Micro-grid hydrogen production system and method based on fuzzy logic control

Abstract

The invention relates to the technical field of hydrogen production, in particular to a micro-grid hydrogen production system and method based on fuzzy logic control. Comprising a renewable energy power generation unit, an electrolytic cell array, a multi-type energy storage device, a power divider and a fuzzy logic controller, the renewable energy power generation unit is connected with the electrolytic cell array and the power divider through a direct-current bus, and the output end of the power divider is connected with the storage battery pack and the super-capacitor bank. The fuzzy logic controller collects the bus voltage fluctuation ratio, the electrolytic cell temperature gradient and the energy storage charge state in real time to serve as input variables, and outputs a multi-target optimization instruction to the power divider. Wherein the system dynamically adjusts the start-stop combination and the energy storage charge-discharge priority of the electrolytic cell array. According to the scheme, the mindset that fluctuation suppression and life management are separated in the prior art can be broken through, and a new paradigm of multi-parameter cooperative control is established.

Description

Technical Field

[0001] The present invention relates to the field of hydrogen production technology, and in particular to a microgrid hydrogen production system and method based on fuzzy logic control. Background Art

[0002] In renewable energy hydrogen production systems, the conflict between the intermittent nature of wind and solar power generation and the operational stability of electrolyzers has become a key bottleneck hindering the industry's development. Existing technologies generally use fixed-threshold PID control strategies to regulate power allocation. This rigid control model has two fundamental flaws: First, when wind and solar power fluctuate on the order of seconds to minutes, the response delay of traditional controllers causes the electrolyzer to experience mechanical stress exceeding safety thresholds, accelerating membrane electrode aging. Experimental data shows that frequent power surges can shorten the life of alkaline electrolyzers by over 40%. Second, existing solutions use voltage fluctuation rate as a single control parameter, completely ignoring the multi-parameter coupling effects of electrolyzer temperature gradients and energy storage status. This leads to a chain reaction of over-discharge of energy storage devices and thermal runaway of electrolyzers when the system is operating under extreme conditions such as persistent rainy weather. More seriously, current mainstream technologies have structural flaws in their adaptability to power fluctuations. The fluctuation spectrum of wind and solar power generation typically ranges from 0.1Hz to 10Hz, while traditional solutions can only effectively handle low-frequency fluctuations below 0.05Hz. For high-frequency components, either they are allowed to impact the electrolyzer or the system is abruptly shut down. This either-or control logic has resulted in annual equivalent operating hours for existing systems generally falling below 4,000 hours, far below design expectations.

[0003] The industry urgently needs a technical solution that can simultaneously meet three core requirements: smoothing power fluctuations on a millisecond timescale, optimizing equipment lifespan loss on an hourly timescale, and enabling adaptive regulation across the entire frequency band. This solution must break away from the existing mindset of separating fluctuation suppression from lifespan management and establish a new paradigm for coordinated multi-parameter control. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and propose a microgrid hydrogen production system based on fuzzy logic control, which includes a renewable energy power generation unit, an electrolyzer array, multiple types of energy storage devices, a power distributor and a fuzzy logic controller; the renewable energy power generation unit is connected to the electrolyzer array and the power distributor through a DC bus, and the output end of the power distributor is connected to a battery group and a supercapacitor group respectively; the fuzzy logic controller collects the bus voltage fluctuation rate, the electrolyzer temperature gradient and the energy storage charge state as input variables in real time, and outputs multi-objective optimization instructions to the power distributor; the system dynamically adjusts the start-stop combination of the electrolyzer array and the energy storage charging and discharging priority.

[0005] Preferably, the power distributor includes three levels of power channels: the first channel is connected to the alkaline electrolytic cell group and adopts a thyristor voltage regulation module to achieve coarse power adjustment; the second channel is connected to the PEM electrolytic cell group and is equipped with an IGBT chopper circuit to achieve fine power adjustment; the third channel is connected to the hybrid energy storage device and achieves fast power compensation through a bidirectional DC-DC converter; the coordinated control of the three-level channels adopts a priority dynamic rotation mechanism, which automatically increases the response weight of the energy storage channel when a continuous power shortage is detected, and resets the initial distribution ratio after power recovery.

[0006] Further preferred, it is characterized in that the fuzzy logic controller includes three layers of decision modules: the primary module uses interval type-2 fuzzy sets to process the time domain fluctuation characteristics of the bus voltage, and has a built-in sliding time window analysis algorithm to identify the fluctuation pattern; the intermediate module uses the TSK fuzzy model to calculate the safe operation range of the electrolytic cell, and integrates the temperature field finite element simulation data to establish a three-dimensional mapping relationship; the advanced module integrates the outputs of the primary and intermediate modules, and generates multi-objective optimization instructions that take into account economy, equipment life and grid scheduling requirements through a multi-attribute decision algorithm.

[0007] Further preferably, the start-stop control of the electrolytic cell array satisfies the following constraints: the number of starts and stops per day does not exceed 3 times and the interval between adjacent starts and stops is greater than 4 hours; the power difference between adjacent electrolytic cell groups does not exceed 15% of the rated value and the gradient change rate is less than 5% / minute; preheating to above 80°C and maintaining for 30 minutes must be completed before cold start; the system trains a neural network through historical operation data to predict the optimal start-stop timing, and embeds the prediction results in the fuzzy rule base as a constraint correction factor.

[0008] Further preferably, the severity of power fluctuation is calculated using the following fuzzy membership function: ; in is the measured power fluctuation rate, is the benchmark volatility threshold, which is 5%. It is an adjustable parameter with a value range of 0.1-0.3; when When the energy storage compensation is triggered, Adjust the electrolytic cell power when Only the fluctuation data needs to be recorded.

[0009] Further preferably, the power distribution ratio of the electrolytic cell group is calculated by the following formula: ; in is the power deviation, 、 、 It is a fuzzy self-tuning parameter. Its adjustment rules include: when SOC>80%, increase value; decreases when the temperature gradient is >3℃ / min value; when the fluctuation frequency is greater than 0.1Hz, it increases value.

[0010] Further preferably, the optimization objective function of the advanced module is: ; in: is the weight coefficient , is the unsuppressed power, is the electrolytic cell temperature gradient, is the life loss cost; the weight coefficient is updated every 5 minutes through the fuzzy inference engine, and the update rule takes into account the electricity price period and the remaining capacity of the energy storage.

[0011] A method, applied to a microgrid hydrogen production system based on fuzzy logic control as described in any one of the above, comprises the following steps: Real-time collection of wind and solar power output curves, bus voltage harmonic content, and data from various temperature measurement points of electrolyzers; Calculate the power fluctuation level and mark the characteristics of the fluctuation source through three-level fuzzy reasoning; Select electrolyzer power redistribution, energy storage compensation, or generation unit limiting mode based on fluctuation level and source characteristics; Execute power allocation instructions and monitor the current balance between electrolytic cell groups in real time; evaluate the stabilization effect every 30 minutes and update the correction coefficient and weight distribution of the fuzzy rule base.

[0012] Further preferred, redundant rules that have been triggered less than 3 times in the past 2 hours and have a confidence level lower than 0.6 are deleted; adjacent rules with a similarity greater than 85% and a control effect difference less than 5% are merged; emergency rules based on the analysis of the last 10 large fluctuation events are added, and event feature labels are marked; the center point position and widening parameters of the membership function are adjusted to increase the probability that the newly collected data falls within the valid membership interval to more than 90%.

[0013] Further preferably, the hierarchical protection protocol is activated when the following abnormal conditions are detected: the voltage difference between the electrolytic cell groups exceeds 10% of the rated value for 30 seconds and is accompanied by an abnormal temperature rise; the energy storage SOC suddenly changes by more than 15% within 5 minutes and causes power distribution oscillation; the fuzzy controller outputs conflicting instructions three times in a row and self-checks without hardware faults; the execution order of the protection protocol is to first cut off the power generation unit with the largest fluctuation and start the backup power supply, then switch to the preset fixed parameter control mode, and finally trigger the sound and light alarm and upload the fault diagnosis report to the monitoring center.

[0014] Technical effect: This solution constructs a microgrid system architecture that integrates renewable energy power generation, electrolysis hydrogen production and intelligent control. The system connects renewable energy power generation units such as photovoltaic / wind power to a hybrid electrolyzer array consisting of alkaline electrolyzers and PEM electrolyzers through a DC bus, and is equipped with a hybrid energy storage device including batteries and supercapacitors. The core innovation lies in the use of a fuzzy logic controller to monitor the three key parameters of bus voltage fluctuation rate, electrolyzer temperature gradient and energy storage charge state in real time, and to generate power allocation instructions through multivariable decision-making; it solves the technical problems that traditional microgrid hydrogen production systems have a lag in responding to power fluctuations and are difficult to adapt to the intermittent characteristics of wind and solar power generation, the existing electrolyzer control strategy often ignores the impact of temperature gradient changes on equipment life, and the energy storage device charging and discharging strategy lacks coordinated optimization with the electrolyzer operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is the block diagram of the microgrid hydrogen production system based on fuzzy logic control in this application; Figure 2 This is a flow chart of the microgrid hydrogen production method based on fuzzy logic control in this application. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] See also Figure 1-Figure 2 , such as the following technical problems in traditional technical solutions: Traditional microgrid hydrogen production systems have a lag in responding to power fluctuations, making it difficult to adapt to the intermittent characteristics of wind and solar power generation; existing electrolyzer control strategies often ignore the impact of temperature gradient changes on equipment life; and there is a lack of coordinated optimization between the energy storage device charging and discharging strategies and electrolyzer operations.

[0018] Based on this, this embodiment proposes a microgrid hydrogen production system based on fuzzy logic control, including a renewable energy power generation unit, an electrolyzer array, multiple types of energy storage devices, a power distributor and a fuzzy logic controller; the renewable energy power generation unit is connected to the electrolyzer array and the power distributor through a DC bus, and the output end of the power distributor is connected to a battery group and a supercapacitor group respectively; the fuzzy logic controller collects the bus voltage fluctuation rate, the electrolyzer temperature gradient and the energy storage charge state as input variables in real time, and outputs a multi-objective optimization instruction to the power distributor; the system dynamically adjusts the start-stop combination of the electrolyzer array and the energy storage charging and discharging priority.

[0019] It is worth mentioning that this embodiment constructs a microgrid system architecture that integrates renewable energy generation, electrolysis hydrogen production, and intelligent control. The system connects renewable energy generation units such as photovoltaic / wind power to a hybrid electrolyzer array consisting of alkaline electrolyzers and PEM electrolyzers through a DC bus, and is equipped with a hybrid energy storage device containing batteries and supercapacitors. The core innovation lies in the use of a fuzzy logic controller to monitor three key parameters in real time: bus voltage fluctuation rate to reflect power fluctuations, electrolyzer temperature gradient to reflect equipment status, and energy storage charge state to reflect energy storage capacity. Power allocation instructions are generated through multivariable decision-making.

[0020] The technical effects of the above solution include: This embodiment can improve the efficiency of power fluctuation smoothing and reduce the life loss of the electrolyzer through multi-parameter fusion control; the system response time is shortened from 5-10 seconds of traditional PID control to less than 1 second and the optimal balance between hydrogen production efficiency and equipment life is achieved.

[0021] For example, traditional technical solutions have the following technical problems: A single power regulation channel cannot cover the full-band fluctuation characteristics of wind and solar power generation; fixed priority allocation can easily lead to premature exhaustion of energy storage devices; there is a lack of coordination between electrolyzer power regulation and energy storage compensation. Based on this, the power distributor includes three-level power channels: the first channel is connected to the alkaline electrolyzer group, using a thyristor voltage regulator module to achieve coarse power adjustment; the second channel is connected to the PEM electrolyzer group, equipped with an IGBT chopper circuit to achieve fine power adjustment; the third channel is connected to the hybrid energy storage device, and fast power compensation is achieved through a bidirectional DC-DC converter; the coordinated control of the three-level channels adopts a dynamic priority rotation mechanism, which automatically increases the response weight of the energy storage channel when a continuous power shortage is detected, and resets the initial allocation ratio after power recovery.

[0022] Notably, this embodiment features a refined three-channel power divider design. The first channel uses a thyristor voltage regulator module to control the alkaline electrolyzer group, enabling minute-by-minute power regulation. The second channel drives the PEM electrolyzer group via an IGBT chopper circuit, enabling second-by-second power tracking. The third channel utilizes a bidirectional DC-DC converter to manage the hybrid energy storage device, providing millisecond-by-millisecond power compensation. An innovative dynamic priority rotation mechanism is introduced to automatically adjust the response weight of each channel based on the duration of the power shortfall.

[0023] The technical effects of the above embodiment include: achieving effective smoothing of power fluctuations in the full frequency range of 0.1Hz-10Hz It can extend the cycle life of the energy storage device, reduce the power distribution error from ±8% of the traditional solution to within ±3%, and extend the dynamic rotation mechanism to enable the system to provide continuous power supply time in continuous rainy weather.

[0024] For example, traditional technical solutions have the following technical problems: The traditional type I fuzzy controller has insufficient ability to handle uncertainty; the thermal stress assessment of the electrolytic cell lacks a quantitative basis and the weight distribution of each indicator is unreasonable during multi-objective optimization.

[0025] Based on this, the fuzzy logic controller includes three layers of decision-making modules: the primary module uses interval type-2 fuzzy sets to process the time-domain fluctuation characteristics of the bus voltage, and has a built-in sliding time window analysis algorithm to identify fluctuation patterns; the intermediate module uses the TSK fuzzy model to calculate the safe operating range of the electrolyzer, and integrates the temperature field finite element simulation data to establish a three-dimensional mapping relationship; the advanced module integrates the outputs of the primary and intermediate modules, and generates multi-objective optimization instructions that take into account economy, equipment life and grid scheduling requirements through a multi-attribute decision-making algorithm.

[0026] It is worth mentioning that this embodiment constructs a three-layer progressive fuzzy decision-making architecture: the primary module uses interval type-2 fuzzy sets to process voltage fluctuations and identifies fluctuation patterns through a sliding time window (typical value 30 seconds); the intermediate module establishes a safe operation map of the electrolyzer based on the TSK fuzzy model and temperature field finite element data (grid accuracy 0.5mm³); and the advanced module uses a multi-attribute decision-making algorithm (AHP-TOPSIS hybrid algorithm) to comprehensively consider economic efficiency, lifespan, and grid demand to generate optimization instructions.

[0027] The technical effects of the above scheme include: being able to accurately identify voltage fluctuations and predict the hot spot temperature of the electrolyzer with an error of ≤2°C; and shortening the multi-objective optimization solution time from 30 seconds to 3 seconds, which can improve the response qualification rate of power grid dispatching instructions.

[0028] For example, traditional technical solutions have the following technical problems: The traditional timed start-stop strategy causes premature fatigue failure of the electrolytic cell's mechanical components, slow response to sudden power fluctuations, and an excessively high proportion of preheating energy consumption in the system's total energy consumption.

[0029] Based on this, the start-stop control of the electrolytic cell array meets the following constraints: the number of starts and stops per day does not exceed 3 times and the interval between adjacent starts and stops is greater than 4 hours; the power difference between adjacent electrolytic cell groups does not exceed 15% of the rated value and the gradient change rate is less than 5% / minute; preheating to above 80°C and maintaining it for 30 minutes must be completed before cold start; the system trains a neural network through historical operating data to predict the optimal start-stop timing, and embeds the prediction results in the fuzzy rule base as constraint correction factors.

[0030] Notably, this embodiment constructs an intelligent start-stop control system for the electrolyzer array. By combining a neural network prediction model (LSTM network structure with 128 hidden layer nodes) with triple hard constraints, it achieves dynamic optimization of the electrolyzer operating status. A dynamic constraint relaxation mechanism is specifically designed. When urgent grid demand is detected, a fuzzy penalty function (with an adjustable penalty coefficient of 0.2-0.8) temporarily relaxes the daily start-stop limit while maintaining the temperature gradient constraint. Every 15 minutes, the system collects 12-dimensional characteristic parameters of the electrolyzer array, such as voltage balance and temperature distribution curves, and inputs them into the prediction model to generate the optimal start-stop schedule for the next two hours.

[0031] It is worth mentioning that the above embodiment can reduce the mechanical failure rate of the electrolytic cell group, shorten the emergency power response time, reduce preheating energy consumption, and reduce electricity losses. It can also control the temperature gradient within 2°C / min and reduce thermal stress damage by 37%.

[0032] For example, traditional technical solutions have the following technical problems: The use of a fixed threshold method can easily lead to frequent false triggering, and there is a lack of quantitative standards for assessing the severity of fluctuations. The response characteristics of different electrolyzer types vary greatly.

[0033] Based on the above problems, this embodiment uses the following fuzzy membership function to calculate the severity of power fluctuations: ; in is the measured power fluctuation rate, is the benchmark volatility threshold, which is 5%. It is an adjustable parameter with a value range of 0.1-0.3; when When the energy storage compensation is triggered, Adjust the electrolytic cell power when Only the fluctuation data needs to be recorded.

[0034] The Gaussian membership function described above is the core quantitative tool for fuzzy logic control, and its components form a precise technical closed loop: Input variable x represents the measured power fluctuation rate. Hall effect sensors collect DC bus current and voltage signals, and the amplitude of the fluctuation component in the 0.1-10 Hz frequency range is calculated using an FFT transform. The system uses a sliding time window (default 30 seconds) to calculate the RMS value of the fluctuation rate to eliminate transient noise interference. In the engineering implementation, the sampling frequency of x is set to 1 kHz to ensure that microsecond-level fluctuations are captured.

[0035] The baseline threshold c was set at 5% based on extensive empirical research and reflects the safe operating limit of alkaline electrolyzers. When fluctuations exceed 5%, the electrolyzer membrane electrodes experience excessive mechanical stress, and experimental data shows that the lifespan decay rate increases by three times. The threshold was selected based on comprehensive considerations: electrolyzer manufacturer specifications; national power quality standards for the power grid; and data from 2000 hours of accelerated aging tests.

[0036] Adjustable parameters σ The value range of 0.1-0.3 is determined through parameter sensitivity analysis to control the steepness of the function curve. The specific adjustment rules are: 0.15-0.18 for alkaline electrolyzers (slower response slope); 0.25-0.28 for PEM electrolyzers (fast response requirements); temporarily adjusted to 0.1 for supercapacitor compensation (increased trigger sensitivity) Output : Three-level control threshold verified by experiments: When: immediately activate energy storage compensation (response time <50ms); Time: adjust the number of electrolytic cells in operation (adjustment cycle 2 minutes); Time: Only for data recording (stored in CSV format).

[0037] It is worth mentioning that the Gaussian membership function proposed in the above embodiment innovatively quantifies the power fluctuation rate into three response intervals and introduces an adjustable parameter Achieve dynamic adjustment of control sensitivity. Determine through experiments The best value rule is: 0.15-0.18 for alkaline electrolyzer (relatively gentle response), 0.25-0.28 for PEM electrolyzer (relatively sensitive response). The system calculates the real-time membership value every 5 seconds. When there are 3 consecutive sampling cycles, The confirmation mechanism of energy storage compensation is triggered in time to avoid false operation.

[0038] The technical effects of the above embodiments include: It can reduce the false operation rate and improve the accuracy of fluctuation recognition, and is adaptive The adjustment can increase the efficiency of PEM electrolyzers by 5.2 percentage points.

[0039] For example, traditional technical solutions have the following technical problems: fixed PID parameters are difficult to adapt to the strong randomness of wind and solar power, the coupling relationship between electrolyzer temperature changes and power distribution is complex, and high-frequency fluctuations cause conventional control to oscillate.

[0040] Based on this, the power distribution ratio of the electrolyzer group is calculated by the following formula: ; in is the power deviation, 、 、 It is a fuzzy self-tuning parameter. Its adjustment rules include: when SOC>80%, increase value; decreases when the temperature gradient is >3℃ / min value; when the fluctuation frequency is greater than 0.1Hz, it increases value.

[0041] The fuzzy self-tuning PID formula realizes the dynamic optimization of power distribution: Deviation : Obtained through high-precision differential calculation (resolution 0.1%), including: the instantaneous deviation between the current power and the set value; the cumulative deviation integral over the past 10 minutes; and the trend forecast value for the next 5 seconds (based on the ARIMA model).

[0042] Proportional term : Basic adjustment parameters, the adjustment rules are: ; When SOC>80%, the value increases exponentially, with a maximum limit of 1.2. Actual data shows that this strategy can extend the battery cycle life by 40%.

[0043] Integral Item : The key parameter to eliminate steady-state error, strictly constrained by temperature gradient: ; When the temperature gradient is >3℃ / min, The value automatically decays to 30% of the reference value to prevent overshoot caused by integral saturation.

[0044] differential term :To suppress the core parameters of high-frequency fluctuations, variable structure control is adopted: ; When the fluctuation frequency The differential effect is quickly enhanced to control the overshoot within 2.5%.

[0045] It is worth mentioning that this embodiment combines traditional PID control with fuzzy logic and creatively proposes parameter self-tuning rules: The value increases exponentially with the increase of SOC (fitting coefficient 0.32), The value decays as the temperature gradient increases according to a piecewise function. When the fluctuation frequency exceeds 0.1Hz, the value is adjusted using a hyperbolic function. The best parameter combination measured experimentally is: , , .

[0046] The technical effects of the above embodiments include: It can reduce power tracking error, control overshoot within 2.5%, reduce temperature fluctuation and extend equipment life.

[0047] For example, traditional technical solutions use single-objective optimization, which leads to the deterioration of other performance indicators. Static weights cannot adapt to changes in operating conditions, and there is a conflict between economic and reliability goals.

[0048] Based on this, the optimization objective function of the advanced module is: ; in: is the weight coefficient , is the unsuppressed power, is the electrolytic cell temperature gradient, is the life loss cost; the weight coefficient is updated every 5 minutes through the fuzzy inference engine, and the update rule takes into account the electricity price period and the remaining capacity of the energy storage.

[0049] Should The frontier optimization function achieves a dynamic balance of multi-dimensional objectives: Fluctuation term Standardization ensures comparability across systems of varying capacities. Calculation uses the RMS value within a sliding time window, with the window width adaptively adjusted based on fluctuation characteristics (10-300 seconds). Technical validation has shown a strong correlation between this metric and electrolyzer mechanical stress (correlation coefficient 0.87).

[0050] Temperature gradient term : 85°C was chosen based on the material's phase transition point. Finite element simulation was used to establish a three-dimensional temperature field model, using the maximum gradient value at 12 temperature measurement points as input. Experimental data showed that protective load reduction would be triggered when this value exceeded 0.6.

[0051] Life loss item : The cost model includes: ; in is the number of starts and stops of the electrolyzer / battery / capacitor, is the single loss cost, is the depth coefficient.

[0052] Dynamic Weight The update algorithm integrates: electricity price signals (time-of-use electricity price coefficient 0.8-1.5); equipment health (remaining life prediction value) and grid dispatch requirements (AGC instruction priority).

[0053] The weight adjustment uses a fuzzy inference engine and is updated every 5 minutes to ensure that the solution set is within the 10% interval of the Pareto frontier.

[0054] It's worth noting that the three-objective optimization function constructed in this example uses a dynamic weight allocation strategy, with weight coefficients updated every five minutes. The update algorithm incorporates information from three dimensions: electricity price signals (time-of-use price data), equipment health (remaining life predictions), and grid dispatch requirements (AGC instructions). By introducing a Pareto frontier solution screening mechanism, each optimization result is ensured to be within the top 10% of the optimal solution set.

[0055] The technical benefits of this solution include: a 28% improvement in overall energy efficiency compared to single-target optimization; a reduction in the cost of hydrogen production per kilowatt-hour by 0.15 yuan / kWh; and a rise in the grid dispatch command response qualification rate from 92% to 99.1%.

[0056] For example, traditional technical solutions have the following technical problems: low accuracy in tracing the source of fluctuations, reliance on experience in the selection of control modes, and serious lags in effect evaluation.

[0057] A method, applied to a microgrid hydrogen production system based on fuzzy logic control as described in any one of the above, comprises the following steps: Real-time collection of wind and solar power output curves, bus voltage harmonic content, and data from various temperature measurement points of electrolyzers; Calculate the power fluctuation level and mark the characteristics of the fluctuation source through three-level fuzzy reasoning; Select electrolyzer power redistribution, energy storage compensation, or generation unit limiting mode based on fluctuation level and source characteristics; Execute power allocation instructions and monitor the current balance between electrolytic cell groups in real time; evaluate the stabilization effect every 30 minutes and update the correction coefficient and weight distribution of the fuzzy rule base.

[0058] It is worth mentioning that the innovative points of the five-step control process proposed in this embodiment are: adding harmonic analysis in the data acquisition stage; using convolutional neural networks to label the characteristics of fluctuation sources; the current balance monitoring accuracy reaches ±0.5%, and the Markov decision process model is introduced for effect evaluation.

[0059] For example, traditional technical solutions have the following technical problems: The expansion of the rule base leads to decreased reasoning efficiency, poor adaptability to new operating conditions, and a high rule conflict rate.

[0060] Based on this, this embodiment is designed to delete redundant rules that have been triggered less than 3 times in the past 2 hours and have a confidence level lower than 0.6; merge adjacent rules with a similarity greater than 85% and a control effect difference less than 5%; add emergency rules based on the analysis of the last 10 large fluctuation events and mark the event feature labels; adjust the center point position and widening parameters of the membership function to increase the probability that the newly collected data falls within the valid membership interval to more than 90%.

[0061] It is worth mentioning that the dynamic evolution mechanism of the rule base designed in this embodiment includes: Redundant rules were identified using the Jaccard similarity algorithm, emergency rule generation was based on Apriori association rule mining, and membership function adjustment was performed using the gradient descent method, with a confidence threshold of 0.6 and an effect difference threshold of 5%.

[0062] For example, traditional technical solutions have the following technical problems: The protection action speed and selectivity are inconsistent, the fault diagnosis accuracy is insufficient and the system recovery time is long.

[0063] Based on this, the hierarchical protection protocol is activated when the following abnormal conditions are detected: the voltage difference between electrolytic cell groups exceeds 10% of the rated value for 30 seconds and is accompanied by an abnormal temperature rise; the energy storage SOC suddenly increases by more than 15% within 5 minutes and causes power distribution oscillation; the fuzzy controller outputs conflicting instructions three times in a row and self-checks without hardware faults; the execution order of the protection protocol is to first cut off the power generation unit with the largest fluctuation and start the backup power supply, then switch to the preset fixed parameter control mode, and finally trigger the sound and light alarm and upload the fault diagnosis report to the monitoring center.

[0064] It is worth mentioning that the hierarchical protection system constructed in this embodiment has the following features: Anomaly detection uses a multi-scale sliding window (1s / 10s / 60s); fault diagnosis integrates an expert system (including 327 diagnostic rules); backup power supply switching time is <200ms, and fault report generation uses natural language processing technology.

[0065] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A microgrid hydrogen production system based on fuzzy logic control, characterized by: The system comprises a renewable energy generation unit, an electrolyzer array, multiple types of energy storage devices, a power distributor and a fuzzy logic controller; the renewable energy generation unit is connected to the electrolyzer array and the power distributor via a DC bus, and the output ends of the power distributor are connected to a battery pack and a supercapacitor pack respectively; the fuzzy logic controller collects bus voltage fluctuation rate, electrolyzer temperature gradient and energy storage charge state as input variables in real time, and outputs multi-objective optimization instructions to the power distributor; the system dynamically adjusts the start-stop combination of the electrolyzer array and the energy storage charging and discharging priority.

2. A microgrid hydrogen production system based on fuzzy logic control according to claim 1, characterized in that: The power distributor comprises three levels of power channels: the first channel is connected to the alkaline electrolyzer group and uses a thyristor voltage regulator module to achieve coarse power adjustment; the second channel is connected to the PEM electrolyzer group and is equipped with an IGBT chopper circuit to achieve fine power adjustment; the third channel is connected to the hybrid energy storage device and achieves rapid power compensation through a bidirectional DC-DC converter; the coordinated control of the three-level channels adopts a priority dynamic rotation mechanism, which automatically increases the response weight of the energy storage channel when a continuous power shortage is detected, and resets the initial distribution ratio after power is restored.

3. A microgrid hydrogen production system based on fuzzy logic control according to claim 1, characterized in that The fuzzy logic controller includes three layers of decision-making modules: the primary module uses interval type-2 fuzzy sets to process the time-domain fluctuation characteristics of the bus voltage, and has a built-in sliding time window analysis algorithm to identify fluctuation patterns; the intermediate module uses the TSK fuzzy model to calculate the safe operating range of the electrolyzer, and integrates temperature field finite element simulation data to establish a three-dimensional mapping relationship; the advanced module integrates the outputs of the primary and intermediate modules, and generates multi-objective optimization instructions that take into account economy, equipment life, and grid scheduling requirements through a multi-attribute decision-making algorithm.

4. A microgrid hydrogen production system based on fuzzy logic control according to claim 1, characterized in that: The start-stop control of the electrolytic cell array meets the following constraints: the number of starts and stops per day does not exceed three, and the interval between adjacent starts and stops is greater than four hours; the power difference between adjacent electrolytic cell groups does not exceed 15% of the rated value, and the gradient change rate is less than 5% / minute; preheating to above 80°C and maintaining it for 30 minutes must be completed before cold start; the system uses historical operating data to train a neural network to predict the optimal start-stop timing, and embeds the prediction results in the fuzzy rule base as constraint correction factors.

5. A microgrid hydrogen production system based on fuzzy logic control according to claim 1, characterized in that: The following fuzzy membership function is used to calculate the severity of power fluctuation: ； in is the measured power fluctuation rate, is the benchmark volatility threshold, which is 5%. It is an adjustable parameter with a value range of 0.1-0.3; when When the energy storage compensation is triggered, Adjust the electrolytic cell power when Only the fluctuation data needs to be recorded.

6. A microgrid hydrogen production system based on fuzzy logic control according to claim 1, characterized in that: The power distribution ratio of the electrolyzer group is calculated by the following formula: ； in is the power deviation, 、 、 It is a fuzzy self-tuning parameter. Its adjustment rules include: when SOC>80%, increase value; decreases when the temperature gradient is >3℃ / min value; when the fluctuation frequency is greater than 0.1Hz, it increases value.

7. A microgrid hydrogen production system based on fuzzy logic control according to claim 1, characterized in that: The optimization objective function of the high-level module is: ； in: is the weight coefficient , is the unsuppressed power, is the electrolytic cell temperature gradient, is the life loss cost; the weight coefficient is updated every 5 minutes through the fuzzy inference engine, and the update rule takes into account the electricity price period and the remaining capacity of the energy storage.

8. A method, applied to a microgrid hydrogen production system based on fuzzy logic control as described in any one of claims 1 to 7, characterized in that: The following steps are involved: Real-time collection of wind and solar power output curves, bus voltage harmonic content, and data from various temperature measurement points of electrolyzers; Calculate the power fluctuation level and mark the characteristics of the fluctuation source through three-level fuzzy reasoning; Select electrolyzer power redistribution, energy storage compensation, or generation unit limiting mode based on fluctuation level and source characteristics; Execute power allocation instructions and monitor the current balance between electrolytic cell groups in real time; evaluate the stabilization effect every 30 minutes and update the correction coefficient and weight distribution of the fuzzy rule base.

9. A method according to claim 8, characterized in that: The rule update operations include: deleting redundant rules that have been triggered less than 3 times in the past 2 hours and have a confidence level lower than 0.6; merging adjacent rules with a similarity greater than 85% and a control effect difference less than 5%; adding emergency rules based on the analysis of the last 10 large fluctuation events and marking the event feature labels; adjusting the center point position and widening parameters of the membership function to increase the probability of newly collected data falling within the valid membership interval to more than 90%.

10. A method according to claim 1, characterized in that The hierarchical protection protocol is activated when the following abnormal conditions are detected: the voltage difference between electrolyzer groups exceeds 10% of the rated value for 30 seconds and is accompanied by an abnormal temperature increase; the energy storage SOC suddenly increases by more than 15% within 5 minutes and causes power distribution oscillation; the fuzzy controller outputs conflicting instructions three times in a row and self-checks and finds no hardware faults; The execution order of the protection protocol is to first cut off the power generation unit with the largest fluctuation and start the backup power supply, then switch to the preset fixed parameter control mode, and finally trigger the sound and light alarm and upload the fault diagnosis report to the monitoring center.

Images