely low frequency test pumping, several sets of water level elevations are collected and the effective lower water level elevation is obtained by median filtering. Then, the target water level and upper and lower limit thresholds are set according to the excavation control elevation and safety margin. A graded test pumping frequency sequence is issued to each well and the water level, flow rate, and pressure response are recorded in the stable window. If there is an anomaly, the frequency is reduced and the low-frequency pumping is extended. Then, the well-specific characteristic table is registered as the control input.

[0013] In a preferred embodiment, the control device synchronously collects groundwater level elevation, pumping flow rate and outlet pressure under a unified sampling period, performs median filtering on the groundwater level sequence to obtain the effective groundwater level elevation, and compares it with the initial groundwater level benchmark for consistency. At the same time, it performs continuous arrival, timestamp monotonicity and range reasonableness checks on the three types of data, and only allows the pump to start after the conditions are met.

[0014] In a preferred embodiment, the pumps are started in stages according to the deviation between the effective groundwater level of each well and the target groundwater level, and the pump start interval is set. The minimum operating frequency is issued for each well first and a stable window is maintained. After confirming that the flow rate is rising and there is no abnormal sudden rise in pressure, the frequency is increased to the commanded operating frequency. If the frequency increases but the flow rate does not increase and the pressure continues to rise, the ramp-up is frozen, the frequency is rolled back, and the low-frequency pumping is extended.

[0015] In a preferred embodiment, during stable operation, the upper and lower thresholds of the groundwater level constitute a control dead zone. If the water level is too high, the frequency of the operating wells is increased first, and then more wells are added. If the water level is too low, the frequency is decreased first, then the wells are stopped, and the pump is stopped and the valve is closed after the frequency is first reduced to the lowest buffer. At the same time, the pump inlet elevation and safety margin are used as hard constraints to trigger priority frequency limiting or pump stoppage. When allocation is required, the total pumping volume demand is given according to the control deviation, and the target pumping volume of each well is allocated according to the deviation ratio between wells. The command operating frequency is calculated in reverse by combining the well characteristic table, and checked with the upper limit of the outlet pressure and the abnormal pressure difference characteristics. If the allowable range is exceeded, the operating frequency of the well point is reduced, and the unfulfilled pumping demand is allocated to other well points.

[0016] In a preferred embodiment, the control device categorizes well points into a ventilated well point set and a non-ventilated well point set based on the installation records and compensation methods, and uses median filtering to obtain the effective groundwater level elevation for each set. Within adjacent sampling periods, the median of the group water level change for each set is calculated, and the difference between the median group water level changes is determined as the inter-group common mode offset. The deviation of the water level change of each well point within the non-ventilated well point set from the median group water level change of the non-ventilated well point set is calculated, and the common mode dispersion is determined based on the deviation to characterize the consistency of water level changes within the non-ventilated well point set. For the operating well point set, the median of the group pumping flow rate change is calculated within adjacent sampling periods. The median change in group pumping flow rate and group effluent pressure is used to determine the pressure-flow evidence matching coefficient based on the matching relationship between the median change in group pumping flow rate, the median change in group effluent pressure, and the inter-group common mode offset. The inter-group common mode offset, common mode dispersion, and pressure-flow evidence matching coefficient are normalized and then weighted according to preset weights to obtain the orifice consistency judgment value C(t). The orifice consistency judgment value C(t) is used to characterize the consistency between the water level orifice and the pressure-flow orifice. When the inter-group common mode offset increases, the common mode dispersion increases, or the pressure-flow evidence matching coefficient decreases, the orifice consistency judgment value C(t) decreases. When the orifice consistency judgment value C(t) is continuously lower than a preset threshold, orifice correction is triggered.

[0017] In a preferred embodiment, after entering the caliber correction state, the time misalignment of the atmospheric pressure link is used as the compensation time shift amount. Candidate shifts are searched in a rolling manner within a preset shift range, and the effective groundwater level elevation of the non-ventilated well point is recalculated. The optimal shift is selected using a rolling window objective function of weighted common mode dispersion based on inter-group variation differences. The compensation time shift amount is then progressively updated using single-cycle limiting and adaptive update coefficients, and the corrected water level caliber is output using the shifted atmospheric pressure.

[0018] In a preferred embodiment, verification wells are selected from the operating non-aerated wells and three micro-amplitude frequency perturbations are applied. Within each window, the median of pumping flow rate and outlet pressure is filtered and the local response slope to the frequency is calculated. The slope is compared with the calibration reference slope in the well characteristic table and the corrected residual common mode offset is superimposed to construct the verification residual. If the verification residual is continuously lower than the threshold, the water level input for staged pump start-up control is switched to the corrected effective lower water level elevation. If the verification residual exceeds the threshold, the water level caliber is frozen and the pump start-up order and frequency allocation rearrangement caused by the non-aerated set are restricted. At the same time, the triggering, shift update, perturbation response and freeze release processes are archived.

[0019] In a preferred embodiment, the control device obtains the effective groundwater level elevation by median filtering and locks the pump stop reference water level. Then, it selects the exit well points according to the batch mechanism and performs amplitude limiting and frequency reduction to the lowest operating frequency for each well, stops the pump, closes the valve for isolation, sets the pump stop interval between batches, and compensates the pumping capacity of the remaining well points after each batch exits.

[0020] During pump shutdown, the water level rebound rate is calculated and a rebound prediction is made. If the prediction approaches the upper limit, the interval is extended and the next batch of shutdowns is suspended. After all pumps are shut down, the rebound amount and rebound rate are continuously monitored. If the upper limit or the rebound rate is exceeded, the pumps are restarted according to the soft start rule to suppress the rebound. If it still cannot be controlled, an alarm is triggered and all data of the process are recorded.

[0021] The technical effects and advantages of the intelligent control system for deep well dewatering in water conservancy projects according to the present invention are as follows:

[0022] This invention unifies the water level elevation by using a wellhead reference elevation, obtains an effective groundwater level elevation through median filtering, and establishes an initial benchmark. Combined with the target water level and threshold, it enables tiered pump start-up, amplitude-limited frequency modulation, well addition and shutdown, and idling protection, making the control actions calculable and auditable. A diameter consistency judgment value is constructed using inter-group common mode offset and operating well pressure flow evidence, triggering compensation time shift correction. A calculable evidence chain is formed through perturbation verification, avoiding repeated rearrangement of multi-well control around diameter drift, improving stability and reducing water hammer and wear risks. During the pump shutdown phase, pumps are withdrawn in batches while continuously monitoring the rebound amount and rate, allowing for rapid handling and archiving of rebound anomalies. Attached Figure Description

[0023] Figure 1 This is a diagram showing the water pressure response of an intelligent control system for deep well dewatering in a water conservancy project, according to the present invention.

[0024] Figure 2 The present invention provides a thermal diagram of well point diameter offset for both ventilated and non-ventilated types in a deep well dewatering intelligent control system for water conservancy projects. Detailed Implementation

[0025] 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.

[0026] This invention discloses an intelligent control system for deep well dewatering in water conservancy projects, comprising: a wellhead benchmark modeling module, a graded pump start-up control module, a common mode correction and verification module, a rebound exit monitoring module, and signal connections between the modules.

[0027] First, in the wellhead benchmark modeling module, based on the excavation plan area, excavation control elevation, and construction zones, the well locations and depths of the dewatering wells are determined. At least one ring of observation wells is then installed outside the excavation to provide a comparison of the groundwater level outside the pit. Construction personnel use a level or total station to transfer the engineering benchmarks to each wellhead, establishing wellhead reference elevations, and recording each wellhead reference elevation as... ;

[0028] Subsequently, drilling was carried out according to the well location. After the borehole reached the designed depth, well casing and filter pipe were lowered, and filter material was backfilled according to the process to form a filter layer. The wellhead was then sealed and protected to prevent debris from falling in, ensuring that the internal structure of the well meets the installation conditions for long-term operation of the submersible pump and long-term immersion of the water level sensor. At the same time, to avoid damage to the wellhead and cables caused by mechanical operations such as on-site earthmoving, stone removal, silt cleaning, and drainage ditch excavation, a protective ring and conspicuous markings were set up in the wellhead area. Signal cables and power cables were equipped with wear-resistant sheaths and laid along a fixed path after passing through protective pipes. When crossing vehicle traffic areas, the cables were raised or protected trenches were buried to ensure uninterrupted data link.

[0029] Next, a groundwater level sensor is installed in each dewatering well and each observation well. The preferred groundwater level sensor is a hydrostatic level gauge, which is fixed above the filter pipe and below the lowest possible groundwater level to ensure that the sensor remains underwater and maintains its effective range even under maximum drawdown conditions. The sensor outputs the raw quantity as the water pressure within the well at a given moment. .

[0030] When using a ventilated hydrostatic level gauge, atmospheric pressure compensation is performed internally by the sensor; when using a non-ventilated hydrostatic level gauge, an atmospheric pressure sensor is simultaneously installed on-site to obtain atmospheric pressure data. The system outputs data with the same sampling period Δt to ensure a clear source of compensation. The control device converts the pressure signal into groundwater level elevation. And strictly based on the wellhead reference elevation As a conversion benchmark, its conversion formula is defined as follows:

[0031] ;

[0032] in, Indicates the groundwater level elevation expressed using a unified elevation benchmark; This indicates the reference elevation at the wellhead. This indicates the water pressure measured by the groundwater level sensor; Indicates atmospheric pressure or equivalent compensation at the same moment; ρ represents the density of the water; g represents the acceleration due to gravity. The control device controls each... Record the well number, sampling time, sensor number, and Δt simultaneously;

[0033] Meanwhile, a flow meter and an outlet pressure sensor are connected in series on the outlet pipe of each dewatering well. The flow meter outputs the pumping flow rate Q(t), and the outlet pressure sensor outputs the outlet pressure. The flow meter and outlet pressure sensor are fixed in the pipe section after the check valve and before the electric valve, and their installation position relationship is fixed in the control device. Then, the control device synchronously collects H(t) and Q(t) at Δt. And the sampling clock is uniformly calibrated;

[0034] After all sensors are online and signals are stable, an initial groundwater level benchmark is established, and the targets and thresholds required for subsequent closed-loop control are generated. The control device continuously collects no fewer than N sets of groundwater level elevation data under conditions where pumping is not initiated or only extremely low-frequency test pumping is performed. Median filtering was performed on this set of data to obtain the effective groundwater level elevation. Defined as:

[0035] ;

[0036] in, This serves as an effective groundwater level elevation for subsequent control and assessment. The control device will... As the initial groundwater level benchmark Write the data into the benchmark table and determine the target groundwater level based on the excavation control elevation and construction safety margin. At the same time, an upper limit threshold is formed. With lower threshold .in, , , All are defined using the same groundwater level elevation as H(t);

[0037] Further parameterized test pumping calibration was performed on each dewatering well: the control device sent a test pumping frequency sequence to the frequency converter of the submersible pump of that well, so that the submersible pump operated step by step within the safe range of low to medium frequency, and maintained at least one stable sampling window at each frequency level, and simultaneously recorded H(t) and Q(t) corresponding to that level within the window. If Q(t) and... Upon detecting a clear anomaly, the control device immediately reverted to the previous frequency setting and extended the low-frequency clean water pumping time until Q(t) and... After the situation stabilizes, the test pumping sequence can continue to avoid disturbing the filter layer, increasing sand carrying capacity in the well, or clogging the filter pipe due to excessive pumping. After the test pumping is completed, the control device writes the frequency, flow rate, pressure, and water level response of each well at each test pumping frequency range into the well-specific characteristic table, and together with the baseline table, it is solidified as the sole data source and control basis.

[0038] After completing the well layout, well completion, and measurement and control benchmark establishment in the wellhead benchmark modeling module, the deep well dewatering initiation and stable operation phase begins. The control objective is to controllably lower and stably maintain the groundwater level H(t) within the foundation pit at the target groundwater level throughout the entire construction process. Nearby, it ensures that the submersible pumps of each dewatering well do not run dry, overload, or experience abnormal hydraulic shocks, and that all pumping control actions are driven by clear data sources and clear triggering logic.

[0039] Specifically, the control device first performs an online validity check on the groundwater level sensors of all precipitation wells and observation wells: continuously acquiring groundwater level elevation H(t) under the same sampling period Δt, and reusing the median filtering of the wellhead benchmark modeling module to obtain the effective groundwater level elevation. ;

[0040] Then the control device will The initial groundwater level benchmark written by the wellhead benchmark modeling module A consistency comparison was performed to confirm that the sensor did not exhibit sudden jumps or prolonged freezing; simultaneously, the pumping flow rate Q(t) and outlet pressure on the outlet pipelines of each dewatering well were read. The current values ​​confirm that the flow meter and outlet pressure sensor are in working order. Only when the groundwater level H(t), pumping flow rate Q(t), and outlet pressure are measured... The control device will only allow the pumping process to begin when all three types of data meet the conditions of continuous arrival, monotonous timestamps, and reasonable range, in order to avoid blindly starting the pump under conditions of missing data, which could lead to uncontrollable depth drawdown or equipment damage.

[0041] When starting pumping in stages, the control device determines the pump start priority based on the groundwater level deviation. For the k-th dewatering well, the control device uses its effective groundwater level elevation. With the target groundwater level The difference is defined as the water level deviation:

[0042] ;

[0043] when This indicates that the current groundwater level in the well is higher than the target groundwater level, and it needs to be lowered by pumping water; when This indicates that the well has reached or fallen below the target groundwater level, and it is not advisable to prioritize increasing pumping during the startup phase.

[0044] Furthermore, the control device monitors all precipitation wells. The pumps are prioritized for wells with large deviations, and a fixed interval is set between adjacent pump starts. To avoid power surges, water hammer in pipelines, or excessively rapid soil submersible pump drawdown caused by simultaneous startup of multiple submersible pumps. Each well's pump start-up operation uses a variable frequency soft start method: the control device sends the minimum operating frequency to the well's frequency converter. And maintain a brief stable window, after confirming that Q(t) rises from zero and After no abnormal spikes are observed, the frequency is gradually increased to the command frequency calculated by the control algorithm using a linear ramping method. Its climbing relationship is defined as:

[0045] ;

[0046] in, Let be the pump start-up time, r be the frequency ramp-up slope, and f(t) be the real-time output frequency during the pump start-up process. During the frequency ramp-up process, the control device continuously monitors Q(t) and... When the flow rate fails to increase with frequency and the outlet pressure continues to rise, the control device determines that the resistance of the well's outlet channel has increased abnormally. It immediately freezes the frequency ramp-up and reverts to the previous stable frequency level. Subsequently, it extends the low-frequency clear water pumping window to restore channel patency until Q(t) and... After the stability condition is met again, the frequency can be increased further. The outlet pressure response relationship is as follows: Figure 1 As shown.

[0047] After completing the initial well point tiered activation, the control device enters the groundwater level closed-loop regulation process, and uses the upper limit threshold established by the wellhead benchmark modeling module. With lower threshold This creates a control dead zone, allowing pumping regulation to respond to groundwater level rises while avoiding equipment wear caused by frequent start-ups, shutdowns, or fine-tuning. The control device calculates the representative effective groundwater level elevation within the foundation pit at each sampling cycle Δt. And calculate the control deviation:

[0048] ;

[0049] when When the control device determines that the groundwater level is higher than the allowable upper limit, it is necessary to increase the pumping capacity. At this time, the control device first implements a frequency-priority control strategy: for the already operational dewatering wells, based on the control deviation... The frequency increment is allocated based on size, allowing wells with larger deviations to receive a higher pumping boost. To ensure that the control actions are calculable and auditable, the control unit defines the frequency adjustment amount for each well as:

[0050] ;

[0051] in, For proportional gain, The maximum frequency adjustment amplitude for a single cycle is defined by clip(·), which is the clipping operator; subsequently, the new command frequency for this well is updated as follows:

[0052] ;

[0053] in, This is the highest permissible operating frequency for this well. This represents the new command operating frequency obtained for the k-th well after frequency adjustment and amplitude limiting in the current sampling period. The control device will... The data is sent to the frequency converter, and Q(t) continues to be monitored. The response to frequency changes is designed to ensure that the increase in pumping capacity resulting from frequency increase actually occurs without introducing abnormal pressure differentials. If the requirements are still met after frequency increase... And all operating wells have reached If the well approaches the risk boundary of idling, the control device will then implement a well-compensation strategy: selecting the deviation from either never-operated or standby dewatering wells. For larger wells, additional pump start-up should be performed, and the same procedures apply as described above. The relationship between intervals and soft-start ramping ensures a smooth and controlled expansion process.

[0054] when When the control device determines that the groundwater level is below the allowable lower limit, indicating a risk of excessive precipitation, it is necessary to reduce the pumping capacity. At this point, the control device reverses the process and implements a frequency reduction priority control strategy, i.e., it operates at the same frequency. The computational framework provides a negative frequency adjustment and continues to verify Q(t) after frequency reduction. The decrease in response ensures a true reduction in pumping volume. When frequency reduction is still insufficient... Once back within the allowable range, the control device then executes the well stoppage compensation strategy: selecting the well point with the smallest deviation or already below the target, and first reducing its frequency to... Maintain a stable window to eliminate the risk of water hammer, then stop the pump and close the electric valve. After stopping the pump, continue to monitor the H(t) recovery process of the well to avoid local water level rebound causing sudden surge or concentrated seepage at the bottom of the pit.

[0055] To prevent submersible pumps from running dry, the control device stores the elevation corresponding to the submersible pump inlet for each well in the installation record of the wellhead reference modeling module. and set a safety margin At any given moment, the groundwater level in the well satisfies: When the control device determines that the water level in the well is close to the pump inlet, it immediately performs frequency limiting or pump shutdown on the well, and this action has a higher priority than the water level deviation. The need for regulation; only when the well water level rises and meets the requirements. And the control device will only allow the well to resume pumping after at least one stable window has been maintained.

[0056] Furthermore, to enable multi-well coordinated control to utilize the well-specific characteristic table established by the wellhead benchmark modeling module, the frequency calculation command allows for allocation using the target pumping rate as an intermediate value and inverse frequency calculation. The control device can first provide the required total pumping rate based on the control deviation e. And according to the deviation of each well The relative proportions of the target pumping volume per well are allocated as follows:

[0057] ;

[0058] ;

[0059] in, The water level deviation of the j-th precipitation well is defined as the difference between the effective groundwater level elevation of that well and the target groundwater level. Let be the allocation coefficient for the k-th well. The target pumping volume is then determined. The control unit subsequently retrieves the frequency-flow rate correspondence table stored in the wellhead baseline modeling module, and performs... Perform reverse lookup or interpolation to obtain the command frequency that satisfies the target pumping volume. And then by The response is subjected to security verification; when the interpolation is reversed... lead to When the allowable upper limit is exceeded or abnormal differential pressure characteristics occur, the control device will... If the operating frequency of the well exceeds the allowable range, reduce the operating frequency of the well and transfer the remaining pumping demand to other wells to ensure that hydraulic anomalies are not introduced while meeting the control objectives.

[0060] It should be noted that in the wellhead benchmark modeling module, some well points use ventilated static pressure level gauges with atmospheric pressure compensation completed internally by the sensor, while other well points use non-ventilated static pressure level gauges with atmospheric pressure provided by on-site atmospheric pressure sensors. And requires with The data is output according to the same sampling period Δt and used to calculate the groundwater level elevation H(t); in the graded pump start-up control module, the control device further uses H(t) and its median filtered value. This serves as the sole water level basis for pump start-up sequencing, threshold dead zone determination, and frequency enhancement or well addition compensation. This can lead to an unconventional compensation common-mode misalignment state: such as... Figure 2 As shown, without changing any hydraulic state of the well body, Due to communication delays, sampling alignment deviations, or timestamp binding errors, the H(t) of non-ventilated well points may experience a uniform shift in the same direction over a period of Δt, while the H(t) of ventilated well points remains unchanged. This shift may still meet the upper limits of continuous arrival, monotonic timestamps, and reasonable range at the single-well level, and may even be consistent with the initial groundwater level benchmark. The consistency comparison does not immediately trigger an anomaly, but it will manifest in multi-well dimensions as follows:

[0061] Water level deviation at some well points The systemic amplification or reduction causes a non-physical rearrangement of the pump start-up priority and frequency allocation within the same time window of the staged pump start-up control module. Furthermore, because the closed-loop control of the staged pump start-up control module immediately uses frequency enhancement priority or well-addition compensation actions to track this shift, a contradictory phenomenon occurs on-site:

[0062] The frequency of control commands and the number of operating wells are changing, but the pumping flow rate Q(t) and the outlet pressure are also changing. However, the response exhibits a weak or inconsistent response that does not match the actual water level disturbance, resulting in a structural decoupling between the water level-driven control action and the hydraulic output reflected by the pressure-flow-driven control. This state is not simply a sensor malfunction or a single-well anomaly, but rather a special system-level operating condition induced by the inconsistency of the compensation diameter and the reliance of multiple wells on a single water level diameter in the closed-loop control: if the water level deviation is continuously tracked in the closed-loop manner according to the staged pump start-up control module, the frequency adjustment and start-up / shutdown strategies will be repeatedly rearranged around the compensation offset, causing the equipment operation organization and well point load distribution to lose their stable basis;

[0063] Therefore, in this embodiment, after the graded pumping and groundwater level closed-loop control of the graded pump start-up control module enter stable operation, in the common-mode calibration and verification module, the control device identifies the well points according to the installation record of the wellhead benchmark modeling module. Well points using ventilated hydrostatic level gauges are grouped into the ventilated well point set, and well points using non-ventilated hydrostatic level gauges are grouped into the non-ventilated well point set. The groundwater level elevation conversion output for ventilated well point k is denoted as... The converted groundwater level elevation output for non-aerated well point k is denoted as... .

[0064] Meanwhile, to reduce transient disturbances, the control device uses the median filtering mechanism of the wellhead reference modeling module and the staged pump start-up control module to obtain an effective groundwater level elevation, denoted as follows: and The number of median filter window points is consistent with that of the wellhead benchmark modeling module and the staged pump start-up control module.

[0065] Next, to generate group-level evidence for compensating for common-mode misalignment, the control device calculates the median of the population water level changes for the two sets in adjacent sampling periods and constructs the inter-group common-mode offset. The median of the group water level change for a ventilated wellpoint set is defined as:

[0066] ;

[0067] The median of the group water level change for a non-aerated wellpoint ensemble is defined as:

[0068] ;

[0069] And define:

[0070] ;

[0071] Where median{⋅} is the median operator; This represents the median change in group water level of the aerated wellpoint set within adjacent sampling periods; This represents the median change in group water level of a set of non-aerated well points within adjacent sampling periods; This is used to characterize the common-mode misalignment feature where only the non-ventilated wellpoint ensemble experiences a uniform shift in the same direction, while the ventilated ensemble does not shift synchronously. To avoid misjudging occasional fluctuations of individual wellpoints as ensemble common-mode, the control device further calculates the common-mode dispersion of the non-ventilated wellpoint ensemble:

[0072] ;

[0073] in, This indicates the degree of dispersion of water level changes within a set of non-aerated well points relative to the change in the set's median. The smaller the value, the more consistent the changes within the set, and the more it conforms to the common mode property.

[0074] Simultaneously, the control device uses the pumping flow rate and outlet pressure, synchronously collected by the staged pump start-up control module, as a second evidence metric to construct the median of the group pressure-flow change of the set of operating wells within adjacent sampling periods. This median is used to identify decoupling characteristics where water level changes abruptly but pressure-flow mismatches occur. The control device reuses the determination results of the operating wells from the staged pump start-up control module, and records the set of operating wells as follows: And defined as follows:

[0075] ;

[0076] ;

[0077] in, Let be the pumping flow rate of the k-th precipitation well at time t. Let be the outflow pressure of the k-th precipitation well at time t, where Represents the set of operating well points The median change in group pumping flow rate within adjacent sampling periods; Represents the set of operating well points The median change in population effluent pressure over adjacent sampling periods.

[0078] Based on the aforementioned evidence of water level changes and pressure-flow changes, the control device defines a diameter consistency judgment value C(t). This value C(t) is a judgment quantity calculated using a normalized weighted average of inter-group common mode offset, common mode dispersion, and pressure-flow evidence matching coefficients. It is used to characterize the degree of consistency between the current water level diameter and the pressure-flow diameter. Preferably, C(t) is defined as follows:

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] Where C(t) represents the consistency judgment value; and Based on the basic scale parameter, and For dynamic scaling coefficients; Indicates within a scrolling window The steady-state fluctuation scale, Indicates within a scrolling window The steady-state fluctuation scales of both can be calculated using the median absolute deviation. and is the scale parameter for pressure-flow changes. By this definition, C(t) will only decrease significantly when there is significant intergroup splitting of the water level aperture and common mode consistency within the ensemble, while the pressure-flow aperture does not show a matching group response. This distinguishes the compensated common mode misalignment from the actual hydraulic changes in the calculated aperture.

[0084] Furthermore, when C(t) decreases and remains below the threshold At this time, the control device enters the caliber correction state and constructs the time misalignment of the atmospheric pressure link into an internal state quantity to compensate for the time shift. Perform self-tracking adjustment. To avoid shift jitter caused by alignment at a single moment, the control device maintains a scrolling window of length L and operates within a preset shift range. A scrolling search is performed within the range, where m represents the candidate shift; This represents the maximum absolute value of the candidate shift; for any candidate shift m, the control device constructs the atmospheric pressure after the shift:

[0085] ;

[0086] Use The groundwater level elevation conversion formula with the wellhead reference modeling module is used to recalculate the groundwater level elevation of non-aerated wellpoints, thus obtaining the recalculated effective groundwater level elevation. Thus, the candidate shifted and The objective function for constructing the scrolling window in the control device is:

[0087] ;

[0088] and take The minimum shift amount is ,in is the weighting coefficient used to balance the contribution of inter-group consistency of variation and common mode dispersion, and L is the rolling window length.

[0089] in, This represents the rolling window objective function corresponding to the candidate shift m at time t;

[0090] This yields the candidate compensation time shift:

[0091] ;

[0092] In order to make It features structural technical movements with dynamic adjustment, speed-limiting convergence, and reversibility; the control device does not directly command... = Instead, it adopts a limited, incremental update, defined as:

[0093] ;

[0094] in, Indicates the candidate compensation time shift amount; Indicates the amount of time shift compensation; This represents the maximum allowable shift adjustment range per single cycle. The coefficients are dynamically updated to adaptively adjust the convergence speed based on the current consistency judgment value C(t). The preferred definition is: ;in and These are the minimum and maximum update coefficients, respectively. Based on the above definitions, The discrete search results are transformed into continuously traceable internal state quantities, enabling the compensation link correction to have dynamic adjustment characteristics and avoid chattering.

[0095] In obtaining Subsequently, the control device updated the atmospheric pressure control calculation method to:

[0096] ;

[0097] Based on this, the self-corrected effective groundwater level elevation is output. Among them, the ventilation well point sampling Non-ventilated well point sampling .

[0098] in, This represents the atmospheric pressure after correction based on the time shift compensation. This represents the corrected effective groundwater level elevation of the k-th well; This represents the effective groundwater level elevation of the non-aerated wellpoint obtained by recalculating under the candidate shift m.

[0099] However, to ensure the structural effectiveness of this correction process and further enhance its ability to eliminate the decoupling phenomenon, a controlled perturbation verification action is introduced under the caliber correction state. This action utilizes the frequency conversion soft start and frequency limiting mechanism already present in the staged pump start-up control module to apply a small, short-term, and reversible frequency perturbation to the selected verification well point. The consistency between the pressure-flow response and the well-specific characteristic table is used to verify the current... Whether the water level has been restored to a usable state elevates the reliability of the correction from an empirical judgment to a calculable chain of evidence.

[0100] Specifically, the control device operates without triggering the idling protection constraint of the graded pump start-up control module and meets the following conditions. Under the premise of this, at least one operating well is selected from the set of non-ventilated well points as a verification well point, and its current command frequency is recorded as . .in, For the perturbation amplitude, the preferred value is to satisfy... , This refers to the maximum frequency adjustment range per cycle in the graded pump start-up control module. and These are the minimum and maximum permitted operating frequencies for the well, respectively.

[0101] The control device sequentially sets the well frequency to [a specific value] within three consecutive verification windows. , , The pumping flow rate is collected synchronously within each verification window according to the sampling period Δt. With water pressure The data in each window are filtered by median to obtain... , , as well as , , The control device calculates and verifies the frequency response slope of the well point near the operating point, defined as:

[0102] ;

[0103] ;

[0104] in, This represents the slope of the local response of the pumping flow rate to the frequency. The slope representing the local response of the outlet water pressure to frequency. , , These represent the pumping flow rates obtained by median filtering at the verification well points within three verification windows: low disturbance, baseline, and positive disturbance, respectively. , , These represent the water pressure obtained by median filtering at the verification well point within three verification windows: low disturbance, baseline, and positive disturbance.

[0105] Simultaneously, the control device calls the well-specific characteristic table generated by the wellhead benchmark modeling module to read and verify the frequency and flow rate discrete point pairs of the well points within the frequency neighborhood. And frequency, pressure discrete point pairs The calibration reference slope is calculated using the difference between adjacent discrete points, and is preferably defined as follows:

[0106] ;

[0107] ;

[0108] in, , This indicates the calibrated reference flow rate slope given in the well characteristic table. This represents the calibration reference pressure slope given in the well characteristic table. The control unit uses this slope to construct perturbations to verify the residuals.

[0109] ;

[0110] ;

[0111] ;

[0112] ;

[0113] in, , , These are the weighting coefficients; To prevent extremely small positive numbers with a denominator of zero; To adopt the corrected water level The recalculated inter-group common-mode offset is used to characterize the degree of residual common-mode after correction. By this definition, only when the local response of pressure flow to frequency is consistent with the calibration slope of the wellhead reference modeling module and the corrected inter-group common-mode offset is significantly reduced... This will reduce the error, thus providing calculable evidence that the compensation misalignment has been effectively corrected. This represents the i-th frequency sampling point in the well-specific characteristic table; Indicates and The corresponding pumping flow rate sampling value; Indicates and The corresponding water pressure sampling value.

[0114] in, according to Recalculated using the same set of criteria, the following results were obtained: ; ; ; The first median is used for the set of non-ventilated well points, and the second median is used for the set of ventilated well points.

[0115] Next, the control device incorporates the perturbation verification results into... In the dynamic adjustment, a structural innovation is achieved to realize closed-loop self-verification and dynamic convergence: when And continue for no less than During each sampling period, the control device determines that the current compensation time shift has reached a usable stable state, allowing the closed-loop control module for deviation calculation, pump start-up sequencing, and threshold judgment of the foundation pit's representative and effective groundwater level elevation to be switched to [the original state]. Calculated ;when At that time, the control device determined that the calibration was not yet reliable. To prevent the closed-loop of the tiered pump start-up control module from being rearranged due to misaligned signals, the control device performed a freeze treatment on the representative water level caliber. That is, only the effective groundwater elevation of the aerated well point set or the observation well point in the foundation pit was used as the closed-loop input of the tiered pump start-up control module, with priority given to the effective groundwater elevation of the aerated well point set. When the number of aerated well points was insufficient or the data did not meet the continuous arrival and reasonable range verification, the effective groundwater elevation of the observation well point in the foundation pit was used instead. Furthermore, within the freeze window, the rearrangement of pump start-up priority caused by changes in the water level of the non-aerated well point set was restricted.

[0116] in, This indicates the verification residual threshold;

[0117] Meanwhile, to accelerate correction convergence and avoid overshoot, the control device will dynamically update the coefficients. Further adaptive adjustment based on verification residuals, preferably... Set as ;

[0118] in, Represents a mapping function;

[0119] Finally, to ensure the traceability and auditability of this structural innovation process, the control device will specify the trigger time and the time within the scrolling window for each entry into the caliber correction state. and Statistics, compensated time shift The update trajectory and perturbation verification process , , , , , and verification residuals The results, along with the solidified archives representing the freezing and release times of the water level, form a complete data chain to demonstrate that the closed-loop control of the graded pump start-up control module is always driven by consistent and verified data.

[0120] After the graded pump start-up control module realizes closed-loop control of groundwater level and the common mode correction verification module ensures stable and usable pumping capacity of dewatering wells, the rebound exit monitoring module performs controlled exit of deep well dewatering when the construction phase requirements are met. That is, without causing a sudden rise in groundwater level, concentrated seepage or local surge risk, the pumping capacity is gradually reduced and the pump is stopped in a calculable order and rate. At the same time, after the pump is stopped, the rebound process of groundwater level is continuously monitored, and the pump is restarted, the pumping capacity is compensated or an alarm is triggered according to the clear trigger conditions.

[0121] Specifically, the control device first enters the pump shutdown preparation state and solidifies the pump shutdown criterion. The control device then acquires the representative and effective groundwater level elevation within the foundation pit during the same sampling period Δt. The calculation continues to use median filtering:

[0122] ;

[0123] During the pump shutdown preparation phase, a reference water level for pump shutdown is established. This is defined as the time to enter the pump shutdown preparation period. The effective groundwater level elevation, i.e. At the same time, the control device will target the groundwater level. Upper limit threshold Lower threshold As the boundary condition that cannot be crossed during pump shutdown remains unchanged, the upper limit of rebound monitoring after pump shutdown is defined as... Its diameter is completely consistent with H(t), ensuring that the entire process of pump shutdown and rebound monitoring is judged using the same groundwater level elevation system.

[0124] After establishing the aforementioned benchmarks, the control device generates a pump shutdown sequence and executes a controlled exit action involving frequency reduction, pump shutdown, and isolation. To prevent a rapid rise in water level caused by simultaneous unloading of multiple wells, a batch exit mechanism is adopted in the rebound exit monitoring module: the control device regulates the water level deviation of each well in the module by adjusting the pump start-up in stages. Based on the size and well type characteristics table, the pumping capacity contribution available, and the spatial zoning of the foundation pit layout, wells with minimal impact on maintaining water levels in key areas of the foundation pit are selected as the first batch of wells to be withdrawn. In each batch of withdrawals, the control device performs the following continuous actions on each well: first, the output frequency of the well's frequency converter is changed from the current command frequency... Reduce to the minimum operating frequency period by period according to the amplitude limit rules. The frequency decrease is still affected Constraints are implemented to ensure gradual changes in pipeline pressure; during the frequency decrease, the control device simultaneously monitors the well's pumping flow rate. With water pressure ,when It decreases as the frequency decreases and If no sudden increase occurs, the unloading process is deemed stable, and pump shutdown is permitted. If a sudden pressure increase or flow rate change occurs, the control device pauses further frequency reduction and maintains a stable window at the current frequency range until... , After the pump returns to a stable state, the load can be reduced further to avoid water hammer or check valve impact during pump shutdown.

[0125] When the well frequency has decreased After maintaining a stable window, the control device sends a pump stop command to the well and closes the well's outlet electric valve after confirming the pump stop, thus isolating the well from the main pipe and preventing backflow disturbance caused by pressure fluctuations in the main pipe after the pump stops.

[0126] Subsequently, the control device immediately invoked the closed-loop control algorithm of the staged pump start-up control module at the global level to compensate the pumping capacity of the still-operating wells: that is, without triggering the idling protection constraint, the command frequency for the remaining wells was appropriately increased. Alternatively, backup wells may be temporarily activated to ensure that the decrease in pumping capacity caused by the shutdown of some pumps does not cause the groundwater level to rise excessively. Pump stop intervals should be set between each batch. The control device only monitors the effective groundwater level elevation within the foundation pit after the previous batch of pumps has stopped. Only when no rapid rise trend is observed within the continuous monitoring window will the next batch of pump shutdowns be allowed, in order to avoid a sudden and uncontrolled rise in water level due to the cumulative effect of continuous pump shutdowns.

[0127] During pump shutdown, the control unit performs calculable predictions of the groundwater level rise trend and uses this data to dynamically adjust the pump shutdown schedule. The control unit operates within a fixed prediction window. Internal calculation of groundwater level change rate:

[0128] ;

[0129] The estimated reduction in pumping capacity due to the expected withdrawal of the next batch is converted into the expected recovery time. This forms a conservative upper bound for prediction:

[0130] ;

[0131] when Approaching or exceeding At that time, the control device automatically extends The next batch of pump shutdowns will be suspended, and the pumping capacity of the still-operating wells will be increased to the upper limit of the allowable limit through the control mechanism of the staged pump start-up control module to stabilize the water level before continuing to shut down; when far below Furthermore, when the recovery rate v(t) is at a low level, the control device allows the pump to continue exiting as planned, thereby transforming the pump shutdown process from a fixed schedule into an adaptive rhythm control driven by both measured and predicted groundwater levels.

[0132] Once all planned pump shutdown actions are completed, the control system enters the groundwater level rebound monitoring phase. During this phase, no further pumping commands are issued, but sampling continues at the predetermined intervals. Obtain the groundwater level H(t) of each observation well and key precipitation well, and calculate the rebound amount. :

[0133] ;

[0134] in, The reference water level for pump shutdown is used. During rebound monitoring, the control device simultaneously assesses the rebound rate, still using the rate of change v(t) as the rate indicator, and establishes a dual-trigger anomaly criterion of upper limit triggering and rate triggering: when the condition is met... Or meet If the sampling period is not less than M, the control device determines that the rebound is abnormal and immediately executes the abnormal handling procedure.

[0135] in, This represents the threshold value corresponding to the judgment quantity;

[0136] The anomaly handling procedure first restores pumping capacity based on the principle of minimal intervention: the control device prioritizes performance scoring in the common-mode calibration verification module. Lower, pumping performance ratio Higher well points are used as targets for restarting the pump, and the pump start-up interval is determined by the staged pump start-up control module. The pumps are started one by one in relation to the soft start ramp-up, and the control target is changed from... Switch to rebound suppression target Through closed-loop regulation Pull back to the allowable range; during pump restart, the control device continues to execute the idling protection constraint, that is, when any well meets the requirement... Do not increase the pump frequency or force pumping at this time to ensure that abnormal handling will not cause equipment risks. If restarting the pump still fails to complete the pumping within the preset time window... When the groundwater level returns to the allowable range, the control device triggers an alarm output and reports the current groundwater level elevation curve, rebound rate, and executed pump start-up well points and frequency commands, enabling on-site management personnel to take further engineering measures based on traceable data.

[0137] Finally, the rebound stability criterion is defined and a closing record is generated. This is done within a continuous stability testing window. Within, the absolute value of the rate of change of groundwater level v(t) satisfies And groundwater level elevation If the rebound process remains within the permissible range and no abnormal handling procedures are triggered, the control device determines that the rebound process is stable and enters the task completion state. After entering the task completion state, the control device will record the pump stop sequence and the pump stop period... With the v(t) curve, during the rebound monitoring period The curves and any abnormal handling records are archived and solidified to form a complete traceable data chain.

[0138] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0139] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0140] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0141] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0142] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0143] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart control system for deep well dewatering in water conservancy projects, characterized in that, include: Wellhead benchmark modeling module, staged pump start-up control module, common mode correction and verification module, rebound exit monitoring module, and signal connections between modules; The wellhead reference modeling module is used to set up dewatering wells and external observation wells according to the pit range and excavation control elevation, measure the wellhead reference elevation, and register the well type characteristic table and the initial groundwater level reference. The graded pump start-up control module is used to perform graded pump start-up after verifying the groundwater level elevation, pumping flow rate, and outlet pressure. It also combines the well type characteristic table to form a control dead zone based on the target water level and upper and lower limit thresholds, and performs frequency regulation, well addition, and well shutdown control. The common-mode correction and verification module is used to divide well points into a ventilated well point set and a non-ventilated well point set, and obtain the effective groundwater level elevation of each set. It calculates the inter-group common-mode offset and the common-mode dispersion of the non-ventilated well point set. Simultaneously, it obtains the median pumping flow rate change and the median outlet pressure change of the operating well point set within adjacent sampling periods, and calculates the pressure-flow evidence matching coefficient based on the correspondence between the median pumping flow rate change, the median outlet pressure change, and the inter-group common-mode offset. The inter-group common-mode offset, common-mode dispersion, and pressure-flow evidence matching coefficient are normalized and then weighted to generate a caliber consistency judgment value. This caliber consistency judgment value characterizes the consistency between the current water level caliber and the pressure-flow caliber. A smaller inter-group common-mode offset, a smaller common-mode dispersion, and a larger pressure-flow evidence matching coefficient result in a larger caliber consistency judgment value. When the caliber consistency judgment value is continuously lower than a preset threshold, it searches for compensation time shift and updates the value within a limited range. Select verification well points, apply frequency perturbation, and obtain verification residuals by combining the well type characteristic table. Switch or freeze water level inputs based on the verification residuals and save the data. The rebound exit monitoring module is used to lock the reference water level for pump shutdown, reduce the frequency of pump shutdown in batches, close valves for isolation and compensate for pumping capacity, and restart the pump or trigger an alarm if the rebound amount or rebound rate exceeds the limit.

2. The intelligent control system for deep well dewatering in water conservancy projects according to claim 1, characterized in that: Based on the excavation pit area and excavation control elevation, the layout of dewatering wells and external observation wells is determined, and the reference elevation of each wellhead is measured. Groundwater level sensors are installed in each well to collect water pressure and atmospheric pressure is collected synchronously when needed. The control device converts the pressure into groundwater level elevation according to a unified sampling cycle and binds the well number and time.

3. The intelligent control system for deep well dewatering in water conservancy projects according to claim 2, characterized in that: A flow meter and a pressure sensor are connected in series in the outlet pipeline of the dewatering well to synchronously collect water level, flow rate and outlet pressure and synchronize the time. Several sets of water level elevations are collected under conditions of no pumping or extremely low frequency test pumping, and the effective lower water level elevation is obtained by median filtering. The target water level and upper and lower limit thresholds are then set according to the excavation control elevation and safety margin. A graded test pumping frequency sequence is issued for each well, and the water level, flow rate and pressure response are recorded in the stable window. If there is an abnormality, the frequency is reduced and the low frequency pumping is extended. Then, the well-specific characteristic table is registered as the control input.

4. The intelligent control system for deep well dewatering in water conservancy projects according to claim 1, characterized in that: The control device synchronously collects groundwater level elevation, pumping flow rate and outlet pressure under a unified sampling period. It performs median filtering on the groundwater level sequence to obtain the effective groundwater level elevation and compares it with the initial groundwater level benchmark. At the same time, it performs continuous arrival, timestamp monotonicity and range reasonableness checks on the three types of data. Only after these conditions are met can the pump be started.

5. The intelligent control system for deep well dewatering in water conservancy projects according to claim 4, characterized in that: Pumps are started in stages according to the deviation between the effective groundwater level and the target groundwater level of each well, and the start-up interval is set. For each well, the minimum operating frequency is issued first and a stable window is maintained. After confirming that the flow rate is rising and there is no abnormal sudden rise in pressure, the frequency is increased to the commanded operating frequency. If the frequency increases but the flow rate does not increase and the pressure continues to rise, the ramp-up is frozen, the frequency is rolled back, and the low-frequency pumping is extended.

6. The intelligent control system for deep well dewatering in water conservancy projects according to claim 5, characterized in that: During stable operation, the upper and lower thresholds of the groundwater level form a control dead zone. If the water level is too high, the frequency of the operating wells is increased first, and then more wells are added. If the water level is too low, the frequency is decreased first, then the wells are stopped. The pump is stopped and the valve is closed after the frequency is reduced to the lowest buffer level. At the same time, the pump inlet elevation and safety margin are used as hard constraints to trigger priority frequency limiting or pump stoppage. When allocation is required, the total pumping volume demand is given according to the overall deviation, and the target pumping volume of each well is allocated according to the deviation ratio between wells. The command operating frequency is calculated in reverse by combining the well characteristic table, and checked with the upper limit of the outlet pressure and the abnormal pressure difference characteristics. If the allowable range is exceeded, the operating frequency of the well point is reduced, and the unfinished pumping demand is allocated to other well points.

7. The intelligent control system for deep well dewatering in water conservancy projects according to claim 1, characterized in that: The control device divides the well points into a ventilated well point set and a non-ventilated well point set according to the compensation method based on the installation records, and uses median filtering to obtain the effective groundwater level elevation of the two sets respectively. Within adjacent sampling periods, the median of the group water level change for each of the two sets is calculated, and the difference between the median group water level changes for the two sets is determined as the inter-group common mode offset. The deviation of the water level change of each well point within the non-aerated well point set relative to the median group water level change of the non-aerated well point set is calculated, and the common mode dispersion is determined based on the deviation to characterize the uniformity of water level changes within the non-aerated well point set. For the operating well point set, the median group pumping flow rate change and the median group outlet pressure change are calculated within adjacent sampling periods, and the common mode dispersion is determined based on the median group pumping flow rate change and the median group outlet pressure change. The matching relationship between the median change and the inter-group common mode offset is used to determine the pressure-flow evidence matching coefficient. After normalizing the inter-group common mode offset, common mode dispersion, and pressure-flow evidence matching coefficient, the values ​​are weighted according to preset weights to obtain the caliber consistency judgment value C(t). The caliber consistency judgment value C(t) is used to characterize the consistency between the water level caliber and the pressure-flow caliber. When the inter-group common mode offset increases, the common mode dispersion increases, or the pressure-flow evidence matching coefficient decreases, the caliber consistency judgment value C(t) decreases. When the caliber consistency judgment value C(t) is continuously lower than a preset threshold, caliber correction is triggered.

8. The intelligent control system for deep well dewatering in water conservancy projects according to claim 7, characterized in that: After entering the caliber correction state, the time misalignment of the atmospheric pressure link is used as the compensation time shift amount. Within the preset shift range, candidate shifts are searched in a rolling manner, and the effective groundwater level elevation of non-ventilated well points is recalculated. The optimal shift is selected using the rolling window objective function of the weighted common mode dispersion of the inter-group variation difference. Then, the compensation time shift amount is progressively updated using single-cycle amplitude limiting and adaptive update coefficients, and the corrected water level caliber is output using the atmospheric pressure after the shift.

9. The intelligent control system for deep well dewatering in water conservancy projects according to claim 8, characterized in that: Three micro-amplitude frequency perturbations are applied to verification wells selected from the operating non-aerated wells. Within each window, the median of pumping flow rate and outlet pressure is filtered and the local response slope to the frequency is calculated. The slope is compared with the calibration reference slope in the well characteristic table and the corrected residual common mode offset is superimposed to construct the verification residual. If the verification residual is continuously lower than the threshold, the water level input for staged pump start-up control is switched to the corrected effective lower water level elevation. If the verification residual exceeds the threshold, the water level caliber is frozen and the pump start-up order and frequency allocation rearrangement caused by the non-aerated set are restricted. At the same time, the triggering, shift update, perturbation response and freeze release processes are archived.

10. The intelligent control system for deep well dewatering in water conservancy projects according to claim 1, characterized in that: The control device obtains the effective groundwater level elevation by median filtering and locks the reference water level for pump shutdown. Then, it selects the well points to be withdrawn according to the batch mechanism and performs amplitude limiting and frequency reduction to the lowest operating frequency for each well, stops the pump, closes the valve for isolation, sets the pump shutdown interval between batches, and compensates the pumping capacity of the remaining well points after each batch is withdrawn. During pump shutdown, the water level rebound rate is calculated and a rebound prediction is made. If the prediction approaches the upper limit, the interval is extended and the next batch of shutdowns is suspended. After all pumps are shut down, the rebound amount and rebound rate are continuously monitored. If the upper limit or the rebound rate is exceeded, the pumps are restarted according to the soft start rule to suppress the rebound. If it still cannot be controlled, an alarm is triggered and all data of the process are recorded.