Intelligent optimization control method of alkaline water hydrogen production system

Abstract

The invention relates to the technical field of industrial process control, and discloses an intelligent optimization control method of an alkaline water hydrogen production system, which comprises the following steps: a controller synchronously acquires real-time voltage, current and temperature data of an electrolytic bath; theoretical reversible voltage is calculated according to a Nernst equation, ohm voltage drop is eliminated, and a bubble additional impedance value representing the bubble coverage degree is analyzed; performing differential processing on the change rate of the bubble additional impedance value to generate a flow feed-forward control instruction, combining a feedback control instruction based on temperature deviation to synthesize a target rotating speed instruction, and driving an alkali liquor circulating pump to adjust the flow speed; according to the method, the bubble impedance observation model is constructed, so that zero-delay sensing and decoupling control over the state of bubbles in the electrolytic cell are achieved, hysteresis of single temperature feedback is eliminated, and the problems of actuator saturation and model aging drifting are solved.

Description

TECHNICAL FIELD

[0001] The application relates to an intelligent optimization control method of an alkaline water hydrogen production system and belongs to the technical field of industrial process control. BACKGROUND

[0002] The current alkaline water hydrogen production technology is a key path for realizing large-scale renewable energy consumption. In conventional industrial applications, a control system generally adopts an electrolytic tank outlet temperature feedback control strategy to adjust the alkali liquid circulating pump flow rate, removes reaction waste heat by adjusting the flow rate to maintain system heat balance, and meets the basic operation safety requirements under the input power relative stability power grid power supply mode based on the thermal inertia control strategy.

[0003] In the off-grid hydrogen production scene using wind energy or solar energy, the input power of the electrolytic tank presents a high-frequency and large-amplitude random fluctuation characteristic. Under this working condition, the existing temperature feedback control strategy exposes a principle defect. The electrolytic tank is a large thermal capacity physical system, and the temperature response time constant is minutes. The power fluctuation causes the change of the electrochemical reaction and the bubble generation rate on the electrode surface to occur in milliseconds. When the input power instantaneously increases sharply, the bubble generation rate rapidly exceeds the bubble detachment rate under the current flow rate, and the gas film is formed on the electrode surface, which sharply reduces the effective reaction area and causes the ohmic impedance to sharply rise. At this time, the control system cannot generate a flow rate adjustment instruction in time because the large inertia temperature signal has not changed, and the electrolytic tank is in a high local voltage and high thermal stress working condition before the overheating protection is triggered. The existing control model usually regards the electrolytic tank impedance as a single-valued function that only changes with temperature and cannot distinguish between the permanent structural impedance drift caused by the aging of the separation membrane or the decay of the catalyst and the transient impedance change caused by the bubble coverage, so that the system is prone to misoperation or control precision decay during the whole life cycle. For example, the utility model patent with the authorization announcement number CN218596533U discloses a hydrogen production system for producing hydrogen by electrolyzing alkaline water. The scheme improves the connection structure of the hydrogen production, gas-liquid separation and water addition and alkali addition subsystems to build a complete process circulation loop. In the control logic, the alkali liquid cooler cooling water flow rate is adjusted mainly according to the monitored temperature signal to control the electrolytic tank outlet temperature, and the voltage monitoring is used as the basis for interlocking shutdown when the system abnormally operates.

[0004] Therefore, how to construct a control method that can realize real-time decoupling of bubble dynamics and system thermodynamic state, has the ability of actuator saturation compensation and can adapt to equipment aging drift has become a technical problem to be solved by the application. SUMMARY

[0005] To solve the problems in the background art, the technical scheme of the application is as follows: an intelligent optimization control method of an alkaline water hydrogen production system, comprising the following steps: Synchronously collecting real-time voltage data, real-time current data and inlet temperature data of the electrolytic tank according to a preset sampling period; The theoretical reversible voltage is calculated based on the Nernst equation and using real-time current data and inlet temperature data. The difference between the real-time voltage data and the theoretical reversible voltage is then calculated to obtain the total overpotential. The ohmic voltage drop component is calculated by multiplying the preset intrinsic structural impedance reference value with the real-time current data. The ohmic voltage drop component is subtracted from the total overpotential to analyze the bubble-added impedance value that characterizes the degree of bubble coverage on the electrode surface. The time change rate of the bubble additional impedance value is calculated and a flow feedforward control command is generated based on the time change rate. Combined with the feedback control command generated based on the outlet temperature deviation of the electrolyzer, the flow feedforward control command and the feedback control command are weighted and synthesized to generate a target speed command to drive the alkali circulation pump to adjust the alkali flow rate. The system monitors the fluctuation rate of input power. When the fluctuation rate is lower than the preset steady-state threshold, it is determined that the system is in a steady-state operating window. Within the steady-state operating window, the collected historical voltage and current data are processed using the recursive least squares method, the actual structural impedance is calculated by inversion, and the intrinsic structural impedance reference value is updated using the actual structural impedance.

[0006] Preferably, the step of subtracting the ohmic voltage drop component from the total overpotential to resolve the bubble-added impedance value, which characterizes the degree of bubble coverage on the electrode surface, specifically includes: constructing an equivalent circuit model of the electrolytic cell containing variable bubble impedance parameters and fixed structural impedance parameters; using real-time voltage data as the observation input to the equivalent circuit model of the electrolytic cell, and setting the bubble-added impedance value as the state variable to be estimated; using the Kalman filter algorithm to iteratively estimate the state variable, calculating the residual between the model prediction value and the observation input, using the residual to correct the estimated value of the state variable, and confirming the converged state variable as the bubble-added impedance value at the current moment.

[0007] Preferably, the step of generating a target rotational speed command to drive the alkali circulation pump to adjust the alkali flow rate further includes an extreme value optimization logic based on impedance-flow sensitivity. Specifically, the extreme value optimization logic includes: superimposing a periodic flow disturbance signal with constant frequency and amplitude onto the base value of the target rotational speed command; extracting a dynamic impedance response component with the same frequency as the periodic flow disturbance signal from the bubble's additional impedance value using a bandpass filter; and calculating the impedance-flow sensitivity index according to the following formula: ,in, The impedance-flow sensitivity index, The magnitude of the dynamic impedance response component. The amplitude of the periodic flow disturbance signal is used; when the impedance-flow sensitivity index is lower than the preset performance threshold, a weight adjustment command is generated to reduce the weight coefficient of the flow feedforward control command when synthesizing the target speed command.

[0008] Preferably, the method further includes an actuator saturation compensation step based on pressure-flow cross-domain coordination: real-time monitoring of the actual feedback speed of the alkali circulation pump; when the actual feedback speed reaches the preset physical speed limit and the bubble additional resistance value continues to be higher than the preset safety threshold, the pressure coordination control mode is activated; in the pressure coordination control mode, a pressure increase command is generated and sent to the system back pressure valve to simultaneously increase the working pressure setpoints on the hydrogen side and oxygen side of the electrolyzer, and the gas phase volume fraction in the electrolyzer is reduced by utilizing the bubble volume contraction effect caused by the pressure increase; when the bubble additional resistance value falls back to the recovery threshold, a pressure recovery command is generated to restore the working pressure setpoint to the initial state.

[0009] Preferably, the method further includes a flow pattern instability suppression step based on voltage frequency domain characteristics: performing high-pass filtering on real-time voltage data to extract high-frequency voltage noise components; calculating the statistical variance of the high-frequency voltage noise components within a sliding time window; comparing the statistical variance with a preset flow pattern instability threshold, and determining that the flow field inside the electrolyzer is in a gas slugging unsteady state when the statistical variance exceeds the flow pattern instability threshold; generating parameter correction instructions to increase the minimum flow limit of the alkali circulation pump and simultaneously reduce the integral gain parameter used to generate feedback control instructions.

[0010] Preferably, the target speed command driving mode includes a steady-state driving mode and a pulse cleaning mode: when the bubble additional impedance value is lower than the preset pinning threshold, the steady-state driving mode is adopted, and a continuous and smooth voltage signal is output to drive the alkali circulation pump; when the bubble additional impedance value exceeds the pinning threshold and the duration exceeds the preset judgment period, the mode is switched to pulse cleaning mode; in pulse cleaning mode, a pulse modulation signal with a preset frequency oscillation component is generated to drive the alkali circulation pump, and the pulse modulation signal induces transient pressure waves in the alkali pipeline to peel off the bubbles on the electrode surface.

[0011] Preferably, the step of generating a weight adjustment instruction to reduce the weight coefficient of the flow feedforward control instruction when synthesizing the target speed instruction specifically follows the following rules: calling the preset positive correlation mapping function between the weight coefficient and the impedance-flow sensitivity index; substituting the real-time calculated impedance-flow sensitivity index into the positive correlation mapping function to calculate the feedforward weight value at the current moment; and using the feedforward weight value to perform a multiplication operation on the flow feedforward control instruction to generate the attenuated flow feedforward control component.

[0012] Preferably, the step of activating the pressure coordination control mode further includes safety interlock logic: before generating the pressure increase command, the hydrogen-oxygen pressure difference value of the electrolyzer is obtained; when the hydrogen-oxygen pressure difference value is within the preset safety balance range, the operation of synchronously increasing the working pressure setting values ​​of the hydrogen side and oxygen side of the electrolyzer is performed; if the hydrogen-oxygen pressure difference value exceeds the safety balance range, the pressure increase command is blocked and a shutdown protection command is generated.

[0013] Preferably, in the step of collecting real-time voltage data, real-time current data and inlet temperature data of the electrolytic cell, a separate high-speed analog-to-digital conversion module is used to collect real-time voltage data and real-time current data, and noise reduction preprocessing is performed on the collected data; the noise reduction preprocessing includes removing outlier points whose amplitude exceeds the preset normal range, and using a low-pass filter to filter out electromagnetic noise components with frequencies higher than power frequency interference.

[0014] Preferably, the step of updating the intrinsic structural impedance reference value specifically includes: writing the updated intrinsic structural impedance reference value into the controller's non-volatile storage unit; and reading the latest intrinsic structural impedance reference value from the non-volatile storage unit as the initial calculation parameter when the controller is started or reset for the next time, so as to inherit the equipment aging status information of the previous operating cycle.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In an alkaline water hydrogen production system, an observation model is constructed that includes a separation mechanism between bubble-added impedance and inherent structural impedance. This model enables zero-delay sensing and decoupled control of the gas-liquid two-phase flow state inside the electrolyzer. Existing technologies rely solely on temperature feedback to regulate flow rate. Due to the high thermal inertia of the electrolyzer and the rapid bubble generation rate, the control response is prone to lag behind power fluctuations. By utilizing the millisecond-level response characteristics of electrical signals, the theoretical thermodynamic voltage and the ohmic voltage drop generated by the inherent structural impedance are removed from the real-time total voltage. The bubble-added impedance component, which characterizes bubble coverage, is then analyzed. Based on this, the controller generates a flow feedforward command before the temperature changes, directly driving the alkaline solution circulation pump for flushing. Based on a soft-sensor observer feedforward control strategy, the high-frequency fluctuation characteristics of the input power are strictly matched on the time scale, eliminating the inherent lag drawbacks of a single temperature feedback loop. This ensures the reaction area on the electrode surface under high power fluctuation conditions and suppresses the risk of local overheating and voltage exceeding limits.

[0016] 2. Based on the pressure and flow cross-domain coordinated actuator saturation compensation mechanism, it breaks through the adjustment limit of a single flow control variable when the physical actuator is saturated. When the speed of the alkaline circulating pump reaches the physical upper limit or the flow rate cannot be increased due to increased pipeline flow resistance, the controller activates the pressure coordination logic to dynamically increase the system back pressure setpoint. Utilizing the physical principle that gas volume contracts with increasing pressure, it reduces the gas phase volume fraction in the electrolyte by increasing the system working pressure. When fluid dynamics methods fail, thermodynamic methods are used to reduce the additional impedance of bubbles. The multivariable decoupling and complementary control strategy expands the safe operation envelope of the hydrogen production system under extreme overshoot conditions or partial filter blockage conditions, avoiding unplanned shutdowns triggered by actuator capacity limitations.

[0017] 3. An adaptive drift correction mechanism for full life-cycle baseline parameters is introduced to solve the problem of control model inaccuracy caused by equipment aging. It identifies the low-level steady-state operation window where input power and temperature fluctuations are both present. Within this window, the regression algorithm is used to inversely calculate the current actual inherent structural impedance and update the baseline parameters in the observation model. This distinguishes between permanent impedance drift caused by membrane fouling and catalyst decay and transient impedance changes caused by bubble accumulation. It prevents the controller from misinterpreting voltage increases caused by equipment aging as bubble accumulation and executing ineffective high-frequency flow regulation. This ensures that the control system maintains the accuracy of bubble state observation throughout the service life of the electrolyzer and avoids over-flushing or under-regulation due to fixed model parameters. Attached Figure Description

[0018] Fig. 1 This is a flowchart of the bubble impedance decoupling and multivariable collaborative control logic of the present invention; Fig. 2 This is a comparison chart of structural impedance drift and misjudgment rate correction under aging conditions according to the present invention; Fig. 3 This is a diagram of the closed-loop system architecture that integrates high-frequency sensing and coordinated control in this invention. Detailed Implementation

[0019] The following embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Any modifications or substitutions made to the methods, steps, or conditions of the invention without departing from the spirit and substance of the invention shall fall within the scope of the invention.

[0020] This embodiment provides an intelligent optimization control method for an alkaline water hydrogen production system, comprising an electrolyzer, an alkaline solution circulation pump, a power module, a back pressure valve, and a controller. The controller collects system operating data in real time through an analog input interface and controls the speed of the alkaline solution circulation pump and the system back pressure through a frequency converter and a regulating valve interface, respectively. This control method is executed internally by the controller, enabling the sensing and control of the bubble state inside the electrolyzer. The controller synchronously collects the real-time voltage of the electrolyzer according to a preset sampling period. Real-time current and inlet temperature Utilizing real-time current With inlet temperature The theoretical reversible voltage under the current operating condition is calculated based on the Nernst equation. Calculate real-time voltage With theoretical reversible voltage The difference is used to obtain the total overpotential, using a preset intrinsic structural impedance reference value. With real-time current The ohmic voltage drop component is calculated by multiplying the product of the two components. Subtracting the ohmic voltage drop component from the total overpotential allows for the analytical extraction of the bubble-added impedance value, which characterizes the degree of bubble coverage on the electrode surface. In a specific implementation, the controller uses a Kalman filter algorithm to add an impedance value to the bubble. Perform state estimation and confirm the converged state variables as the bubble's additional impedance value at the current moment. Construct discrete state-space equations Define state variables Add an impedance scalar to the bubble to be estimated, and the observation equation The observation matrix For real-time current Observation vector The residual voltage after subtracting the inherent ohmic voltage drop from the total overpotential. Initialize the error covariance matrix. and process noise covariance Measurement of noise covariance Specifically, the process noise covariance Covariance of measurement noise The method for determining this is as follows: measure the noise covariance. The calibration is performed by continuously collecting data under the baseline condition that the electrolytic cell is not powered or is operating at a stable low power. Groups (e.g.) ) Calculate the statistical variance of the residual voltage data after subtracting the inherent ohmic voltage drop from the total overpotential, and set the statistical variance value as . initial value, The value reflects the inherent physical noise level of the voltage / current sensor and sampling circuit; process noise covariance. Tuning: Based on the determination of the physical time constant of bubble formation and desorption, in this embodiment, a combination of offline simulation and actual measurement is adopted: Initialize to Value to The convergence speed of the filter is observed through a step response experiment; if the convergence is too slow, the value is increased. Value; if noise suppression is insufficient, then decrease. In this preferred embodiment, the value is for a sampling period of The working conditions, settings The range of values ​​is ,set up The range of values ​​is Then, according to Periodically perform recursive operations: calculate the Kalman gain Utilizing residuals Corrected state estimate And update the posterior error covariance. For the output sequence Perform moving average filtering, window length After filtering out high-frequency electrochemical noise, the final output is... Furthermore, this embodiment quantifies each key control threshold based on the following physical mechanism: the dynamic response threshold used to trigger flow feedforward control is set to the bubble additional impedance reference value. to This method accurately identifies abnormal bubble buildup rates based on impedance change rate; and sets the safety threshold and recovery threshold involved in pressure-coordinated control as the intrinsic structural impedance reference values, respectively. of and This allows for the construction of a pressure regulation range with hysteresis to prevent actuator oscillation; the pinning threshold used to activate the pulse cleaning mode is set to a baseline value. And the judgment period is set to to Seconds are used to effectively eliminate transient interference; and the flow pattern instability threshold based on the statistical variance of voltage high-frequency noise is set to the steady-state variance value. to The safe balance range of the hydrogen-oxygen pressure difference is limited to [number] times, based on the mechanical resilience of the diaphragm. Within.

[0021] The controller calculates the additional resistance value of the bubble. The time rate of change is used to generate flow feedforward control commands. When the bubble adds resistance value When the flow rate shows an upward trend and the rate of change over time exceeds the preset dynamic response threshold, the flow feedforward control command is activated. Amplitude and bubble additional impedance value The rate of change is positively correlated with the controller, which combines feedback control commands generated based on the outlet temperature deviation of the electrolyzer. For flow feedforward control commands With feedback control commands Perform weighted synthesis to generate the target speed command. The alkali solution circulation pump is driven to regulate the alkali solution flow rate. This process adjusts the flow rate before temperature parameters change, suppressing bubble accumulation. The controller monitors the fluctuation rate of the input power. When volatility When the voltage drops below a preset steady-state threshold, the system is determined to be in a steady-state operating window. Within this window, the collected historical voltage and current data are processed using a recursive least squares method to invert and calculate the actual structural impedance. The recursive least squares method employs a forgetting factor. In the form of the forgetting factor Preferred range is to Initial covariance matrix Set as a larger diagonal matrix (e.g.) When the actual structural impedance is compared with the currently stored intrinsic structural impedance reference value, When the deviation exceeds the preset update dead zone, the update dead zone is set to the intrinsic structural impedance reference value. of to Between, for example This setting is used to filter out model drift caused by slight measurement noise, avoiding unnecessary parameter update operations; the intrinsic structural impedance reference value is updated using this actual structural impedance. The updated intrinsic structural impedance reference value The data is written to the controller's non-volatile memory and read as initial calculation parameters upon the controller's next startup or reset; a target speed command is generated. The steps also include extreme value optimization logic based on impedance-flow sensitivity, whereby the controller responds to the target speed command. A periodic flow disturbance signal with constant frequency and amplitude is superimposed on the base value. The preferred frequency range of the periodic flow disturbance signal is: to The amplitude is preferably set to the current baseline value of the alkali circulation flow rate. Due to the additional resistance value of the bubble Kalman filtering and moving average processing have been performed (window length) This eliminates high-frequency electrochemical noise, and its accuracy and dynamic response are sufficient to support low frequencies. to Dynamic impedance extraction; in addition, due to the additional impedance value of the bubble. Kalman filtering and moving average processing have been performed (window length) This eliminates high-frequency electrochemical noise, and its accuracy and dynamic response are sufficient to support low frequencies. to Dynamic impedance extraction of bubbles is performed using a bandpass filter to extract the additional impedance value. Extracting periodic flow disturbance signals Dynamic impedance response components at the same frequency According to the formula Calculate the impedance-flow sensitivity index The performance threshold is set based on the minimum detectable sensitivity of the system under optimal degassing conditions, and its preferred range is the initial sensitivity of the system under rated operating conditions. peak to When the impedance-flow sensitivity index When the performance falls below a preset performance threshold, a weight adjustment command is generated to reduce the flow feedforward control command. Synthetic target speed command The weighting coefficient at that time.

[0022] The method also includes an actuator saturation compensation step based on pressure-flow cross-domain coordination, and real-time monitoring of the actual feedback speed of the alkali circulation pump. When the actual feedback speed Reaching the preset physical rotation speed limit and the bubble's additional resistance value When the pressure consistently exceeds the preset safety threshold, the pressure coordination control mode is activated. In this mode, the controller acquires the hydrogen-oxygen pressure difference in the electrolyzer, confirms it is within the preset safety balance range, generates a pressure increase command, and sends it to the system back pressure valve. This simultaneously increases the working pressure setpoints on both the hydrogen and oxygen sides of the electrolyzer. The increased pressure causes bubble volume contraction, reducing the gas phase volume fraction within the electrolyzer. When the bubble's additional resistance value... After the pressure drops back to the recovery threshold, a pressure recovery command is generated to restore the working pressure setpoint to its initial state; and a flow pattern instability suppression step based on voltage frequency domain characteristics is implemented for real-time voltage. Perform high-pass filtering to extract high-frequency noise components of the voltage. Calculate the high-frequency noise component of the voltage. Statistical variance within the sliding time window , statistical variance value Compared with the preset flow pattern instability threshold, when the statistical variance value When the flow pattern instability threshold is exceeded, the internal flow field of the electrolyzer is determined to be in a gas-lock unsteady state. A parameter correction command is generated, increasing the minimum flow limit of the alkali circulation pump and simultaneously decreasing the value used to generate the feedback control command. Integral gain parameter; target speed command The driving modes include steady-state driving mode and pulse cleaning mode, when the bubble has an additional resistance value When the voltage drops below the preset pinning threshold, a steady-state drive mode is adopted, outputting a continuous and smooth voltage signal to drive the alkali circulation pump. When the added impedance value of the air bubbles... When the pinning threshold is exceeded and the duration exceeds the preset judgment period, the system switches to pulse cleaning mode. In pulse cleaning mode, a pulse modulation signal with a preset frequency oscillation component is generated to drive the alkaline solution circulation pump, which induces transient pressure waves in the alkaline solution pipeline to peel off the bubbles on the electrode surface.

[0023] Preferably, the pulse modulation signal adopts a square wave or trapezoidal wave form, the preset frequency range is 0.5Hz to 5.0Hz, the amplitude of the oscillation component is 10% to 30% of the current reference speed command, and the duty cycle is set to 40% to 60%.

[0024] Example 1 In industrial scenarios involving off-grid wind power coupled with alkaline hydrogen production, the electrolyzer is directly powered by a wind farm exhibiting high-frequency random fluctuations. A typical operating condition for this system is that the input power may fluctuate dramatically within seconds. Rated load jumps to Rated load causes a sharp change in the current density of the electrolyzer. Existing alkaline water hydrogen production systems mainly rely on temperature feedback loops with high thermal inertia to regulate flow rate, which cannot respond promptly to such millisecond-level gas production rate fluctuations. This easily leads to localized overheating and voltage exceeding limits due to bubble accumulation on the electrode surface. When the system detects a sudden surge in wind power, the control method of this invention intervenes through a parallel-running bubble impedance observer and dual-channel decoupled control logic. The controller uses... The frequency synchronously acquires the real-time voltage of the electrolytic cell. Real-time current and inlet temperature The theoretical reversible voltage under the current operating condition is calculated using the Nernst equation. And subtract the reference value of the inherent structural impedance. The resulting ohmic voltage drop The bubble-added impedance value, which characterizes the degree of bubble coverage, was determined. In the initial stage of a sharp increase in power, although the outlet temperature of the electrolyzer has not yet changed, the additional impedance value of the desorbed bubbles... The rate of change has shown an upward trend.

[0025] At this time, the feedforward control channel responds to the bubble's additional impedance value. The controller calculates the additional resistance value of the bubble due to rapid changes. The time change rate is used to generate a large flow feedforward control command when the change rate exceeds a preset threshold. This instruction is the same as the instruction generated by the temperature feedback loop. After weighted synthesis, the driving alkali circulation pump significantly increases its speed before the temperature field responds. The high-speed alkali flow powerfully washes the electrode surface, using fluid shear force to peel off the newly formed bubble layer. As the bubbles are removed, the bubble-added impedance value... The input power fluctuation rate falls back to a safe range. During continuous operation, if the input power fluctuation rate decreases and enters the steady-state window, the controller uses the recursive least squares method to process the steady-state data, invert the actual structural impedance, and update the intrinsic structural impedance reference value. To eliminate the impact of equipment aging on observation accuracy, when extreme operating conditions cause the circulating pump speed to reach its physical limit and the bubble impedance is still not eliminated, the system automatically activates the pressure coordination control mode. By synchronously increasing the system back pressure to compress the bubble volume and reduce the gas phase volume fraction, the system can still maintain safe operation under overload or abnormal flow resistance conditions. This embodiment verifies that the control method based on bubble impedance observation can effectively overcome the lag of traditional temperature feedback control through its application in a wind power coupled hydrogen production scenario. This method utilizes the fast response characteristics of electrical signals to achieve real-time perception and active intervention of the bubble state inside the electrolyzer. Without adding additional detection hardware, it improves the adaptability and operational safety of the alkaline hydrogen production system to wide power fluctuation power supplies.

[0026] Example 2 This embodiment demonstrates... Experiments were conducted on a pilot-scale alkaline water electrolysis hydrogen production platform. The platform was equipped with a programmable DC power supply to simulate dynamic fluctuations in wind power. High-frequency pressure and temperature sensors were installed at the inlet and outlet of the electrolyzer and at key flow path nodes. The data acquisition system... The sampling rate synchronously records voltage, current, temperature, and pressure data; the test is set up with a typical wind power step condition: the input power is at... From within seconds Rated load step to Under rated load, during this process, a comparative test was conducted using the existing PID temperature feedback-based control strategy (control group) and the feedforward control strategy based on bubble impedance observation of this invention (sample group). In the control group experiment, when the power increased by a step, the electrolytic cell voltage rose rapidly. Due to the lag in temperature response, the outlet temperature initially decreased after the power increase. Within seconds, there was no significant change, causing the circulating pump speed to remain low. During this time, bubbles on the electrode surface could not be removed in time, resulting in a sharp increase in bubble coverage. This manifested as a non-linear rise in the electrolytic cell voltage, occurring after a power step. Seconds, the average voltage of a single cell in the electrolytic cell reaches Exceeding the set limit The alarm threshold triggers the system's overvoltage protection action; in the sample experiment of this invention, the controller calculates the additional impedance value of the bubble in real time. When a power step occurs, the controller in Detected within milliseconds As the rate of change increases, the feedforward channel immediately generates a flow increment command. The speed of the driving circulation pump is at From within seconds Upgraded to The high-speed alkaline solution flow rapidly washes the electrode surface, suppressing the accumulation of bubbles. Table 1 shows the comparison data of key performance indicators under the two control strategies.

[0027] Table 1. Comparison of key performance indicators under the two control strategies

[0028] As shown in Table 1, the sample of this invention achieves an order-of-magnitude improvement in flow response speed (from greater than...). seconds shortened to less than (seconds), effectively controlling peak bubble coverage within The safety level was far lower than that of the control group. The peak voltage of a single cell is limited to To avoid the risk of voltage exceeding limits, and furthermore, the system of the present invention reaches steady state in a shorter time. Furthermore, the rate of temperature rise due to local overheating decreased; to further verify the effectiveness of the pressure-flow cross-domain coordinated compensation mechanism, in another set of experiments, the upper limit of the circulating pump speed was artificially set to... Simulating pump performance limitations or abnormally increased flow resistance, when the power step jumps to At rated load, single flow control cannot suppress bubble coverage below the safe threshold. In this case, the controller activates pressure coordination mode to reduce system back pressure from... Upgraded to Experimental data show that after the back pressure is increased Within seconds, the additional resistance value of the bubble Decrease approximately The system successfully remained in a safe operating range and did not trigger shutdown protection.

[0029] Example 3 This embodiment combines Figs. 1 to 3 This paper describes an intelligent optimization control method for an alkaline water hydrogen production system, such as... Fig. 1 As shown, the control logic begins with the data synchronization acquisition module, which is responsible for the real-time acquisition and noise reduction of voltage, current, and inlet temperature. The acquired data flows to the bubble impedance observation model, where theoretical voltage and ohmic voltage drop are removed to analyze the bubble state. At the same time, the inherent structural impedance reference adaptive update module uses the recursive least squares method to correct the model parameters and update the structural impedance reference value under the steady-state window. Based on the analyzed state system, a flow feedforward control command is generated, and dynamic compensation is generated based on the bubble impedance change rate. The system then enters the target speed command synthesis stage, where a weighted algorithm of flow feedforward and temperature feedback is executed. The synthesized command drives the alkaline circulating pump to adjust the flow rate and actively remove electrode bubbles. If the pump speed reaches the physical upper limit, the pressure and flow cross-domain collaborative compensation module is activated, which dynamically increases the system back pressure when the pump speed is saturated, and finally acts on the system back pressure valve.

[0030] like Fig. 2As shown in the figure, the horizontal axis represents the running time in hours, and the vertical axis represents the percentage (%) or impedance value (mΩ). The figure illustrates the process of the actual structural impedance drift curve increasing from 5mΩ to over 12mΩ over time. Affected by this, the uncorrected bubble coverage misjudgment rate curve shows an upward trend, indicating that the error accumulates with equipment aging. However, the corrected bubble coverage misjudgment rate curve, after introducing an adaptive update mechanism, shows that the impedance drift misjudgment rate continues to decrease and remains low, confirming the algorithm's effectiveness in suppressing model aging drift. Fig. 3 As shown, the core of the architecture is the intelligent optimization controller algorithm, which integrates four major functional modules: bubble impedance decoupling, aging adaptive correction, pressure cross-domain coordination, and flow pattern instability monitoring. It receives input from the high-frequency sensing array through high-frequency data stream and sends coordinated control commands to the execution and control terminal. The execution and control terminal adjusts the flow or pressure of the hydrogen production physical scenario. The state feedback generated by the physical scenario returns to the sensing array, thus forming a complete closed-loop control loop that includes sensing, decision-making, and execution.

[0031] Example 4 This embodiment addresses the issue of control model inaccuracies caused by equipment aging during the entire lifecycle operation of an alkaline electrolyzer. It verifies the effectiveness of a lifecycle-wide baseline parameter adaptive correction mechanism. In practical industrial applications, impurity deposition on the electrolyzer membrane and degradation of the catalyst active layer can lead to inherent structural impedance. Slow and irreversible drift occurs. If the control system uses constant impedance parameters, the increase in ohmic voltage drop caused by aging will be incorrectly attributed to bubble coverage, leading to incorrect bubble coverage observations. Inflated input power levels can lead to ineffective high-flow-rate flushing operations, increasing energy consumption and accelerating equipment wear. To address this issue, this invention employs a steady-state window-based data inversion correction logic, with the controller monitoring the input power fluctuation rate in real time. and rate of temperature change ,when Persistently below and Persistently below Exceeding the preset time, such as At the specified time, the system is determined to have entered the quasi-steady-state correction window. Within this window, the controller... Frequency acquisition of multiple voltage groups With current The data, combined with recursive least squares (RLS) and a pre-defined polarization curve model, is used to identify the current estimated value of the actual structural impedance online. If newly identified Compared with the current benchmark value If the deviation exceeds the preset dead zone, the controller uses an exponentially weighted moving average strategy to smoothly update the benchmark value and then updates the value accordingly. The data is written to non-volatile memory as the reference parameters for subsequent bubble impedance calculation. Table 2 shows the comparison data of key performance indicators of the system under simulated aging conditions before and after the adaptive correction mechanism is enabled.

[0032] Table 2 Comparison of the effects of adaptive correction mechanisms under aging chemical conditions

[0033] As shown in Table 2, after enabling adaptive correction, the system can track the slow drift of the electrolyzer impedance, reducing the misjudgment rate of bubble coverage under steady-state conditions from... Reduce to This improvement eliminates most of the ineffective flushing operations, reducing the proportion of high-speed operation time of the circulating pump from [previous percentage]. Significantly decreased to less than This effectively extends the service life of the actuator and controls the additional energy consumption caused by misoperation within a certain range. The low level is better than the uncorrected level. .

[0034] Example 5 In large-scale hydrogen production parks integrating wind and solar power generation, this invention introduces a standardized baseline calibration and data initialization procedure to address the initial impedance differences between different batches of electrolyzers caused by manufacturing tolerances. Before the system is first put into operation or before replacing core components, the controller enters an initialization calibration mode. In this mode, the controller drives the power module to output a series of stepped constant current excitation signals, covering the electrolyzer from... to Within the rated current range, at each current step, after the system waits for the voltage to stabilize, multiple sets of voltage and current data are recorded synchronously. The initial volt-ampere characteristic curve of this specific electrolytic cell is fitted using the least squares method, and the intrinsic structural impedance reference value is analyzed. .

[0035] Furthermore, considering the unique characteristics of the cold start-up phase of electrolytic cells under extreme climatic conditions, such as extremely cold environments, this invention constructs a start-up parameter correction logic based on ambient temperature. When the ambient temperature sensor detects that the temperature is lower than a preset low-temperature threshold, such as... When the controller automatically calls the low-temperature start-up parameter table, it sets the applicable start-up current ramp-up rate limit and the target circulation flow rate setpoint for the preheating stage under low-temperature conditions. The system strictly follows this correction logic in the early stage of start-up, limiting the rate of increase of input power and maintaining a high electrolyte circulation flow rate to prevent the risk of local overheating caused by excessive electrolyte viscosity or insufficient conductivity, until the inlet temperature of the electrolytic cell rises back to the normal operating range, at which point the system automatically switches back to the conventional control mode.

[0036] Example 6 This embodiment establishes a standardized system initialization parameter calibration and adaptive tuning procedure. After the system hardware is installed and the electrical insulation test is completed, the self-test and calibration module built into the controller automatically executes the procedure. The calibration procedure performs static parameter identification, and the controller drives the power supply module to inject a series of stepped current signals with increasing amplitude into the electrolytic cell. The current range covers the rated current. to The holding time for each step is set to To ensure voltage stability, the controller synchronously collects steady-state voltage values ​​under each step current during this period, constructing a measured volt-ampere characteristic dataset. Linear regression analysis of this dataset is then performed using the least squares method to determine the initial intrinsic structural impedance reference value for this specific electrolytic cell. and zero current intercept voltage, calculated The parameters written to the controller's non-volatile memory are used as initial reference parameters for the subsequent real-time bubble impedance observation model, thereby eliminating initial model deviations caused by manufacturing tolerances. A hard constraint check of the linear regression fit is then performed: the Pearson product-moment correlation coefficient of the regression model is calculated. and residual sum of squares ;like or Exceeding the preset dispersion threshold To determine if the electrolytic cell electrode plate assembly consistency is incompatible or if the sensor zero-point drift is abnormal, the following measures are taken: The storage write operation is performed, and the failure status flag is set; if the absolute value of the deviation between the zero current intercept voltage and the theoretical open circuit voltage is... If the short circuit or bypass current leakage alarm of the electrolysis chamber is triggered, the closed-loop operation mode will be prohibited.

[0037] And the dynamic response parameter setting of the procedure, the controller at the rated current At the operating point, the superimposed amplitude is , frequency is to The controller analyzes the frequency response characteristics of the electrolytic cell voltage to the variable-frequency sinusoidal disturbance current signal, particularly the amplitude and phase frequency responses in the high-frequency range. This analysis allows the controller to identify the equivalent double-layer capacitance and electrochemical reaction time constant of the electrolytic cell. Based on this identified time constant, the controller automatically calculates and tunes the filter time constant in the bubble impedance observer and the differential gain coefficient of the feedforward control channel using the internal model control principle. Specifically, the filter time constant is set to the electrochemical reaction time constant. The differential gain coefficient is adjusted by normalization according to the disturbance response amplitude to ensure dynamic matching between the feedforward compensation and the bubble generation rate. Finally, the procedure performs a safety boundary check. After the system is heated to the rated operating temperature, the controller gradually increases the speed of the circulating pump while monitoring the pressure difference between the inlet and outlet of the electrolytic cell. When the pressure difference reaches the preset mechanical safety upper limit, the pump speed is adjusted accordingly. At this time, record the pump speed value and set it as the maximum allowable speed of the system under the current pipeline resistance characteristics. ,Should The value is used as the saturation limit parameter in the feedforward control logic and the pressure collaborative compensation logic.

[0038] Example 7 This embodiment constructs a real-time bubble state perception and feedforward control strategy for electrical signals. During implementation, a high-frequency sampling module is used to... The frequency synchronously acquires the real-time voltage of the electrolytic cell. Real-time current and inlet temperature Using the Nernst equation, the controller calculates the theoretical reversible voltage under the current operating condition. And subtract the reference value of the inherent structural impedance. The resulting ohmic voltage drop The bubble-added impedance value, which characterizes the degree of bubble coverage, was determined. When the input power is monitored at From within seconds Rated load jumps to At rated load, the controller responds to the additional impedance value of the bubble. The rapid upward trend was observed by calculating the additional resistance value of the bubble. The time change rate is considered, and when this change rate exceeds a preset threshold, the controller directly generates a large-flow feedforward control command. The alkali circulation pump is driven to significantly increase its speed before the temperature field responds, utilizing fluid shear force to peel off the newly formed bubble layer on the electrode surface. Experimental data show that this feedforward control strategy reduces the flow response delay time from that of traditional temperature feedback control. shortened from more than a second to Within seconds, the peak bubble coverage is effectively controlled within... The safety level was far lower than that of the control group. .

[0039] To address the equipment aging issue caused by long-term operation of electrolytic cells, this embodiment also integrates a full life-cycle baseline parameter adaptive correction mechanism. This mechanism is applied when the system experiences input power fluctuations. Below And the rate of temperature change Below During the steady-state window, the controller uses the recursive least squares method to process the steady-state data, invert the actual structural impedance, and update the intrinsic structural impedance reference value. This eliminates the risk of erroneously attributing the increase in ohmic voltage drop caused by aging to bubble coverage, thus reducing the misjudgment rate of bubble coverage under steady-state conditions from... Reduce to And reduce the proportion of ineffective high-speed operation time of the circulating pump to less than In addition, to address the actuator saturation problem under extreme operating conditions, the system is configured with pressure-flow cross-domain collaborative compensation logic. When the circulating pump speed reaches the physical upper limit and the bubble resistance is still not eliminated, the controller automatically activates the pressure collaborative mode to synchronously increase the system back pressure to compress the bubble volume and reduce the gas phase volume fraction.

[0040] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for intelligent optimization control of an alkaline water electrolysis hydrogen production system, characterized in that, The method comprises the following steps: Synchronously collecting real-time voltage data, real-time current data and inlet temperature data of the electrolytic cell according to a preset sampling period; Calculating a theoretical reversible voltage according to the Nernst equation and using the real-time current data and the inlet temperature data, and calculating a difference between the real-time voltage data and the theoretical reversible voltage to obtain a total overpotential; Calculating an ohmic pressure drop component by multiplying a preset inherent structural impedance reference value by the real-time current data, subtracting the ohmic pressure drop component from the total overpotential, and analyzing a bubble additional impedance value representing a bubble coverage degree of an electrode surface; Calculating a time variation rate of the bubble additional impedance value, generating a flow rate feedforward control instruction based on the time variation rate, combining a feedback control instruction generated based on an outlet temperature deviation of the electrolytic cell, weighting and synthesizing the flow rate feedforward control instruction and the feedback control instruction, and generating a target rotating speed instruction to drive the caustic lye circulating pump to adjust a caustic lye flow rate; Monitoring a fluctuation rate of input power, determining that the system is in a steady state operation window when the fluctuation rate is lower than a preset steady state threshold, processing collected historical voltage and current data by using a recursive least square method in the steady state operation window, and inversely calculating an actual structural impedance and updating the inherent structural impedance reference value by using the actual structural impedance.

2. The intelligent optimization control method of the alkaline water hydrogen production system according to claim 1, characterized in that, The step of subtracting the ohmic pressure drop component from the total overpotential to analyze the bubble additional impedance value representing the bubble coverage degree of the electrode surface specifically comprises: constructing an equivalent circuit model of the electrolytic cell containing a variable bubble impedance parameter and a fixed structural impedance parameter; taking the real-time voltage data as an observation input of the equivalent circuit model of the electrolytic cell, and setting the bubble additional impedance value as a state variable to be estimated; iteratively estimating the state variable by using a Kalman filtering algorithm, calculating a residual between a model prediction value and the observation input, correcting an estimated value of the state variable by using the residual, and confirming the converged state variable as the bubble additional impedance value at the current time.

3. The intelligent optimization control method of the alkaline water hydrogen production system according to claim 1, characterized in that, The step of generating the target rotating speed instruction to drive the lye circulating pump to adjust the lye flow rate further comprises an extremum optimization logic based on impedance-flow sensitivity, and the extremum optimization logic specifically comprises: superimposing a periodic flow disturbance signal with constant frequency and amplitude on a base value of the target rotating speed instruction; extracting a dynamic impedance response component with the same frequency as the periodic flow disturbance signal from the bubble additional impedance value by using a band-pass filter; and calculating an impedance-flow sensitivity index according to the following formula: wherein, is the impedance-flow sensitivity index, is an amplitude of the dynamic impedance response component, is an amplitude of the periodic flow disturbance signal; and when the impedance-flow sensitivity index is lower than a preset efficiency threshold, generating a weight adjustment instruction to reduce a weight coefficient of the flow feedforward control instruction in synthesizing the target rotating speed instruction.

4. The intelligent optimization control method of the alkaline water hydrogen production system according to claim 1, characterized in that, The intelligent optimization control method further comprises an actuator saturation compensation step based on pressure-flow cross-domain cooperation: monitoring an actual feedback rotating speed of the caustic lye circulating pump in real time; when the actual feedback rotating speed reaches a preset physical rotating speed upper limit and the bubble additional impedance value continuously exceeds a preset safety threshold, a pressure cooperation control mode is activated; in the pressure cooperation control mode, a pressure lifting instruction is generated and sent to a system back pressure valve, the working pressure set values of the hydrogen side and the oxygen side of the electrolytic cell are synchronously increased, and the gas phase volume fraction in the electrolytic cell is reduced by using a bubble volume shrinkage effect caused by the pressure rise; when the bubble additional impedance value falls back to a recovery threshold, a pressure recovery instruction is generated to restore the working pressure set values to the initial state.

5. The intelligent optimization control method of an alkaline water electrolysis hydrogen production system according to claim 1, characterized in that, The intelligent optimization control method further comprises a flow pattern instability suppression step based on voltage frequency domain characteristics: performing a high-pass filtering process on the real-time voltage data to extract a voltage high-frequency noise component; calculating a statistical variance value of the voltage high-frequency noise component in a sliding time window; comparing the statistical variance value with a preset flow pattern instability threshold; when the statistical variance value exceeds the flow pattern instability threshold, it is determined that the internal flow field of the electrolytic cell is in a gas plug non-steady state; a parameter correction instruction is generated to increase a minimum flow rate limit value of the caustic lye circulating pump and simultaneously reduce an integral gain parameter used to generate the feedback control instruction.

6. The intelligent optimization control method of an alkaline water electrolysis hydrogen production system according to claim 1, characterized in that, The driving mode of the target speed instruction includes a steady driving mode and a pulse cleaning mode: when the bubble additional impedance value is lower than a preset pinning threshold, the steady driving mode is adopted to output a continuous and smooth voltage signal to drive the lye circulating pump; When the bubble additional impedance value exceeds the pinning threshold and the duration exceeds a preset determination period, the pulse cleaning mode is switched to; in the pulse cleaning mode, a pulse modulation signal with a preset frequency oscillation component is generated to drive the lye circulating pump, and transient pressure waves are induced in the lye pipeline to peel off the bubbles on the electrode surface.

7. The intelligent optimization control method of the alkaline water hydrogen production system according to claim 3, characterized in that, The step of generating a weight adjustment instruction to reduce the weight coefficient of the flow feedforward control instruction in synthesizing the target speed instruction specifically follows the following rules: a preset positive correlation mapping function of the weight coefficient and the impedance-flow sensitivity index is called; the impedance-flow sensitivity index calculated in real time is substituted into the positive correlation mapping function to calculate the front weighting value at the current time; the flow feedforward control instruction is multiplied by the front weighting value to generate the decayed flow feedforward control component.

8. The intelligent optimization control method of an alkaline water electrolysis hydrogen production system according to claim 4, characterized in that, The step of activating the pressure cooperative control mode further includes safety interlocking logic: before generating the pressure lifting instruction, the hydrogen-oxygen pressure difference value of the electrolytic cell is obtained; when the hydrogen-oxygen pressure difference value is within a preset safety balance range, the operation of simultaneously increasing the working pressure set values of the hydrogen side and the oxygen side of the electrolytic cell is performed; if the hydrogen-oxygen pressure difference value exceeds the safety balance range, the pressure lifting instruction is locked and a shutdown protection instruction is generated.

9. The intelligent optimization control method of the alkaline water hydrogen production system according to claim 1, characterized in that, In the step of collecting real-time voltage data, real-time current data and inlet temperature data of the electrolytic cell, independent high-speed analog-to-digital conversion modules are used to collect real-time voltage data and real-time current data, and the collected data are subjected to denoising pretreatment; The denoising pretreatment includes rejecting wild value points with amplitudes exceeding a preset normal range, and filtering out electromagnetic noise components with frequencies higher than power frequency interference by using a low-pass filter.

10. The intelligent optimization control method of an alkaline water electrolysis hydrogen production system according to claim 1, characterized in that, The step of updating the inherent structure impedance reference value specifically includes: writing the updated inherent structure impedance reference value into a non-volatile storage unit of the controller; when the controller is next started or reset, reading the latest inherent structure impedance reference value in the non-volatile storage unit as an initial calculation parameter to inherit the device aging state information of the last running period.

Images