Alkaline water electrolysis hydrogen production system optimization method based on multi-parameter cooperative control

Abstract

The invention relates to an optimization method of an alkaline water electrolysis hydrogen production system based on multi-parameter cooperative control. The method comprises the following steps: acquiring real-time power and historical output data of renewable energy sources, environmental parameters and operation data of an electrolytic cell, and preprocessing the real-time power and historical output data, environmental parameters and operation data of the electrolytic cell; constructing a fusion renewable energy fluctuation prediction model, and inputting historical output data, environmental parameters and electrolytic cell operation data into the model to predict renewable energy fluctuation characteristics; establishing a wind-light output prediction model, and determining an optimal current density interval based on the real-time power of the renewable energy and the wind-light output prediction model; correcting the real-time electrolyte concentration by adopting a conductivity-pH dual channel, constructing a concentration-current two-dimensional compensation model, and inputting the corrected real-time electrolyte concentration into the model to obtain a target electrolyte concentration; and dynamically adjusting related parameters of the hydrogen production system, calculating a self-adaptive threshold value of the system, determining a grading response threshold value based on the self-adaptive threshold value, and controlling the hydrogen production system to operate. And high-efficiency hydrogen production and stable operation of the system are realized.

Description

Technical Field

[0001] The present application relates to the technical field of green hydrogen production, and particularly to an optimization method for an alkaline water electrolysis hydrogen production system based on multi-parameter collaborative control. Background Art

[0002] As a key equipment for green hydrogen production, the alkaline water electrolysis hydrogen production system has become a core technology in the fields of renewable energy consumption, industrial green hydrogen production, and energy storage peak shaving due to its high technical maturity, strong system stability, high product purity and other significant advantages. This system drives water to decompose into hydrogen and oxygen by applying a direct current electric field in an alkaline electrolyte, and its energy conversion efficiency and operation reliability have a decisive impact on the production cost of green hydrogen.

[0003] The global installed capacity of alkaline electrolyzers continues to grow, but the existing technologies still face some key bottlenecks. First, the problem of dynamic response hysteresis. Traditional control algorithms are difficult to adapt to the rapid fluctuations of renewable energy power generation, resulting in a lag in current density regulation and a significant decrease in hydrogen production efficiency. Especially under fluctuating operating conditions, the deviation of the current density from the optimal operating range will greatly increase the unit hydrogen energy consumption cost. Second, the problem of electrolyte transport imbalance. During the operation of the system, the loss of alkali solution will cause an imbalance in the electrolyte concentration, which will further exacerbate the kinetic mismatch between the hydrogen evolution and oxygen evolution reactions, and at the same time accelerate the corrosion rate of the electrode material, affecting the device life. Third, the lack of intelligent diagnosis ability. The existing early warning system relies on fixed threshold settings and cannot adapt to the dynamic operating conditions of the electrolyzer, resulting in a high false alarm rate of faults and making it difficult to achieve accurate early fault warning.

[0004] Therefore, in the related technologies, there is an urgent need for a method that can improve the energy conversion efficiency and stability of the alkaline water electrolysis hydrogen production system. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide an optimization method for an alkaline water electrolysis hydrogen production system based on multi-parameter collaborative control, which can improve the energy conversion efficiency and stability of the alkaline water electrolysis hydrogen production system.

[0006] In a first aspect, the present application provides an optimization method for an alkaline water electrolysis hydrogen production system based on multi-parameter collaborative control. The method includes:

[0007] Obtain the real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data, and perform preprocessing;

[0008] Construct a fusion renewable energy fluctuation prediction model, including random forest, LSTM neural network, and support vector machine, and input the historical output data, environmental parameters, and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics;

[0009] Build a wind-solar power output prediction model, and determine the optimal current density range based on the real-time power of the renewable energy and the wind-solar power output prediction model;

[0010] Adopt a dual-channel conductivity-pH correction to real-time electrolyte concentration, construct a two-dimensional concentration-current compensation model based on an adaptive temperature correction coefficient, input the corrected real-time electrolyte concentration into the model to obtain the target electrolyte concentration;

[0011] Dynamically adjust the relevant parameters of the hydrogen production system based on the renewable energy fluctuation characteristics, the optimal current density range and the target electrolyte concentration, calculate the system adaptive threshold, determine the hierarchical response threshold based on the adaptive threshold, and control the operation of the hydrogen production system.

[0012] Optionally, in an embodiment of the present application, the preprocessing includes:

[0013] Adopt the 3σ criterion to eliminate abnormal data, linearly interpolate and fill the missing data, and normalize all data.

[0014] Optionally, in an embodiment of the present application, the construction of the integrated renewable energy fluctuation prediction model, including random forest, LSTM neural network and support vector machine, input the historical power output data, environmental parameters, and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics, including:

[0015] Input the historical power output data, environmental parameters, and electrolyzer operation data into the initial random forest, perform random sampling, quantization and dimensionality reduction, and output 15-dimensional key features;

[0016] Input the 15-dimensional key features into the initial LSTM neural network and the initial support vector machine respectively, and output the predicted power fluctuation value;

[0017] Compare the predicted power fluctuation value with the real-time power, adjust the parameters of the random forest, LSTM neural network and support vector machine to obtain the integrated renewable energy fluctuation prediction model;

[0018] Input the historical power output data, real-time environmental parameters, and real-time electrolyzer operation data into the adjusted integrated renewable energy fluctuation prediction model, and output the renewable energy power prediction value.

[0019] Optionally, in an embodiment of the present application, the wind-solar power output prediction model is:

[0020] η(J,T,RH)=η0·e -αJ ·[1-β1(T-T ref )-β2(RH-RH ref )]

[0021] where J is the current density, T is the ambient temperature, RH is the relative humidity, RH ref = 50%, η0 = 85% is the reference efficiency, α = 0.12 cm 2 / A is the attenuation coefficient, β1 = 0.0023 °C -1 and β2 = 0.0023 are the temperature and humidity compensation coefficients.

[0022] Optionally, in an embodiment of the present application, the concentration-current two-dimensional compensation model is:

[0023] C target = C0 + k1·ΔJ + k2·ΔT

[0024] k2(t) = k 2,base ·[1 + α·sigmoid(β·ΔT)]

[0025] where C target is the target electrolyte concentration, C0 is the corrected real-time electrolyte concentration, k1 = 0.12 mol / (A·h) is the current density compensation coefficient, ΔJ is the current density deviation value, ΔT is the temperature deviation value, k2 is the temperature correction coefficient, k 2,base = 0.008 mol / °C is the base temperature correction coefficient, α is the sensitivity adjustment factor dynamically adjusted according to the electrolyte circulation rate, β is the temperature response gradient, and sigmoid(·) is the activation function that limits the coefficient fluctuation range.

[0026] Optionally, in an embodiment of the present application, the method further includes:

[0027] Using the PID algorithm to maintain the liquid level set value of the liquid level system.

[0028] Optionally, in an embodiment of the present application, the method further includes:

[0029] Performing multi-parameter fusion feature extraction based on the system operation parameters, and inputting the extracted principal component features into the random forest classification model for system fault location.

[0030] In a second aspect, the present application also provides an optimization device for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control. The device includes:

[0031] A data acquisition and preprocessing module, configured to obtain the real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data, and perform preprocessing;

[0032] A renewable energy fluctuation characteristic prediction module, which is used to construct a renewable energy fluctuation prediction model integrating a random forest, an LSTM neural network, and a support vector machine, and input the historical output data, environmental parameters, and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics;

[0033] An optimal current density interval determination module, which is used to establish a wind-solar output prediction model and determine the optimal current density interval based on the real-time power of the renewable energy and the wind-solar output prediction model;

[0034] A target electrolyte concentration determination module, which is used to correct the real-time electrolyte concentration through a conductivity-pH dual-channel, construct a concentration-current two-dimensional compensation model based on an adaptive temperature correction coefficient, and input the corrected real-time electrolyte concentration into the model to obtain the target electrolyte concentration;

[0035] A system optimization module, which is used to dynamically adjust the relevant parameters of the hydrogen production system based on the renewable energy fluctuation characteristics, the optimal current density interval, and the target electrolyte concentration, calculate the system adaptive threshold, determine the hierarchical response threshold based on the adaptive threshold, and control the operation of the hydrogen production system.

[0036] In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor executes the steps of the methods described in the above respective embodiments.

[0037] In a fourth aspect, the present application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the methods described in the above respective embodiments are implemented.

[0038] The above optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control first obtains the real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data, and performs preprocessing. Then, a fusion renewable energy fluctuation prediction model is constructed, including a random forest, an LSTM neural network, and a support vector machine, and the historical output data, environmental parameters, and electrolyzer operation data are input into the model to predict the renewable energy fluctuation characteristics. Then, a wind-solar output prediction model is established, and the optimal current density interval is determined based on the real-time power of the renewable energy and the wind-solar output prediction model. Then, the real-time electrolyte concentration is corrected through a conductivity-pH dual-channel, and a concentration-current two-dimensional compensation model is constructed based on an adaptive temperature correction coefficient, and the corrected real-time electrolyte concentration is input into the model to obtain the target electrolyte concentration. Finally, the relevant parameters of the hydrogen production system are dynamically adjusted based on the renewable energy fluctuation characteristics, the optimal current density interval, and the target electrolyte concentration, the system adaptive threshold is calculated, and the hierarchical response threshold is determined based on the adaptive threshold to control the operation of the hydrogen production system. That is to say, through dynamic current density regulation, closed-loop control of electrolyte concentration, and an intelligent system stability guarantee mechanism, efficient hydrogen production and stable operation of the system are achieved. Description of the Drawings

[0039] Figure 1 It is an application environment diagram of an optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control in an embodiment;

[0040] Figure 2 It is a flowchart of an optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control in an embodiment;

[0041] Figure 3 It is a flowchart of the dynamic current density optimization in an embodiment;

[0042] Figure 4 It is a flowchart of the electrolyte concentration closed-loop control strategy in an embodiment;

[0043] Figure 5 It is a flowchart of the intelligent early warning threshold adjustment hierarchical response mechanism in an embodiment;

[0044] Figure 6 It is a structural block diagram of an optimization device for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control in an embodiment;

[0045] Figure 7 It is an internal structure diagram of a computer device in an embodiment. Detailed Embodiments

[0046] To make the objectives, technical solutions and advantages of this application more clear and understandable, the following further elaborates on this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for explaining this application and are not used to limit this application.

[0047] An optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control provided by an embodiment of this application can be applied to an application environment as Figure 1 shown. Among them, the terminal communicates with the server through the network. The data storage system can store the data that the server needs to process. The data storage system can be integrated on the server, or placed in the cloud or other network servers. Among them, the terminal can be, but is not limited to, various personal computers, laptop computers, smartphones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0048] In one embodiment, as Figure 2 shown, an optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control is provided. Taking the server in Figure 1 as an example for illustration, it includes the following steps:

[0049] S201: Obtain the real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data, and perform preprocessing.

[0050] In an embodiment of this application, first, obtain the real-time power output and historical output data of renewable energy (wind energy, solar energy), environmental temperature (-20°C to 50°C), humidity (10% to 95% RH), wind speed, and electrolyzer operation parameters (voltage, current, temperature, etc.). Specifically, as Figure 3 shown, deploy a sensor group at key positions such as the electrolyzer body, gas-liquid separator, and circulation pipeline to collect parameters such as stack voltage / current, electrolyte flow rate (accuracy ±2%), plate temperature (±0.5°C), gas purity (99.9% level) in real time, and perform data cleaning preprocessing on the collected data.

[0051] Specifically, in an embodiment of this application, the preprocessing includes:

[0052] Adopt the 3σ criterion to eliminate abnormal data, perform linear interpolation filling on missing data, and perform normalization processing on all data.

[0053] In an embodiment of this application, the calculation process of eliminating outliers by the 3σ criterion is as follows:

[0054] Calculate the mean: For a set of collected data x1, x2, …, x n , its mean The calculation formula is: Calculate the standard deviation: The calculation formula for the standard deviation σ is: Determine the outlier range: According to the 3σ criterion, data falling outside is regarded as an outlier. That is, if or then is considered an outlier and can be removed.

[0055] The process of linearly interpolating and filling missing values is as follows:

[0056] Assume that in the data sequence, x i and x i+1 are two adjacent known data points, and there is a missing value x m between them, and the corresponding position is m (i < m < i + 1); Calculate the interpolation coefficient: According to the principle of linear interpolation, the calculation formula for the interpolation coefficient k is: Fill the missing value: Then the linear interpolation calculation formula for the missing value x m is: x m =(1 - k)x i +kx i+1 , that is, x m =(1 - (m - i))x i +(m - i)x i+1 .

[0057] At the same time, use the Min - Max method to unify the data to the [0, 1] interval, and the formula is:

[0058]

[0059] S203: Construct a fusion renewable energy fluctuation prediction model, including a random forest, an LSTM neural network, and a support vector machine, and input the historical output data, environmental parameters, and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics.

[0060] In the embodiment of this application, as Figure 3 shown, combine the random forest algorithm, the LSTM neural network, and the support vector machine (SVM) to predict the renewable energy fluctuation characteristics. The random forest algorithm can perform random sampling and analysis on the data, provide diverse data inputs for the LSTM neural network and the SVM, and enhance the generalization ability of the model. It specifically includes two parts: model training and model application.

[0061] Specifically, in an embodiment of the present application, the construction of the integrated renewable energy fluctuation prediction model, including random forest, LSTM neural network, and support vector machine, and inputting the historical output data, environmental parameters, and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics includes:

[0062] S301: Input the historical output data, environmental parameters, and electrolyzer operation data into the initial random forest for random sampling, quantization, and dimensionality reduction, and output 15-dimensional key features.

[0063] S303: Input the 15-dimensional key features into the initial LSTM neural network and the initial support vector machine respectively, and output the predicted power fluctuation values.

[0064] S305: Compare the predicted power fluctuation values with the real-time power, and adjust the parameters of the random forest, LSTM neural network, and support vector machine to obtain the integrated renewable energy fluctuation prediction model.

[0065] S307: Input the historical output data, real-time environmental parameters, and real-time electrolyzer operation data into the adjusted integrated renewable energy fluctuation prediction model, and output the predicted renewable energy power values.

[0066] In an embodiment of the present application, first, input the historical output data, environmental parameters, and electrolyzer operation data into the initial random forest for random sampling and analysis of the data, solve the high-dimensional data redundancy problem, and improve the input quality of the subsequent model. Input data: The original training data set X ∈ R n×m (n = number of samples, m = 32-dimensional features, including power, temperature, humidity, etc.); Gini coefficient calculation: where D is the current node data set, and C k is the number of samples in the k-th class, used to quantify the contribution of features to the data purity, and key parameters such as power volatility and temperature change are selected.

[0067] Decision tree generation parameters: Number of trees: 100 (balancing calculation efficiency and accuracy), Minimum number of samples in leaf nodes: 5 (preventing overfitting), Feature subset sampling rate: (m = 32 is the number of original features).

[0068] Output and downstream coordination: Remove redundant features with importance < 5%, generate a dimensionality-reduced data set, that is, a 15-dimensional key feature vector X ′ ∈ R n×15 , as the input of LSTM and SVM. Dynamic update mechanism: When the prediction error > 10%, add 10% decision trees and adjust the sampling rate to 0.8m.

[0069] After that, the LSTM time series prediction algorithm is used to capture the time dependence of the renewable energy output, and solve the long-term and short-term fluctuation correlation problem of the wind and light power. The network structure of LSTM includes a forget gate (controlling information discard), an input gate (updating the cell state), and an output gate (generating a prediction value). Output data: the power fluctuation value P in the next 15 minutes LSTM ∈[0,1] (after normalization);

[0070] Forget gate:

[0071] f t =σ(W f ·[h t-1 ,x t +b f )

[0072] Among them, W f : weight matrix (dimension 128×256), h t-1 : the previous hidden state, x t : the current input feature vector (15-dimensional), σ: sigmoid activation function, f t : the output result at the current time step t.

[0073] Input gate:

[0074] i t =σ(W i ·[h t-1 ,x t +b i )

[0075]

[0076] Among them, tanh: hyperbolic tangent activation function, Candidate cell state.

[0077] Cell state update: ⊙: represents element-wise multiplication.

[0078] Output gate:

[0079] o t =σ(W o ·[h t-1 ,x t +b o )

[0080] h t =o t ·tanh(C t )

[0081] Among them, o t is the output of the output gate, h tis the hidden state at the current moment, W o is the weight matrix of the output gate, b o is the bias of the output gate.

[0082] Weight matrix W f , W i , W C , W o It is trained by the Adam optimizer, and its calculation logic is as follows:

[0083] First, parameter initialization: (for each weight matrix) Let W ∈ {W f , W i , W C , W o} and initialize: first moment estimate (momentum term): Second moment estimate (gradient square term): Time step: t = 1, hyperparameters (default reference values) include: momentum decay rate: β 1＝0.9 , gradient square decay rate: β 2＝0.999 , learning rate: α = 0.001, numerical stability constant: ∈ = 10 -8 .

[0084] After that, gradient calculation: At time step t, calculate the gradient of the current batch through the backpropagation algorithm where: L: is the loss function. The dimension of the gradient matrix needs to be consistent with the weight matrix. For example, the gradient of W f is (n is the number of hidden units, d is the input dimension).

[0085] After that, moment estimate update: Update each weight matrix independently:

[0086]

[0087]

[0088] After that, perform bias correction: Correct the estimation bias in the initial stage:

[0089]

[0090] Weight update:

[0091]

[0092] In this step, the update amount of each weight matrix is jointly determined by the adaptive learning rate and .

[0093] At the same time, the SVM (Support Vector Machine) is used for non-linear correction, and the regression power correction value of the renewable energy power fluctuation is output. Specifically, the radial basis function (RBF) kernel function is adopted:

[0094] K(x i ,x j )=exp(-γ∥x i -x j ∥ 2 )(γ=0.01)

[0095] Among them, x i ,x j are the input sample features (samples in the 15-dimensional feature X output by the random forest), γ is the kernel function parameter, which controls the function scope.

[0096] The training is to optimize the objective through regression and minimize the loss function, and the loss function is expressed as:

[0097]

[0098] Constraint conditions:

[0099]

[0100] Among them, w is the weight vector in the high-dimensional feature space, which is automatically calculated, b is the bias term, which is dynamically adjusted and automatically calculated according to the optimization, C = 1.5 is the penalty factor, ∈ = 0.01 is the tolerance, ξ i 、 are slack variables, and being greater than 0 means that the sample is allowed to cross the classification interval or be misclassified.

[0101] After that, the time decay function is used to adjust the weights, dynamically allocate the weights, and calculate the final prediction value. The formula is:

[0102]

[0103] Among them, k = 0.1 is the decay coefficient, and T = 15 minutes is the prediction period.

[0104] It should be noted that when |P LSTM -P SVM |> 10%, incremental learning is triggered, 10 new decision trees are added to the random forest, and the feature subset sampling rate is adjusted from to 0.8m (m = 15 dimensions).

[0105] After the model training is completed, it is applied. Data cleaning is performed on the collected real-time wind and solar power output, environmental sensor data, and electrolyzer operation data. The time series data is segmented using the sliding window method (window size = 30 minutes, step size = 5 minutes), and outlier filtering and data standard normalization are carried out. Then, RF feature screening is performed, and the feature weights are updated every 5 minutes. The 30-minute historical time series is input into the LSTM for real-time prediction, and the predicted value of the power fluctuation in the next 15 minutes is output. The 15-dimensional feature X^ output by the random forest is input into the SVM for non-linear correction, and the predicted value of the regression is output. Finally, the final predicted value is calculated by combining the dynamic weights, and the predicted value including the time series is output.

[0106] S205: Establish a wind and solar power output prediction model, and determine the optimal current density range based on the real-time power of the renewable energy and the wind and solar power output prediction model.

[0107] In the embodiment of the present application, a wind and solar power output prediction model is established. Considering the characteristics of different types of electrolyzers and the influence of environmental factors (such as temperature and humidity) on the electrolyzer efficiency curve, the adaptive optimal current density range is obtained through dynamic programming. This range can be adjusted steplessly within the range of 0.3 - 0.8 A / cm 2

[0108] Specifically, in an embodiment of the present application, the wind and solar power output prediction model is:

[0109] η(J,T,RH) = η0·e -αJ ·[1 - β1(T - T ref ) - β2(RH - RH ref )]

[0110] where J is the current density, T is the environmental temperature, RH is the relative humidity, RH ref = 50%, η0 = 85% is the reference efficiency, α = 0.12 cm 2 / A is the attenuation coefficient, β1 = 0.0023 °C -1 and β2 = 0.0023 are the temperature and humidity compensation coefficients.

[0111] The objective function is:

[0112]

[0113] where P re (t) is the real-time power of the renewable energy, γ = 0.15 A 2 / (cm 4 ·min) is the current density change penalty factor.

[0114] The constraint conditions are:

[0115] ​S207: Correct the real-time electrolyte concentration through dual channels of conductivity-pH, construct a two-dimensional compensation model of concentration-current based on the adaptive temperature correction coefficient, input the corrected real-time electrolyte concentration into the model, and obtain the target electrolyte concentration.

[0116] In the embodiment of the present application, as Figure 4 shown, dual channels are adopted to online monitor the concentration of electrolyte KOH, that is, through the conductivity sensor and pH feedback, and at the same time, the monitoring of the impurity content in the electrolyte is increased to achieve accurate monitoring of the electrolyte concentration. And a two-dimensional compensation model of concentration-current is constructed to adjust the liquid replenishment rate in real time according to the monitored concentration, current, temperature and impurity content data (error < ±0.5 mol / L) to ensure that the electrolyte concentration is stable within a suitable range. Considering the indirect influence of temperature on concentration monitoring and reaction, an adaptive temperature correction coefficient is added to the model and dynamically adjusted according to different stages of system operation and environmental conditions to further improve the accuracy of the model.

[0117] C 监测 = w1·C σ + w2·C pH

[0118] Among them, C σ is the concentration converted from conductivity, C pH is the concentration converted from pH value, w1 and w2 are dynamic weights, w1 = 0.7 - 0.1·ΔT / 10, w2 = 1 - w1.

[0119]

[0120] σ is the measured value of conductivity, k cell is the electrode constant, Λ m is the molar conductivity,

[0121] Λ m = Λ m,25 ·[1 + α(T - 25)]

[0122] Λ m,25 is the reference molar conductivity at 25°C (248 S·cm 2 / mol for KOH), α is the temperature coefficient (empirical value 0.019 / °C), and T is the actual temperature (°C).

[0123]

[0124] ρ sol is the electrolyte concentration (g / L, collected in real time by a densitometer), M KOH is the molar mass (56.11 g / mol)

[0125]

[0126] Among them, [X i is the impurity ion concentration (such as Fe3+, Cl-), and γ i is the interference factor (Fe3+: 0.15 / mol, Cl-: 0.08 / mol).

[0127] In an embodiment of the present application, the concentration-current two-dimensional compensation model is as follows:

[0128] C target = C0 + k1·ΔJ + k2·ΔT

[0129] k2(t) = k 2,base ·[1 + α·sigmoid(β·ΔT)]

[0130] Among them, C target is the target electrolyte concentration, C0 is the corrected real-time electrolyte concentration, k1 = 0.12 mol / (A·h) is the current density compensation coefficient, ΔJ is the current density deviation value, ΔT is the temperature deviation value, k2 is the temperature correction coefficient, and k 2,base = 0.008 mol / °C is the basic temperature correction coefficient, α is the sensitivity adjustment factor dynamically adjusted according to the electrolyte circulation rate, β is the temperature response gradient, and sigmoid(·) is the activation function, which limits the coefficient fluctuation range.

[0131] In an embodiment of the present application, the method further includes:

[0132] Adopting the PID algorithm to maintain the liquid level set value of the liquid level system.

[0133] In an embodiment of the present application, as Figure 4 shown, the liquid level is maintained stable through the PID algorithm. The liquid level system can be abstracted as a dynamic balance model, and the control target is to maintain the liquid level set value by adjusting the opening of the liquid inlet / drain valve. The PID algorithm dynamically adjusts the output through the error feedback mechanism, and the formula is expressed as:

[0134]

[0135] Among them, e(t) = set liquid level - actual liquid level, K p , K i , K d are the proportional, integral, and differential coefficients respectively.

[0136] When it is detected that the input power drops by 20%, the multi-model fusion prediction module is started, and in combination with the adaptive optimal current density interval, a current density adjustment instruction is generated (such as from 0.55 A / cm 2 dropping to 0.45 A / cm 2) The adjustment rate ≤ 5 A / s. Synchronously match the rotation speed of the electrolyte circulation pump (for example, from 3000 rpm to 2400 rpm), and maintain the liquid level fluctuation < ±3 mm through the PID algorithm to avoid concentration imbalance caused by sudden changes in the caustic solution flow rate.

[0137] S209: Dynamically adjust the relevant parameters of the hydrogen production system based on the renewable energy fluctuation characteristics, the optimal current density range, and the target electrolyte concentration, calculate the system adaptive threshold, determine the hierarchical response threshold based on the adaptive threshold, and control the operation of the hydrogen production system.

[0138] In the embodiments of the present application, the relevant parameters of the hydrogen production system are dynamically adjusted based on the renewable energy fluctuation characteristics, the optimal current density range, and the target electrolyte concentration, and the system adaptive threshold is calculated. The formula is as follows:

[0139]

[0140] Wherein, T adjust (t) is the real-time threshold after dynamic adjustment (such as hydrogen purity, leakage amount, etc.), T base is the initial value (such as purity 99.5%, leakage amount 50 ml / h), λ is the adjustment factor (value range 0.2 - 0.8), obtained by training with historical fault data, S(t) is the current operating state parameter of the system (including temperature, degree of electrolyzer aging, etc.), S avg 、S max 、S min are the historical average values, maximum values, and minimum values of the parameters. And determine the hierarchical response threshold based on the adaptive threshold. When the corresponding warning is triggered, execute the corresponding processing procedures, such as nitrogen purging, electrode activation, power off, etc., to ensure the stable operation of the hydrogen production system. Specifically, as Figure 5 shown, Level I warning (abnormal gas purity): When the hydrogen purity < 99.5%, start the nitrogen purging program (flow rate 50 L / min, duration 30 seconds), and perform secondary purification through the adsorption tower (molecular sieve + activated carbon) to restore the purity to 99.99%. Dynamically adjust the warning threshold of hydrogen purity according to the operating time of the system, historical fault data, etc.

[0141] Level II warning (deterioration of electrode performance): If the inter-electrode voltage difference > 2 V, execute the electrode activation program: reverse current pulse (0.1 A / cm 2 , duration 5 minutes) to remove the surface passivation layer and restore the electrochemically active area. Similarly, the warning threshold of the inter-electrode voltage difference can be dynamically adjusted.

[0142] Level Ⅲ Early Warning (Alkali Solution Leakage): When the leakage rate > 50 ml / h, turn off the power supply of the electrolyzer and activate the emergency drainage valve. At the same time, activate the standby alkali solution storage tank (with a capacity of 200 L) for temporary liquid supply to prevent the system from shutting down. The early warning threshold of the leakage rate can also be dynamically adjusted according to the actual situation.

[0143] Level Ⅳ Early Warning (Model Deviation Early Warning): Regularly evaluate the two-dimensional compensation model of concentration-current. When the deviation between the model prediction value and the actual measurement value exceeds a certain threshold (such as 5%), start the model correction program, and use the latest operation data to optimize the model to ensure the accuracy of the model and the stable operation of the system. At the same time, this threshold can also be dynamically adjusted according to the system operation situation.

[0144] In one embodiment of the present application, the method further includes:

[0145] Extract multi-parameter fusion features based on system operation parameters, and input the extracted principal component features into a random forest classification model for system fault location.

[0146] In one embodiment of the present application, introduce a machine learning algorithm to comprehensively analyze various operation parameters of the system, such as the stack voltage / current, electrolyte flow rate, plate temperature, gas purity, etc., to achieve multi-modal fault diagnosis and more quickly and accurately locate faults. Specifically, first, use principal component analysis to extract multi-parameter fusion features, and select the first k principal components (cumulative contribution rate > 85%) as the diagnostic input. The formula is as follows:

[0147] Z = W T ·X + ∈

[0148] Where X is the original parameter matrix (n-dimensional data such as voltage / current, temperature, flow rate, etc.), W is the feature vector matrix, and ∈ is the residual term.

[0149] Use a random forest classification model for system fault location, and the fault type decision function is:

[0150] F(x) = mode{h1(x), h2(x),..., h m (x)}

[0151] Where h m (x) is the output of a single decision tree.

[0152] Feature importance weight calculation:

[0153]

[0154] Key faults such as abnormal voltage (weight > 0.3) and temperature mutation (weight > 0.25) can be identified.

[0155] In the above optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control, first, the real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data are obtained and preprocessed; then, a fusion renewable energy fluctuation prediction model, including random forest, LSTM neural network, and support vector machine, is constructed, and the historical output data, environmental parameters, and electrolyzer operation data are input into the model to predict the renewable energy fluctuation characteristics; then, a wind-solar output prediction model is established, and the optimal current density range is determined based on the real-time power of the renewable energy and the wind-solar output prediction model; then, the real-time electrolyte concentration is corrected through a conductivity-pH dual-channel, and a concentration-current two-dimensional compensation model is constructed based on an adaptive temperature correction coefficient, and the corrected real-time electrolyte concentration is input into the model to obtain the target electrolyte concentration; finally, the relevant parameters of the hydrogen production system are dynamically adjusted based on the renewable energy fluctuation characteristics, the optimal current density range, and the target electrolyte concentration, the system adaptive threshold is calculated, and the hierarchical response threshold is determined based on the adaptive threshold to control the operation of the hydrogen production system. That is to say, through dynamic current density regulation, closed-loop control of electrolyte concentration, and an intelligent system stability guarantee mechanism, efficient hydrogen production and stable operation of the system are achieved.

[0156] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily have to be executed at the same time, but can be executed at different times. The execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or in turn with at least some of the steps or stages in other steps or other steps.

[0157] Based on the same inventive concept, the embodiments of the present application also provide an optimization device for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control for implementing the above-mentioned optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the optimization device for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control provided below can refer to the limitations on the optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control in the above text, and will not be repeated here.

[0158] In one embodiment, as Figure 6As shown in the figure, an optimization device 600 for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control is provided, including: a data acquisition and preprocessing module 601, a renewable energy fluctuation characteristic prediction module 603, an optimal current density range determination module 605, a target electrolyte concentration determination module 607, and a system optimization module 609, where:

[0159] The data acquisition and preprocessing module 601 is used to obtain the real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data, and perform preprocessing.

[0160] The renewable energy fluctuation characteristic prediction module 603 is used to construct a fusion renewable energy fluctuation prediction model, including a random forest, an LSTM neural network, and a support vector machine, and input the historical output data, environmental parameters, and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics.

[0161] The optimal current density range determination module 605 is used to establish a wind and light output prediction model, and determine the optimal current density range based on the real-time power of the renewable energy and the wind and light output prediction model.

[0162] The target electrolyte concentration determination module 607 is used to correct the real-time electrolyte concentration through a conductivity-pH dual-channel, construct a concentration-current two-dimensional compensation model based on an adaptive temperature correction coefficient, and input the corrected real-time electrolyte concentration into the model to obtain the target electrolyte concentration.

[0163] The system optimization module 609 is used to dynamically adjust the relevant parameters of the hydrogen production system based on the renewable energy fluctuation characteristics, the optimal current density range, and the target electrolyte concentration, calculate the system adaptive threshold, determine the hierarchical response threshold based on the adaptive threshold, and control the operation of the hydrogen production system.

[0164] In an embodiment of the present application, the data acquisition and preprocessing module is further used for:

[0165] Using the 3σ criterion to eliminate abnormal data, linearly interpolating and filling missing data, and normalizing all data.

[0166] In an embodiment of the present application, the renewable energy fluctuation characteristic prediction module is further used for:

[0167] Input the historical output data, environmental parameters, and electrolyzer operation data into the initial random forest, perform random sampling, quantization, and dimensionality reduction, and output 15-dimensional key features;

[0168] Input the 15-dimensional key features into the initial LSTM neural network and the initial support vector machine respectively, and output the predicted power fluctuation value;

[0169] Compare the predicted power fluctuation value with the real-time power, adjust the parameters of the random forest, LSTM neural network, and support vector machine to obtain a fused renewable energy fluctuation prediction model;

[0170] Input the historical output data, real-time environmental parameters, and real-time electrolyzer operation data into the adjusted fused renewable energy fluctuation prediction model to output the renewable energy power prediction value.

[0171] In an embodiment of the present application, the wind and light output prediction model is:

[0172] η(J, T, RH) = η0·e -αJ ·[1 - β1(T - T ref ) - β2(RH - RH ref )]

[0173] where J is the current density, T is the environmental temperature, RH is the relative humidity, RH ref = 50%, η0 = 85% is the reference efficiency, α = 0.12 cm 2 / A is the attenuation coefficient, β1 = 0.0023 °C -1 and β2 = 0.0023 are the temperature and humidity compensation coefficients.

[0174] In an embodiment of the present application, the concentration-current two-dimensional compensation model is:

[0175] C target = C0 + k1·ΔJ + k2·ΔT

[0176] k2(t) = k 2,base ·[1 + α·sigmoid(β·ΔT)]

[0177] where C target is the target electrolyte concentration, C0 is the corrected real-time electrolyte concentration, k1 = 0.12 mol / (A·h) is the current density compensation coefficient, ΔJ is the current density deviation value, ΔT is the temperature deviation value, k2 is the temperature correction coefficient, k 2,base = 0.008 mol / °C is the basic temperature correction coefficient, α is the sensitivity adjustment factor dynamically adjusted according to the electrolyte circulation rate, β is the temperature response gradient, and sigmoid(·) is the activation function to limit the coefficient fluctuation range.

[0178] In an embodiment of the present application, the method further includes:

[0179] Adopt the PID algorithm to maintain the liquid level set value of the liquid level system.

[0180] In an embodiment of the present application, the method further includes:

[0181] Multi-parameter fusion feature extraction is performed based on system operation parameters, and the extracted principal component features are input into a random forest classification model for system fault location.

[0182] Each module in the above-mentioned optimization device for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0183] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 7 shown. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements an optimization method for an alkaline electrolytic water hydrogen production system based on multi-parameter collaborative control. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0184] Those skilled in the art can understand that Figure 7 the structure shown in

[0185] is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0186] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it implements the steps in the above-mentioned method embodiments.

[0187] In one embodiment, a computer program product is provided, including a computer program which, when executed by a processor, implements the steps in the above method embodiments.

[0188] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties.

[0189] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0190] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0191] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several deformations and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. An optimization method for an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control, characterized in that, The method comprises: Obtain real-time power and historical output data of renewable energy, environmental parameters, and electrolyzer operation data, and perform preprocessing; Constructing a fusion renewable energy fluctuation prediction model, including random forest, LSTM neural network and support vector machine, and inputting the historical output data, environmental parameters and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics; Establishing a wind and solar power output prediction model, and determining an optimal current density range based on the real-time power of the renewable energy and the wind and solar power output prediction model; The conductivity-pH dual channel is used to correct the real-time electrolyte concentration, and a concentration-current two-dimensional compensation model is constructed based on the adaptive temperature correction coefficient. The corrected real-time electrolyte concentration is input into the model to obtain the target electrolyte concentration. Based on the renewable energy fluctuation characteristics, the optimal current density range and the target electrolyte concentration, the relevant parameters of the hydrogen production system are dynamically adjusted, the system adaptive threshold is calculated, and the graded response threshold is determined based on the adaptive threshold to control the operation of the hydrogen production system.

2. The optimization method of an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control according to claim 1, wherein The pre-processing comprises: The 3σ criterion was used to eliminate abnormal data, linear interpolation was used to fill missing data, and all data were normalized.

3. The optimization method of an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control according to claim 1, characterized in that, The construction of the integrated renewable energy fluctuation prediction model includes random forest, LSTM neural network and support vector machine, and inputting the historical output data, environmental parameters and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics includes: Input the historical output data, environmental parameters, and electrolyzer operation data into the initial random forest, perform random sampling, quantization, and dimensionality reduction, and output 15-dimensional key features; The 15-dimensional key features are respectively input into the initial LSTM neural network and the initial support vector machine, and the predicted power fluctuation value is output; Comparing the predicted power fluctuation value with the real-time power, adjusting the parameters of the random forest, LSTM neural network and support vector machine, and obtaining a fusion renewable energy fluctuation prediction model; The historical output data, real-time environmental parameters, and real-time electrolyzer operation data are input into the adjusted fusion renewable energy fluctuation prediction model to output the renewable energy power prediction value.

4. A method for optimizing an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control according to claim 1, characterized in that The wind and solar power output prediction model is: η(J, T, RH) = η0·e -αJ ·[1 - β1(T - T ref ) - β2(RH - RH ref )] where J is the current density, T is the ambient temperature, RH is the relative humidity, RH ref = 50%, η0 = 85% is the reference efficiency, α = 0.12 cm 2 / A is the attenuation coefficient, β1 = 0.0023 °C -1 and β2 = 0.0023 are the temperature and humidity compensation coefficients.

5. A method for optimizing an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control according to claim 1, characterized in that The concentration-current two-dimensional compensation model is: C target = C0 + k1·ΔJ + k2·ΔT k2(t) = k 2,base ·[1 + α·sigmoid(β·ΔT)] Among them, C target is the target electrolyte concentration, C0 is the corrected real-time electrolyte concentration, k1 = 0.12 mol / (A·h) is the current density compensation coefficient, ΔJ is the current density deviation value, ΔT is the temperature deviation value, k2 is the temperature correction coefficient, k 2,base = 0.008 mol / ℃ is the base temperature correction coefficient, α is the sensitivity adjustment factor dynamically adjusted according to the electrolyte circulation rate, β is the temperature response gradient, sigmoid(·) is the activation function, which limits the coefficient fluctuation range.

6. The optimization method of an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control according to claim 1, characterized in that The method further comprises: The PID algorithm is used to maintain the liquid level set point of the liquid level system.

7. A method for optimizing an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control according to claim 1, characterized in that, The method further comprises: Multi-parameter fusion feature extraction is performed based on system operating parameters, and the extracted principal component features are input into the random forest classification model to locate system faults.

8. An optimization device for an alkaline electrolyzed water hydrogen production system based on multi-parameter collaborative control, characterized in that, The device comprises: Data acquisition and preprocessing module, used to obtain real-time power and historical output data of renewable energy, environmental parameters, electrolyzer operation data, and perform preprocessing; A renewable energy fluctuation characteristic prediction module is used to construct a fusion renewable energy fluctuation prediction model, including a random forest, an LSTM neural network and a support vector machine, and input the historical output data, environmental parameters and electrolyzer operation data into the model to predict the renewable energy fluctuation characteristics; An optimal current density interval determination module is used to establish a wind and solar power output prediction model, and determine the optimal current density interval based on the real-time power of the renewable energy and the wind and solar power output prediction model; The target electrolyte concentration determination module is used to correct the real-time electrolyte concentration through the conductivity-pH dual-channel, construct a two-dimensional concentration-current compensation model based on the adaptive temperature correction coefficient, input the corrected real-time electrolyte concentration into the model, and obtain the target electrolyte concentration; The system optimization module is used to dynamically adjust the relevant parameters of the hydrogen production system based on the renewable energy fluctuation characteristics, the optimal current density range, and the target electrolyte concentration, calculate the system adaptive threshold, determine the hierarchical response threshold based on the adaptive threshold, and control the operation of the hydrogen production system.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images