Water demand closed-loop collaborative control system based on infiltration model solution

Abstract

The application relates to the technical field of automatic control systems, and discloses a water demand closed-loop cooperative control system based on an infiltration model analysis, which comprises the following: a state variable high-frequency driving module outputting a total water demand state deviation scalar and a differential evolution slope; a characteristic parameter low-frequency correction module outputting a preferential flow and a piston flow distribution coefficient; and a scheduling instruction analysis module acquiring the total water demand state deviation scalar, wherein when the absolute value of the differential evolution slope is not greater than a response hysteresis dead zone threshold value, the scalar is decomposed by the preferential flow and the piston flow distribution coefficient to output a first physical driving signal; and when the absolute value of the differential evolution slope is greater than the response hysteresis dead zone threshold value, an interrupt signal is used to block a parameter latch module and a second physical driving signal is output by calling a forward weight matrix. The application reconstructs a control loop time domain correlation, eliminates data flow conversion conflicts, takes over control rights, and suppresses the action overshoot of an executing mechanism.

Description

Technical Field

[0001] This invention belongs to the field of automatic control system technology, and in particular relates to a water demand closed-loop collaborative control system based on infiltration model analysis. Background Technology

[0002] Currently, for closed-loop control links containing multi-source asynchronous input signals, conventional engineering control strategies generally adopt a hierarchical parameter mapping architecture. To maintain continuous scheduling of fluid resources, automated control loops need to synchronously integrate high-frequency sampled continuous environmental variables with low-frequency acquired discrete physical mechanism parameters. The current mainstream control logic adopts a zero-order hold mechanism, which mainly maintains the discrete characteristic parameters acquired by the low-frequency sampling nodes in a static latched state, using this as a benchmark weight to constrain the high-frequency environmental state variables. This is used to calculate and generate the physical drive signal sent to the end control node. However, when this static latching mechanism is applied to dynamic physical fields with transient disturbances, the inherent limitations of this control architecture will inevitably become apparent. The internal dynamic boundary of the controlled object will be instantly reshaped when encountering sudden environmental changes such as meteorological fluctuations. The physical mechanism parameters that determine the resource allocation ratio are constrained by their inherent mass balance analysis period and cannot achieve high-frequency synchronous updates. If the high-frequency control inner loop is forced to hibernate to wait for the low-frequency data analysis to complete, it will inevitably cause control response lag. If a linear interpolation algorithm is directly introduced, it will also be unable to eliminate the fusion deviation of asynchronous data in the time domain due to the lack of real physical evolution slope as support.

[0003] Besides the limitations of conduction hysteresis caused by the hardware configuration of fluid distribution, software control methods also have shortcomings in handling the coordinated regulation of multi-source information. For example, Chinese invention patent CN115480480B discloses a closed-loop tracking system and method for coordinated control of miss distance and trajectory prediction. This method uses the target's three-dimensional motion information to extrapolate the trajectory and project it onto a two-dimensional image plane to generate dynamic error feedback control quantities for composite closed-loop control. This type of control method relies on rigid body spatial coordinate calculation with deterministic kinematic laws. The basic constraint is that the controlled object follows a continuous and predictable spatial trajectory evolution. However, when facing the multiphase flow dynamics evolution under non-ideal soil infiltration conditions... The safety boundary of water transport and infiltration is determined by the mass balance mechanism of deep preferential flow leakage and matrix piston flow replenishment. Stable isotope tracing is needed to analyze the source of soil water and the relative interaction between preferential flow and piston flow during its replenishment of shallow groundwater. The fluid distribution boundary is determined by the nonlinear preferential flow and piston flow mass balance mechanism, rather than spatial geometric displacement. When the external environment undergoes transient changes, the existing technology lacks dynamic latching and blocking and adaptive feedforward substitution mechanisms for multi-rate asynchronous hydrodynamic characteristic parameters. Applying spatial trajectory projection algorithms leads to a fundamental mismatch between the prediction model and the nonlinear water transport boundary, and the erroneous dynamic extrapolation feedforward amplifies the control feedback oscillation amplitude.

[0004] Therefore, the technical problem to be solved by this invention is how to reconstruct the time-domain handshake protocol and trigger interrupt rules inside the control logic unit, and actively block the parameter misalignment transmission chain under transient and drastic conditions without shortening the sampling period of the underlying physical and chemical sensors, and eliminate the instruction feedforward deviation caused by asynchronous data fusion. Summary of the Invention

[0005] This invention aims to solve the problems of command feedforward distortion and system regulation oscillation caused by the reliance on static historical parameter latching in multi-rate asynchronous control loops when dealing with transient environmental changes.

[0006] In this technical solution, a water demand closed-loop coordinated control system based on infiltration model analysis includes: The state variable high-frequency driving module is equipped with a transient variability monitoring operator to receive continuous soil state signals and continuous precipitation signals at the first time step, runs the basic infiltration model to output the total water demand state deviation scalar, and uses the transient variability monitoring operator to extract the differential evolution slope of the continuous precipitation signal. The low-frequency correction module for characteristic parameters inputs isotope characteristic values ​​with a second time step larger than the first time step and obtains the priority flow and piston flow allocation coefficients. The parameter latching module maintains a constant output of the priority flow and piston flow allocation coefficients. The scheduling instruction parsing module, connected to the state variable high-frequency drive module and the parameter latching module, is used to obtain the total water demand state deviation scalar. When the absolute value of the differential evolution slope is not greater than the set response hysteresis dead zone threshold, it calculates the water demand compensation components of the priority flow and the piston flow in combination with the priority flow and piston flow allocation coefficients and converts them into the first physical drive signal output to the fluid allocation module. When the absolute value is greater than the set response hysteresis dead zone threshold, it generates an over-limit interruption signal and blocks the reading link to the parameter latching module. Simultaneously, the adaptive takeover operator retrieves the extreme operating condition feedforward weight matrix stored in the storage module and generates the second physical drive signal, which is then sent to the fluid allocation module.

[0007] Preferably, the low-frequency correction module for characteristic parameters further includes a smoothing latch submodule, which, when obtaining newly generated priority flow and piston flow allocation coefficients, retrieves the priority flow and piston flow allocation coefficients from the previous acquisition cycle and generates a dynamic decay forgetting factor by combining them with the environmental evolution slope. Based on the dynamic decay forgetting factor, an interpolation transition function is established and written into the parameter latch module, so that the scheduling instruction parsing module receives the priority flow and piston flow allocation coefficients that change according to the interpolation transition function during runtime, in order to convert the first physical drive signal.

[0008] Preferably, the scheduling instruction parsing module also includes a time-domain waveform peak-shifting scheduling module; after decomposing the total water demand state deviation scalar and generating the priority flow water demand compensation component and the piston flow water demand compensation component, based on the fluid dynamic response hysteresis characteristics of the fluid distribution module, specifically, the first physical transmission resistance of the priority flow water demand compensation component is determined by the transient hydraulic response time of the switching action of the switch control sub-module in the fluid distribution module, and the second physical transmission resistance of the piston flow water demand compensation component is determined by the mechanical hydraulic response inertia of the valve core displacement of the flow regulation sub-module and the pipeline hydraulic resistance; for the priority flow water demand compensation component, the single water supply duration T in the large pore priority flow trigger threshold and infiltration safety constraint parameters is used as the basis for determining the priority flow water demand compensation component. on With interval time T off A high-frequency short-pulse drive waveform is set to prevent surface water accumulation. Based on the hysteresis characteristics of water lateral diffusion caused by the suction of the soil microporous matrix, a low-frequency long-period drive waveform is set for normal matrix infiltration, targeting the piston flow water demand compensation component. When the priority flow water demand compensation component and the piston flow water demand compensation component are detected to exist synchronously, the high-frequency short-pulse drive waveform is controlled to be embedded in the interval of the low-frequency long-period drive waveform for output. The matrix capillary suction during the intermittent period is used to promote the lateral absorption of water and block the gravity infiltration channel of the priority flow in large pores.

[0009] Preferably, the high-frequency driving module for state variables also includes a soil sensor network interface and a field meteorological observation interface; the soil sensor network interface is connected to a multi-layer soil physical detection array deployed outside the system to continuously read continuous soil state signals at different depths of longitudinal profiles on site; the field meteorological observation interface is connected to an external tipping bucket rain gauge to acquire continuous precipitation signals in real time and transmit continuous independent variable data streams to the transient rate monitoring operator.

[0010] Preferably, the low-frequency correction module for characteristic parameters further includes a hydrogen and oxygen isotope input interface and a balance analysis calculation module; the hydrogen and oxygen isotope ratios of precipitation, soil water, and plant xylem measured by an external isotope mass spectrometer are received through the hydrogen and oxygen isotope input interface; the hydrogen and oxygen isotope ratios of precipitation, soil water, and plant xylem are input into the balance analysis calculation module, the dual-isotope linear mixing mass balance model is run, the preferential flow and piston flow distribution coefficients are calculated, and the results are transmitted to the parameter latching module.

[0011] Preferably, the instruction mapping rule matrix in the scheduling instruction parsing module includes an amplitude conversion logarithmic relationship and a time opening conversion rule; when the scheduling instruction parsing module runs the instruction mapping rule matrix, it converts the priority flow water demand compensation component into the first control pulse width for controlling the action of the field solenoid valve, and converts the piston flow water demand compensation component into the first valve opening duty cycle for controlling the displacement of the valve core of the field regulating valve; wherein, the combination of the first control pulse width and the first valve opening duty cycle constitutes the first physical drive signal.

[0012] Preferably, the scheduling instruction parsing module includes an adaptive takeover operator; after disconnecting the signal reading link to the parameter latching module, the adaptive takeover operator reads the extreme condition feedforward weight matrix stored in the storage module; the precipitation state is determined based on the polarity of the differential evolution slope, and the component coefficients in the extreme condition feedforward weight matrix are matched according to the magnitude of the absolute value of the differential evolution slope to determine the dynamic feedforward compensation weight; the dynamic feedforward compensation weight is used to replace the priority flow and piston flow allocation coefficients, and the total water demand state deviation scalar is linearly scaled by a product to obtain the second control pulse width and the second valve opening duty cycle used to combine and constitute the second physical drive signal.

[0013] Preferably, the scheduling instruction parsing module further includes an error feedback correction loop, which is used to cyclically read the actual response value of the soil hydrodynamic state of the controlled root zone and calculate the real-time residual between it and the set value within the current control cycle after either the first physical driving signal or the second physical driving signal is issued, and input the real-time residual into the state variable high-frequency driving module to complete the dynamic feedback bias correction of the total water demand state deviation scalar within the current control cycle.

[0014] Preferably, the system further includes a fluid network distribution execution module deployed at the controlled site. The fluid network distribution execution module includes a flow regulation sub-module, a switch control sub-module, and a pressure transmission sub-module. Both the flow regulation sub-module and the switch control sub-module are connected to the output interface of the scheduling command parsing module via a signal transmission bus. The flow regulation sub-module is used to adjust the opening resistance based on the received first valve opening duty cycle or second valve opening duty cycle. The switch control sub-module is used to change the switching cycle based on the received first control pulse width or second control pulse width.

[0015] Compared with existing technologies, the water demand closed-loop coordinated control system based on infiltration model analysis of the present invention has the following advantages: 1. In the closed-loop collaborative control of water demand based on infiltration model analysis, the high-frequency drive loop of state variables collects soil moisture content, matrix potential, and rain gauge pulse signals at multiple depths according to the first time step, and calculates the total water demand state deviation scalar in real time; the low-frequency correction loop of characteristic parameters acquires the hydrogen-oxygen stable isotope ratio of input water, soil water at multiple depths, and shallow groundwater at the second time step, analyzes d-excess, LMWL, and lc-excess, and runs the end-member hybrid model to obtain infiltration safety constraint parameters, and maintains discrete parameter output with the help of parameter latching module; this two-layer cascade topology reconstructs the time domain characteristics of the control loop, blocks the time domain conflict between the continuous fluctuation of high-frequency environmental state over time and the intermittent acquisition of low-frequency mechanism characteristics, and enables the fluid distribution adjustment node to decompose the total water demand state deviation scalar according to the currently latched reasonable distribution coefficient, eliminating the cumulative deviation caused by the lack of discrete parameter updates in the control link.

[0016] 2. The high-frequency drive loop of the state variable is equipped with a transient rate monitoring operator to calculate the differential evolution slope of the precipitation pulse accumulation signal in real time within a preset sliding time window, and compare the absolute value of the slope with the preset physical response hysteresis dead zone threshold. When the system encounters sudden and severe environmental fluctuations and the absolute value of the differential evolution slope suddenly exceeds the limit, the regulation and control system generates an over-limit interruption signal, instantly and forcibly isolating the reading link of the scheduling instruction parsing module to the parameter latching module, avoiding the continued constraint of high-frequency variable input by expired discrete historical weights, solving the step jump of output instructions caused by data misalignment in the multi-rate asynchronous control loop, and suppressing the overshoot of the end fluid distribution regulation component.

[0017] 3. The scheduling instruction parsing module, in conjunction with the over-limit interrupt signal, simultaneously activates the adaptive takeover operator while blocking the parameter latching module's extraction channel. This operator retrieves the pre-stored extreme condition feedforward weight matrix from the non-volatile memory. The operator adjusts the polarity and amplitude of the differential evolution slope to quickly calculate the high-frequency dynamic compensation weight. It then takes over the allocation ratio that has failed due to asynchronous data misalignment, rapidly reducing the upper limit of water supply intensity or urgently cutting off water supply. This prevents the superposition of sudden rainstorms and artificial water supply from causing surface water accumulation and exacerbating deep preferential flow leakage. This intertwined feedback control and feedforward compensation ensures that the control system still has the ability to accurately respond to hydrodynamic boundary reshaping when the external physical transport channel mechanism undergoes drastic deformation, maintaining the stability of water resource coordinated allocation instruction output under all operating conditions. Attached Figure Description

[0018] Figure 1 This is a flowchart of the multi-rate asynchronous data decoupling and adjustment process of the present invention; Figure 2 This is the topology diagram of the time-domain staggered collaborative sensing network of the present invention; Figure 3 This is a flowchart illustrating the switching between normal and extreme operating conditions in this invention; Figure 4 This is a flowchart of the real-time feedback correction process of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A water demand closed-loop coordinated control system based on infiltration model analysis includes: The state variable high-frequency driving module is equipped with a transient variability monitoring operator to receive continuous soil state signals and continuous precipitation signals at the first time step, runs the basic infiltration model to output the total water demand state deviation scalar, and uses the transient variability monitoring operator to extract the differential evolution slope of the continuous precipitation signal. The low-frequency correction module for characteristic parameters inputs isotope characteristic values ​​with a second time step larger than the first time step and obtains the priority flow and piston flow allocation coefficients. The parameter latching module maintains a constant output of the priority flow and piston flow allocation coefficients. The scheduling instruction parsing module, connected to the state variable high-frequency drive module and the parameter latching module, is used to obtain the total water demand state deviation scalar. When the absolute value of the differential evolution slope is not greater than the set response hysteresis dead zone threshold, it calculates the water demand compensation components of the priority flow and the piston flow in combination with the priority flow and piston flow allocation coefficients and converts them into the first physical drive signal output to the fluid allocation module. When the absolute value is greater than the set response hysteresis dead zone threshold, it generates an over-limit interruption signal and blocks the reading link to the parameter latching module. Simultaneously, the adaptive takeover operator retrieves the extreme operating condition feedforward weight matrix stored in the storage module and generates the second physical drive signal, which is then sent to the fluid allocation module.

[0021] Preferably, the low-frequency correction module for characteristic parameters further includes a smoothing latch submodule, which, when obtaining newly generated priority flow and piston flow allocation coefficients, retrieves the priority flow and piston flow allocation coefficients from the previous acquisition cycle and generates a dynamic decay forgetting factor by combining them with the environmental evolution slope. Based on the dynamic decay forgetting factor, an interpolation transition function is established and written into the parameter latch module, so that the scheduling instruction parsing module receives the priority flow and piston flow allocation coefficients that change according to the interpolation transition function during runtime, in order to convert the first physical drive signal.

[0022] Preferably, the scheduling instruction parsing module also includes a time-domain waveform peak-shifting scheduling module; after decomposing the total water demand state deviation scalar and generating the priority flow water demand compensation component and the piston flow water demand compensation component, based on the fluid dynamic response hysteresis characteristics of the fluid distribution module, specifically, the first physical transmission resistance of the priority flow water demand compensation component is determined by the transient hydraulic response time of the switching action of the switch control sub-module in the fluid distribution module, and the second physical transmission resistance of the piston flow water demand compensation component is determined by the mechanical hydraulic response inertia of the valve core displacement of the flow regulation sub-module and the pipeline hydraulic resistance; for the priority flow water demand compensation component, the single water supply duration T in the large pore priority flow trigger threshold and infiltration safety constraint parameters is used as the basis for determining the priority flow water demand compensation component. on With interval time T off A high-frequency short-pulse drive waveform is set to prevent surface water accumulation. Based on the hysteresis characteristics of water lateral diffusion caused by the suction of the soil microporous matrix, a low-frequency long-period drive waveform is set for normal matrix infiltration, targeting the piston flow water demand compensation component. When the priority flow water demand compensation component and the piston flow water demand compensation component are detected to exist synchronously, the high-frequency short-pulse drive waveform is controlled to be embedded in the interval of the low-frequency long-period drive waveform for output. The matrix capillary suction during the intermittent period is used to promote the lateral absorption of water and block the gravity infiltration channel of the priority flow in large pores.

[0023] Preferably, the high-frequency driving module for state variables also includes a soil sensor network interface and a field meteorological observation interface; the soil sensor network interface is connected to a multi-layer soil physical detection array deployed outside the system to continuously read continuous soil state signals at different depths of longitudinal profiles on site; the field meteorological observation interface is connected to an external tipping bucket rain gauge to acquire continuous precipitation signals in real time and transmit continuous independent variable data streams to the transient rate monitoring operator.

[0024] Preferably, the low-frequency correction module for characteristic parameters further includes a hydrogen and oxygen isotope input interface and a balance analysis calculation module; the hydrogen and oxygen isotope ratios of precipitation, soil water, and plant xylem measured by an external isotope mass spectrometer are received through the hydrogen and oxygen isotope input interface; the hydrogen and oxygen isotope ratios of precipitation, soil water, and plant xylem are input into the balance analysis calculation module, the dual-isotope linear mixing mass balance model is run, the preferential flow and piston flow distribution coefficients are calculated, and the results are transmitted to the parameter latching module.

[0025] Preferably, the instruction mapping rule matrix in the scheduling instruction parsing module includes an amplitude conversion logarithmic relationship and a time opening conversion rule; when the scheduling instruction parsing module runs the instruction mapping rule matrix, it converts the priority flow water demand compensation component into the first control pulse width for controlling the action of the field solenoid valve, and converts the piston flow water demand compensation component into the first valve opening duty cycle for controlling the displacement of the valve core of the field regulating valve; wherein, the combination of the first control pulse width and the first valve opening duty cycle constitutes the first physical drive signal.

[0026] Preferably, the scheduling instruction parsing module includes an adaptive takeover operator; after disconnecting the signal reading link to the parameter latching module, the adaptive takeover operator reads the extreme condition feedforward weight matrix stored in the storage module; the precipitation state is determined based on the polarity of the differential evolution slope, and the component coefficients in the extreme condition feedforward weight matrix are matched according to the magnitude of the absolute value of the differential evolution slope to determine the dynamic feedforward compensation weight; the dynamic feedforward compensation weight is used to replace the priority flow and piston flow allocation coefficients, and the total water demand state deviation scalar is linearly scaled by a product to obtain the second control pulse width and the second valve opening duty cycle used to combine and constitute the second physical drive signal.

[0027] Preferably, the scheduling instruction parsing module further includes an error feedback correction loop, which is used to cyclically read the actual response value of the soil hydrodynamic state of the controlled root zone and calculate the real-time residual between it and the set value within the current control cycle after either the first physical driving signal or the second physical driving signal is issued, and input the real-time residual into the state variable high-frequency driving module to complete the dynamic feedback bias correction of the total water demand state deviation scalar within the current control cycle.

[0028] Preferably, the system further includes a fluid network distribution execution module deployed at the controlled site. The fluid network distribution execution module includes a flow regulation sub-module, a switch control sub-module, and a pressure transmission sub-module. Both the flow regulation sub-module and the switch control sub-module are connected to the output interface of the scheduling command parsing module via a signal transmission bus. The flow regulation sub-module is used to adjust the opening resistance based on the received first valve opening duty cycle or second valve opening duty cycle. The switch control sub-module is used to change the switching cycle based on the received first control pulse width or second control pulse width.

[0029] Example 1: In a specific irrigation district water resource scheduling scenario, the water demand closed-loop collaborative control system is deployed at the distribution execution node of the fluid transmission pipeline network. In this scenario, the environmental sensing unit periodically monitors the volumetric water content of soil at multiple depths, matrix potential, and the pulse count sequence of the tipping bucket rain gauge with a sliding calculation step of 5 minutes. However, the characteristic parameters required for isotope mass balance analysis are limited by the timeliness of field sampling and laboratory mass spectrometry analysis, and are input to the system with a low-frequency cycle of weekly or precipitation events. The existing zero-order hold control logic lacks physical mechanism support. When encountering sudden rainstorms that change the infiltration and transport mechanism, the system continuously calls the lagging static weight coefficient, causing a mismatch between resource scheduling and water demand, resulting in a drive lag in the fluid distribution and regulation mechanism.

[0030] The water demand closed-loop collaborative control system adopts a two-layer cascade regulation architecture for logical reconstruction. The high-frequency drive module for state variables serves as the inner loop, receiving environmental variables in real time and calculating the scalar of water demand state deviation. Simultaneously, a built-in transient rate monitoring operator calculates the differential evolution slope of rainfall intensity within a preset 15-minute sliding time window. Specifically, during operation, the basic soil moisture balance model acquires multi-depth soil volumetric water content within the current calculation step. Based on the root growth depth of the controlled crop at the current growth stage, it dynamically determines the active root layer and its depth. The weighted average water content of the active root layer is calculated using the proportion of soil layer thickness represented by each monitored depth within the active root layer to the total active root layer depth as the weight allocation basis. The amount of water required for the root zone is calculated by subtracting the pre-set field water holding capacity (based on the capillary maximum water holding capacity determined by controlled field soil geology) from the pre-set field water holding capacity to obtain the soil water deficit in the root zone. This deficit is then combined with the active root layer depth to obtain the theoretical water requirement of the crop. Finally, the effective precipitation (based on the actual amount of water infiltrated into the active root layer within the sliding time window, calculated using the tipping bucket rain gauge pulse cumulative precipitation) is subtracted from the theoretical water requirement to obtain the effective precipitation (based on the pre-set field water requirement). This allows for the calculation of the total water demand state deviation scalar in the time domain. The characteristic parameter low-frequency correction module serves as the outer loop, receiving the δ¹ values ​​of input water (irrigation water and precipitation), soil water at multiple depths of 10cm, 30cm, 60cm, and 100cm, as well as shallow groundwater or seepage water from the lower boundary, through the hydrogen and oxygen isotope input interface. 8 The ratio of stable isotopes O to δ²H; the equilibrium analytical calculation module first constructs the local atmospheric precipitation line LMWL (δ²H=a·δ¹). 8 O+b), where a is the slope and b is the intercept, which is set based on the linear regression fitting equation of the measured isotope ratios of precipitation samples collected in a controlled field, and the deuterium surplus d-excess=δ²H-8·δ¹ of each water body is calculated. 8 O and the evaporation deviation index lc-excess=δ²H-a·δ¹ 8Ob is used to identify the degree of evaporative fractionation and recharge source of soil water. The criteria are as follows: the negative deviation of lc-excess less than zero characterizes the intensity of the influence of evaporative fractionation on soil water; the distribution of soil water isotope ratios at each depth along the local precipitation line and their Euclidean distance from the input water isotope ratio determine the water recharge source. Subsequently, a dual-isotope multi-endmember linear mixing mass balance model is run. Using input water and shallow soil water as mixing endmembers, the mass balance equations are solved to calculate the contribution rate of precipitation to the recharge of shallow groundwater or lower boundary seepage water, thus determining the contribution ratio f of preferential flow in large pores during the rapid recharge of shallow groundwater by precipitation. pf The displacement transport characteristics of matrix piston flow were determined by the hysteresis depth of the peak value of deep soil water isotope ratio relative to the surface layer; when the analysis showed that the preferential flow contribution ratio f pf When the preset safety threshold is exceeded (based on the maximum allowable priority flow recharge ratio for controlled soils that do not generate deep gravity bypass leakage), it indicates that bypass leakage is likely to occur in the large pores of the soil. Based on this, the equilibrium analysis calculation module constructs a system including the maximum allowable water supply intensity upper limit I. max Maximum single irrigation volume V max Duration T of a single water supply on and the interval time T off The infiltration safety constraint parameters, specifically the mapping rule, are as follows: in f pf When the preset safety threshold is exceeded, the maximum allowable water supply intensity limit I is reduced proportionally. max With the upper limit of single irrigation volume V max Shorten the duration of a single water supply (T) on And extend the interval time T off To suppress preferential flow through large pores; in f pf When the water supply intensity and pulse duration parameters set by the basic hydraulic conditions are not greater than the preset safety threshold, the upper limit of the water supply intensity and pulse duration parameters are maintained; and the parsed infiltration safety constraint parameters are written into the parameter latch module to maintain the output, so as to match and correct the high-frequency drive inner loop data of the 15-minute sliding time window across the step size within the low-frequency update cycle.

[0031] Combination Figure 3As shown, the system uses the relationship between the absolute value of the differential evolution slope and the set response hysteresis dead zone threshold as the basis for determining the operating condition switch: when the absolute value of the differential evolution slope is not greater than the set response hysteresis dead zone threshold, the system is in the response hysteresis dead zone, and executes the reading of infiltration safety constraint parameters and converts them into the first physical drive signal; when the transient rate monitoring operator detects a sudden change in the absolute value of the differential evolution slope, and the absolute value exceeds the set response hysteresis dead zone threshold, an over-limit interruption signal is generated. Based on the interruption signal, the scheduling instruction parsing module automatically cuts off the reading link to the parameter latching module and directly runs the adaptive takeover operator. By extracting the extreme operating condition feedforward weight matrix stored in the non-volatile memory, a second physical drive signal is generated and sent to the end execution node. Specifically, the extreme operating condition feedforward weight matrix pre-stored in the storage module is designed as a 2-row, 2-column constant lookup table. Its horizontal row dimension represents the discrete classification interval of the absolute value of the differential evolution slope of precipitation, which is divided into a moderate rain interval of 5 mm to 15 mm per hour and a heavy rain interval of more than 15 mm per hour. The vertical column dimension represents the polarity of the differential evolution slope, corresponding to negative polarity representing weakening precipitation and positive polarity representing strengthening precipitation, respectively. The initial component coefficients in this matrix are preset to 0.75 and 0.25 for moderate rain intervals under negative and positive polarities, respectively, and 0.85 and 0.15 for heavy rain intervals under negative and positive polarities, respectively. After the adaptive takeover operator is activated, it directly classifies the precipitation intensity based on the polarity and amplitude of the currently extracted differential evolution slope, according to the absolute value of the differential evolution slope. The system matches horizontal rows of intervals and vertical columns of precipitation increase / decrease trends represented by positive and negative signs. It selects smaller component coefficients under the positive polarity of increased precipitation to forcibly lower the upper limit of water supply and larger component coefficients under the negative polarity of decreased precipitation to mitigate excessive shutdown. The system retrieves and matches a unique dynamic feedforward compensation weight in the lookup table. It then uses this weight to quickly force a reduction in the upper limit of water supply intensity or issue a valve shut-off command to generate a second physical drive signal. This prevents surface water stagnation and strong preferential leakage induced by heavy rain and artificial water supply.

[0032] This collaborative logic enables rapid switching of physical resource scheduling weights under extreme precipitation conditions. Within the steady-state range, the system maintains allocation accuracy using low-frequency characteristic parameters. Furthermore, during operation within the aforementioned steady-state range, the environmental evolution slope is the rate of change of trend obtained by calculating the difference between the soil moisture measurement values ​​of the current hour and the previous hour. The dynamic decay forgetting factor is defined as a proportional value that decreases linearly from 1.0 to 0.1 as the absolute value of the environmental evolution slope increases. Its specific mapping relationship is as follows: (when hour, In the formula For dynamic decay forgetting factor, The absolute value of the slope of environmental evolution. The preset upper limit threshold of the environmental evolution slope is used to quantitatively indicate the confidence decay rate of historical characteristic parameters. The interpolation transition function is a time-domain linear gradient function with a total duration of 600 seconds between two adjacent low-frequency update cycles. It uses a dynamic decay forgetting factor as the step weight of the adjustment amplitude per second to control the infiltration safety constraint parameter to smoothly transition from the historical latched value of the previous cycle to the newly generated value of the current cycle, thereby eliminating the step jump of the control weight. At the moment of environmental drastic change, the system blocks the impact of expired parameters on the high-frequency control loop through parameter latching based on the physical state variability. The operation results show that when the fluid distribution execution node encounters precipitation fluctuations, the step amplitude of the execution end drive signal converges to the preset range, avoiding mechanical fatigue wear of the regulating valve and resource allocation overshoot, and realizing the synchronous operation of high-frequency state perception and low-frequency mechanism benchmark.

[0033] Example 2: In the water distribution and regulation scenario of this irrigation district, to verify the system's parameter self-adaptation capability and regulation stability under high-frequency environmental fluctuations, a test section with real-time acquisition capability of water flow characteristic variables was selected as the test object. The test bench includes a variable frequency fluid drive pump group with closed-loop control function, whose outlet is connected to a distributed pipe network system with a working pressure upper limit of 0.5MPa. The monitoring end consists of a high-precision flow meter group and a pressure sensor array, and the sensor sampling frequency is set to 100Hz to ensure high-frequency acquisition of water pressure fluctuations in the pipe network. To simulate real industrial interference, a 50Hz harmonic disturbance and random flow pulse are superimposed on the controlled water flow through an external disturbance injection module to observe the system's response trajectory to transient deviations. The test parameters are set according to engineering logic, with the sampling period set according to the Nyquist sampling theorem to ensure non-aliasing acquisition of transient hydraulic characteristic parameters in the pipe network; the proportional gain coefficient is selected based on the system's inertial time constant. To ensure that the actuator can quickly enter the convergence state during sudden load changes, during the offline calibration phase before the system is put into operation, the raw data stream of the sensor under steady-state precipitation conditions is continuously collected by the transient rate monitoring operator. The maximum fluctuation amplitude of the differential evolution slope reference noise caused by external wind speed interference and mechanical vibration of the tipping bucket rain gauge is measured to be 0.3 mm per minute. In order to balance the sensor's anti-noise and anti-malfunction requirements with the safety response sensitivity under sudden rainstorm conditions, this invention sets the response hysteresis dead zone threshold precisely at 0.5 mm per minute. This value corresponds to 1.6 times the maximum fluctuation amplitude of the reference noise. Engineering experiments show that if the threshold is set below 0.3 mm per minute, the sensor reference noise will frequently trigger the interrupt signal erroneously, leading to control link deadlock; if the threshold is set above 1.5 mm per minute, the system will not be able to cut off the expired parameters in time at the beginning of a strong convective rainstorm, thus losing the practical engineering significance of feedforward control protection.

[0034] During the experiment, the control system collected the instantaneous pressure and flow values ​​at the pump outlet through the high-frequency drive module and used them as the input benchmark for the water demand deviation scalar. In the control group experiment, the system adopted the parameter latching mode and set the allocation coefficient to a constant 0.5. In the experimental group experiment, the system activated the real-time adaptive parameter update mechanism. When the introduced flow disturbance caused the pipeline pressure fluctuation to exceed the preset pressure deviation threshold of 15%, the transient rate monitoring operator captured the pressure change signal. At this time, the control logic triggered the reconstruction action of the feedforward weight matrix and adjusted the pump drive frequency to the target correction point. In actual operation, in order to overcome the inherent mechanical hydraulic response hysteresis and inertial dead zone of the overall pipeline network and the actuator valve, a set of high-frequency response energy storage hydraulic buffer chamber and high-speed switching pilot valve group were configured at the output end of the fluid distribution module. When the scheduling command parsing module detected that the high-frequency short pulse drive waveform and the low-frequency long period drive waveform existed synchronously, it was not a direct Instead of driving the large-diameter regulating valve in the main pipeline to perform high-frequency reciprocating pumping, the system maintains the large-diameter regulating valve in a stable opening according to a long-period pattern to maintain the upper limit of the reference flow rate. The adaptive control output drives the high-speed switching pilot valve group to perform high-frequency pulse pressure relief within the valve opening gap of the low-frequency long-period waveform. The elastic deformation of the energy storage hydraulic buffer chamber absorbs and smooths the instantaneous water hammer impact fluctuations, so that the overall large inertia hydraulic system can perform millisecond-level high-frequency short pulse time-domain staggered nested scheduling without physical loss. The recorded data shows that in the control group, because the allocation coefficient cannot be updated in real time with transient disturbances, the actuator exhibits obvious lag when facing pressure pulses, and the regulating pressure value has an overshoot of more than 20% within 0.8s. In the test group, the system completes the feedforward parameter correction within 0.12s after detecting the pressure deviation, the overshoot of the actuator is limited to 5.2%, and the time for the pressure to recover to steady state is shortened to 0.35s.

[0035] Further testing showed that if the adjustment step size of the distribution coefficient was increased to 1.5 times the set value, the pipeline pressure exhibited high-frequency fluctuations; if the step size was decreased to 0.5 times the set value, the system's response time delay to flow pulses increased to 0.9s, losing its effective ability to suppress transient disturbances. Based on the above measurement data, the parameter latch update coefficient between 0.5 and 1.2 was determined as the system's working window. As the load on the controlled fluid system increased, the control command remained synchronized with the physical fluid transport state. The experimental results verified the dynamic stability of the constructed water demand closed-loop collaborative control system when dealing with high-frequency environmental disturbances, and quantitative comparison proved that by introducing real-time rate monitoring and dynamic collaboration with the feedforward weight matrix, the overshoot fluctuations of the actuator's actions could be reduced.

[0036] Example 3: This example combines Figures 1 to 2 The following describes the water demand closed-loop coordinated control system based on infiltration model analysis, such as... Figure 1As shown, the output of the first time step receiver of the multi-depth soil hydrodynamic and precipitation pulse accumulation signals is connected to the state variable high-frequency drive module to run the basic soil moisture balance model. The output of the state variable high-frequency drive module running the basic soil moisture balance model is connected to the transient variability monitoring operator to extract the differential evolution slope. The output of the transient variability monitoring operator to extract the differential evolution slope is connected to the scheduling instruction parsing module to calculate compensation using coefficients. Simultaneously, the left output of the state variable high-frequency drive module running the basic soil moisture balance model is unidirectionally connected downwards to the left input of the scheduling instruction parsing module to calculate compensation using coefficients, used to transmit the total water demand status to the scheduling instruction parsing module. The output of the stable isotope eigenvalue second time step input is connected to the characteristic parameter low-frequency correction module to maintain the infiltration safety constraint parameter constant. The output of the characteristic parameter low-frequency correction module to maintain the infiltration safety constraint parameter constant is connected to the scheduling instruction parsing module to calculate the water demand control component in combination with the constraint parameter. The output of the scheduling instruction parsing module to calculate compensation in combination with the coefficient is connected to the fluid distribution module to receive the physical drive signal. The output of the fluid distribution module to receive the physical drive signal is connected to the adaptive pipe operator to retrieve the feedforward weight matrix. The output of the adaptive pipe operator to retrieve the feedforward weight matrix is ​​connected in reverse to the fluid distribution module to receive the physical drive signal.

[0037] like Figure 2 As shown, the state variable high-frequency driving module, storage module, scheduling instruction parsing module, fluid distribution module, and characteristic parameter low-frequency correction module are arranged sequentially from top to bottom. The right output of the state variable high-frequency driving module is unidirectionally connected downwards to the right input of the scheduling instruction parsing module. A bidirectional vertical connection channel exists between the storage module and the scheduling instruction parsing module. The lower end of the scheduling instruction parsing module is unidirectionally connected downwards to the fluid distribution module. A unidirectional upward-extending signal bus extends from the left side of the characteristic parameter low-frequency correction module. This signal bus branches off midway, pointing unidirectionally to the left input of the scheduling instruction parsing module, and continues unidirectionally upwards to finally connect to the left input of the state variable high-frequency driving module.

[0038] Example 4: In the water distribution and regulation scenario of this irrigation district, to verify the system's ability to defend against and repair data recording injection and model drift caused by implicit negative events, a distributed monitoring node with multi-source data fusion and analysis capabilities was selected as the test object. The test bench includes a set of high-frequency flow and pressure sensor arrays with a sampling frequency of 100Hz, and is connected to a logic control unit with local edge computing capabilities via an RS485 bus. In normal operation mode, the control unit calls a preset state transition matrix to perform water demand deduction based on the real-time input of the sensors. The preset state transition matrix refers to the matrix used to convert the data collected at the current calculation step size into a state transition matrix. The mapping of multi-depth soil volumetric water content and precipitation data is recursively used to derive the state transition parameter matrix of the controlled root zone water demand state at the next moment. Water demand extrapolation refers to the calculation of the root zone soil moisture deficit and total water demand state deviation scalar based on this matrix. The basis for setting is the state transition coefficient obtained by offline calibration based on the basic soil moisture balance model and the historical time series monitoring data of the controlled field before the system is put into operation. The experiment introduces external interference injection equipment to simulate abnormal operation signals or data frames contaminated by noise at specific physical nodes, which are used as implicit negative event records to test the control unit's identification, isolation and compensation path for abnormal data injection.

[0039] The control unit is internally configured with a data flow feature verification module and a logic mapping processing module. The data flow feature verification module establishes a sliding sampling time window for the input flow of each sensor. By calculating the variance change rate and jump amplitude of the data sequence within the window, it generates a data reliability index in real time. When the logic mapping processing module receives a set of data input from the flow sensor, the data flow feature verification module extracts the instantaneous gradient of the flow signal within the window period. If the gradient value exceeds a preset threshold of the physical system inertia limit, the preset threshold is determined as follows: before the system is put into operation, the physical system inertial time constant is identified by injecting a step drive command into the fluid distribution execution node. The maximum physical flow change rate per unit time limited by the physical system inertial time constant is used as the preset threshold, which means that the input data is interfered with by implicit negative event recording. At this time, the control unit immediately activates the defensive repair logic, suspends the input of damaged data records, starts a state estimation algorithm based on historical time series data, and infers the current logical input value based on the steady-state parameters that passed the last verification before the damage. The inference result is then forcibly injected into the logic mapping processing module to replace the original record that was interfered with.

[0040] For model drift scenarios, the logical mapping processing module deploys a gradient compensation adaptive operator. This operator monitors the accumulation of residuals between the system's predicted output and the actual sensor feedback using a sliding window. When the accumulated residual value exceeds a set warning limit (based on the maximum allowable cumulative deviation of the soil hydrodynamic state in the controlled root zone), the operator executes specific adjustment rules according to the residual polarity: when the residual is positive, a negative reduction adjustment is applied to the corresponding weight parameter in the state transition matrix; when the residual is negative, a positive increment adjustment is applied to the corresponding weight parameter in the state transition matrix, until the residual returns to the value of the controlled root zone. The allowable range of steady-state control accuracy for soil hydrodynamic state in the region is within the baseline bandwidth set by the system. Through this mechanism, the system not only eliminates logical misleading caused by the injection of implicit negative event records, but also realizes dynamic repair of the input path of the damaged model. Experimental data shows that under the condition of continuous injection of pulse disturbances, the system after introducing feature verification and gradient compensation maintains an accuracy of over 98.5% in calculating the scalar of water demand deviation. No systemic control failure caused by damage to a single data link occurred, verifying the technical effectiveness of the constructed defense and repair mechanism in improving the stability of the closed-loop system.

[0041] Example 5: In a data acquisition and control environment with multiple sensor nodes, the controlled physical system operates under non-ideal industrial conditions with data link delays and noise interference. To ensure that the controller can stably output commands in the absence of prior model parameters, the control system is configured with an offline calibration and data initialization process. Before the system is officially put into operation, a set of calibration conditions with different flow rates, velocities, and ambient temperatures is preset. Step drive commands are injected into the fluid distribution execution nodes through the control drive module. At the same time, the transient rate monitoring operator records the response time, overshoot, and steady-state error fed back by the execution nodes in real time. The characteristic parameter low-frequency correction module collects the response data sequence and uses the recursive least squares method to identify the model parameters. The identified physical system inertial time constant and damping ratio are stored in the reference parameter library of non-volatile memory. This calibration process is performed after the initial deployment of the system or after the replacement of major components to ensure that all subsequent calculations of water demand state deviation scalars are based on the current response characteristics of the physical entity.

[0042] When the control system encounters sudden noise interference with the input signal during actual operation, the transient rate monitoring operator detects an abnormally steep slope in the differential calculation of the tipping bucket rain gauge pulse counting sequence or soil water suction data within the sliding window. To prevent this unrealistic jump from causing erroneous responses in the regulating mechanism, the system configures a dynamic smoothing algorithm in the scheduling command parsing module. This algorithm uses the state deviation scalar of the current moment and the previous sampling period as a benchmark to calculate the input increment and compare it with the preset maximum physical change limit. When the input increment exceeds the limit, the control unit automatically enters a smooth transition mode, limiting the update amplitude of the output valve control command to the extreme value range allowed by the unit sampling step, thereby filtering out non-physically continuous sudden command fluctuations. After the signal returns to normal, the scheduling command parsing module recalculates based on the actual pressure deviation of the current system and the theoretical water demand model, and sends a smooth convergence command to the execution node to eliminate residual deviations during the transition process. This mechanism maintains stable control output to the fluid distribution execution node in tests where sensors are damaged or the transmission link experiences intermittent interruptions, and no malfunctions of the actuator induced by link noise occur.

[0043] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A water demand closed-loop coordinated control system based on infiltration model analysis, characterized in that, include: The state variable high-frequency driving module is equipped with a transient variability monitoring operator to receive multi-depth soil hydrodynamic state monitoring signals and precipitation pulse accumulation signals at the first time step. It runs the basic soil moisture balance model to output the total water demand state deviation scalar and uses the transient variability monitoring operator to extract the differential evolution slope of precipitation within the sliding time window. The low-frequency correction module for characteristic parameters inputs stable isotope characteristic values ​​at a second time step greater than the first time step and obtains infiltration safety constraint parameters characterizing the preferential flow replenishment and piston flow replacement mechanism. The parameter latching module maintains a constant output of the infiltration safety constraint parameters. The scheduling instruction parsing module, connected to the state variable high-frequency driving module and the parameter latching module, is used to obtain the total water demand state deviation scalar. When the absolute value of the differential evolution slope is not greater than the set response hysteresis dead zone threshold, it constructs the upper limit of water supply intensity and pulse intermittent constraints in combination with infiltration safety constraint parameters, and converts the total water demand state deviation scalar into a first physical driving signal for suppressing deep bypass leakage, which is then output to the fluid distribution module. When the absolute value is greater than the set response hysteresis dead zone threshold, an over-limit interruption signal is generated and the reading link to the parameter latching module is blocked. The adaptive takeover operator is run synchronously to retrieve the extreme operating condition feedforward weight matrix stored in the storage module and generate a second physical driving signal, which is then sent to the fluid distribution module.

2. The water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The low-frequency correction module for characteristic parameters also includes a smoothing latch submodule, which, when obtaining newly generated infiltration safety constraint parameters, retrieves the infiltration safety constraint parameters from the previous acquisition cycle and generates a dynamic decay forgetting factor in combination with the environmental evolution slope. Based on the dynamic decay forgetting factor, an interpolation transition function is established and written into the parameter latch module, so that the scheduling instruction parsing module receives the infiltration safety constraint parameters that change according to the interpolation transition function during runtime, so as to smoothly update the control boundary of the first physical drive signal.

3. The water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The scheduling instruction parsing module also includes a time-domain waveform peak-shifting scheduling module; when analyzing the total water demand state deviation scalar in conjunction with the infiltration safety constraint parameters, based on the water lateral diffusion hysteresis characteristics of soil microporous matrix suction and the trigger threshold of macroporous preferential flow, a low-frequency long-period driving waveform for normal matrix infiltration and a high-frequency short-pulse driving waveform for preventing surface water accumulation are set; when executing water supply control, the high-frequency short-pulse driving waveform is controlled to be embedded in the intermittent period of the low-frequency long-period driving waveform for output, and the matrix capillary suction during the intermittent period is used to promote the lateral absorption of water and block the gravity infiltration channel of macroporous preferential flow.

4. The water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The state variable high-frequency driving module also includes a soil sensor network interface and a field meteorological observation interface. The soil sensor network interface is connected to a multi-layer soil physical detection array deployed outside the system to periodically read the soil volumetric water content and matrix suction signals at different depths of the longitudinal profile. The field meteorological observation interface is connected to an external tipping bucket rain gauge to acquire tipping bucket pulse counts in real time and accumulate them into a rainfall intensity sequence within a preset sliding time window, so as to send a continuous independent variable data stream to the transient rate monitoring operator.

5. The water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The low-frequency correction module for characteristic parameters also includes a hydrogen and oxygen isotope input interface and a balance analysis calculation module. It receives stable isotope ratios of input water, multi-depth soil water, and shallow groundwater or lower boundary seepage water measured by an external isotope mass spectrometer via the hydrogen and oxygen isotope input interface. The received stable isotope ratios are input into the balance analysis calculation module to calculate deuterium surplus, local precipitation line, and evaporation deviation index. A dual-isotope multi-terminal linear mixing mass balance model is then run to calculate the contribution rate of soil water sources, the proportion of preferential flow in precipitation recharge to shallow groundwater, and piston flow displacement characteristics. Based on these, infiltration safety constraints, including the maximum allowable water supply intensity limit, the maximum single irrigation volume limit, pulse duration, and pulse interval time, are constructed and transmitted to the parameter latching module.

6. The water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The instruction mapping rule matrix in the dispatch instruction parsing module includes the amplitude conversion logarithmic relationship and the time opening conversion rule. When the dispatch instruction parsing module runs the instruction mapping rule matrix, it generates the first control pulse width to control the on-site valve action based on the pulse duration and pulse interval time in the infiltration safety constraint parameters, and generates the first valve opening duty cycle to control the valve core displacement of the on-site regulating valve based on the maximum allowable water supply intensity limit and the total water demand state deviation scalar. The combination of the first control pulse width and the first valve opening duty cycle constitutes the first physical drive signal.

7. A water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The scheduling instruction parsing module is equipped with an adaptive takeover operator. After the signal reading link to the parameter latching module is cut off, the adaptive takeover operator reads the extreme condition feedforward weight matrix stored in the storage module. The precipitation state is determined based on the polarity of the differential evolution slope, and the component coefficients in the extreme condition feedforward weight matrix are matched according to the magnitude of the absolute value of the differential evolution slope to determine the dynamic feedforward compensation weight. The dynamic feedforward compensation weight is used to perform emergency correction or forcibly reduce the upper limit of water supply for the current infiltration safety constraint parameters, and to perform linear proportional product scaling on the total water demand state deviation scalar to obtain the second control pulse width and the second valve opening duty cycle used to combine and constitute the second physical drive signal.

8. A water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The scheduling instruction parsing module also includes an error feedback correction loop, which is used to cyclically read the actual response value of the soil hydrodynamic state of the controlled root zone and calculate the real-time residual between it and the set value after either the first physical drive signal or the second physical drive signal is issued within the current control cycle. The real-time residual is then input to the state variable high-frequency drive module to complete the dynamic feedback bias correction of the total water demand state deviation scalar within the current control cycle.

9. A water demand closed-loop coordinated control system based on infiltration model analysis according to claim 1, characterized in that, The system also includes a fluid pipeline distribution and execution module deployed at the controlled site. The fluid pipeline distribution and execution module includes a flow regulation sub-module, a switch control sub-module, and a pressure transmission sub-module. The flow regulation sub-module and the switch control sub-module are both connected to the output interface of the scheduling command parsing module through a signal transmission bus. The flow regulation module is used to adjust the instantaneous flow limit of the water supply pipeline based on the received duty cycle of the first valve opening or the duty cycle of the second valve opening. The switch control module is used to change the on / off state of pulse water supply and intermittent cycle based on the width of the first or second control pulse received.

Images