PEM electrolysis water hydrogen production multi-tank hybrid optimization method

Abstract

The present application relates to a kind of PEM electrolytic water hydrogen production multi-tank hybrid optimization method, comprising the following steps: S1, the data of photovoltaic power generation is obtained, and data preprocessing is carried out;S2, using K-means clustering algorithm, the daytime data of photovoltaic power generation is divided into sunny day and non-sunny day according to weather mode, and the average power of different weather is calculated;S3, scheduling strategy is constructed;S4, using improved particle swarm optimization PSO algorithm for global search and optimization, find the optimal PEM scheduling strategy;S5, record the running time of each PEM electrolytic tank and state switching frequency, and evaluate the generated PEM scheduling strategy;The present application combines the influence of weather on photovoltaic power generation with PEM electrolytic tank allocation scheduling by K-means clustering algorithm, designs optimal coupling control strategy, multi-objective optimization and running time management, significantly reduces the switching frequency of PEM electrolytic tank, reduces equipment wear and maintenance cost, improves the efficiency of PEM electrolytic water hydrogen production.

Description

Technical Field

[0001] This invention relates to the field of renewable energy power generation and hydrogen production technology, and in particular to an optimized method for multi-tank hybrid operation of PEM electrolysis for hydrogen production. Background Technology

[0002] With the continued growth of global energy demand and the fact that fossil fuels have exacerbated environmental pollution, the development and utilization of renewable energy has gradually become an important issue. Compared with traditional fossil fuels, renewable energy is more environmentally friendly, does not produce greenhouse gas emissions, and helps to mitigate climate change. For example, solar energy has become the fastest-growing renewable energy source. Renewable energy can replace traditional fossil fuel power generation methods by building corresponding power plants, such as photovoltaic power plants, wind power plants, and hydropower plants, thereby achieving widespread application and sustainable development.

[0003] However, the intermittency and instability of renewable energy are the biggest reasons limiting its widespread application. Photovoltaic power generation is intermittent and volatile, and how to efficiently utilize photovoltaic power for hydrogen production has become a current research hotspot. PEM water electrolysis technology, as a way of energy conversion, can convert renewable energy into hydrogen energy, realizing energy storage and conversion. In addition, PEM water electrolysis technology has the advantages of high efficiency, environmental protection and easy control, and has broad application prospects in the field of energy storage and conversion. The development of PEM water electrolysis technology will help promote the optimization and upgrading of the energy structure, reduce dependence on fossil fuels, and achieve sustainable development.

[0004] The optimized scheduling of PEM electrolysis for hydrogen production can be effectively combined with photovoltaic power generation, converting the energy lost from curtailed solar power into hydrogen storage. The goals of optimized scheduling typically include maximizing hydrogen production, increasing power absorption rate, and reducing equipment start-ups and shutdowns to extend the lifespan of the PEM electrolyzer and reduce maintenance costs. Currently, the technologies for optimized scheduling of PEM electrolyzers mainly focus on the following aspects:

[0005] (1) Rule-based scheduling strategy usually adopts preset start-up and shutdown rules, that is, simply setting thresholds manually without considering the variability of data, and determining the start-up and shutdown of PEM electrolyzers based on whether the real-time power or predicted power of photovoltaic power generation reaches or exceeds the preset threshold.

[0006] (2) Prediction-based optimization scheduling: Using the predicted data of photovoltaic power generation, optimization algorithms (such as linear programming, dynamic programming, etc.) are used to formulate the scheduling plan of PEM electrolyzers. By predicting future photovoltaic power generation and demand, the hydrogen production plan is optimized to maximize output or minimize cost.

[0007] (3) Based on the scheduling strategy of optimization algorithm, various optimization algorithms are used to optimize the start-up and shutdown strategy of PEM electrolyzer to obtain multi-objective optimization results. For example, the particle swarm optimization algorithm (PSO) is used to find the combination of the maximum power point of photovoltaic power generation and the number of PEMs to maximize hydrogen production while minimizing the number of switching.

[0008] Although current technologies have made some progress in the scheduling optimization of PEM electrolyzers, there are still shortcomings and deficiencies. Existing scheduling optimization methods focus on maximizing hydrogen production or minimizing the number of switching operations, and cannot set other objectives for optimization. Prediction-based optimization methods rely too much on the accuracy of the prediction model, and the effectiveness of the scheduling strategy will be affected if the prediction error is large. At the same time, there is a lack of scheduling strategies that optimize for the number of switching operations, making it difficult for the system to achieve a balance between high switching frequency and high hydrogen production. Summary of the Invention

[0009] In view of this, the present invention provides a multi-cell hybrid optimization method for PEM electrolysis to produce hydrogen, which combines the impact of weather on photovoltaic power generation with the allocation and scheduling of PEM electrolyzers, designs an optimal coupling control strategy, and improves the efficiency of PEM electrolysis to produce hydrogen.

[0010] To achieve the above objectives, a multi-tank hybrid optimization method for PEM water electrolysis hydrogen production is characterized by the following steps:

[0011] S1. Acquire photovoltaic power generation data, separate the data into daytime data and nighttime data, and perform data preprocessing;

[0012] S2. Use the K-means clustering algorithm to divide the daytime photovoltaic power generation data into sunny days and non-sunny days according to the weather pattern, and calculate the average power under different weather conditions;

[0013] S3. Construct scheduling strategy;

[0014] S301. Set the parameters of the PEM electrolyzer, including the rated power of each PEM electrolyzer and the total number of PEM electrolyzers in the system.

[0015] S302, Calculate the power threshold and the number of activated PEM electrolyzers;

[0016] Based on the average photovoltaic power of each weather pattern and the safety factor, the applicable power threshold and the number of PEM electrolyzers that can be activated for each weather pattern are calculated.

[0017] S4. Use the improved Particle Swarm Optimization (PSO) algorithm for global search and optimization to find the optimal PEM scheduling strategy.

[0018] S401. Set a multi-objective optimization function, where the multi-objectives include maximizing hydrogen production, improving power absorption rate, and minimizing the number of switching operations.

[0019] S402. Initialize the particle swarm optimization algorithm parameters, including initializing the number of particles, number of iterations, and inertia weight, introducing mutation operation, dynamically adjusting parameters, and adding a local search mechanism;

[0020] S403. By iteratively updating the velocity and position of the particles, find the optimal PEM electrolyzer scheduling strategy, including the start-stop threshold and the PEM electrolyzer standby duration.

[0021] S5. Record the running time and number of state transitions for each PEM electrolyzer, and evaluate the generated PEM scheduling strategy.

[0022] The generated PEM scheduling strategy was evaluated by calculating hydrogen production and electricity absorption rate.

[0023] Preferably, data preprocessing includes extracting temperature, humidity, and solar radiation characteristics from the acquired daytime photovoltaic power generation data and performing normalization processing.

[0024] Preferably, the multi-objective optimization function is a weighted sum of hydrogen production, number of switching operations, and power absorption rate, and the weights of hydrogen production, number of switching operations, and power absorption rate are preset according to requirements.

[0025] Preferably, the local search mechanism involves performing a small-scale search and adjustment near the generated PEM electrolyzer scheduling strategy to obtain the optimal PEM electrolyzer scheduling strategy.

[0026] Compared with the prior art, the beneficial effects of the present invention are:

[0027] 1. This invention uses the K-means clustering algorithm to identify weather patterns in photovoltaic power generation data, simplifying complex meteorological data into two typical categories: sunny days and non-sunny days. By using the average power and features of different clusters, the start-up and shutdown of PEM electrolyzers are dynamically adjusted to improve the system's adaptability and optimization capabilities under different weather conditions.

[0028] 2. This invention designs a multi-objective optimization function that comprehensively considers hydrogen production, power absorption rate and switching times, dynamically adjusts parameters such as particle number, iteration number, inertia weight and acceleration coefficient, and uses a local search enhancement method to perform a detailed search near the global optimal solution, which can find better local solutions and further improve the overall optimization effect.

[0029] 3. This invention combines the K-means clustering algorithm and the particle swarm optimization algorithm (PSO) to combine the impact of weather on photovoltaic power generation with the allocation and scheduling of hydrogen production from PEM electrolyzer water electrolysis, and designs an optimal coupled control strategy for PEM electrolyzers.

[0030] 4. This invention significantly reduces the number of PEM electrolyzer switching operations through multi-objective optimization and runtime management, thereby reducing equipment wear and maintenance costs, extending the service life of PEM electrolyzer equipment, and improving the efficiency and stability of the PEM water electrolysis hydrogen production system. Attached Figure Description

[0031] Figure 1 This is a flowchart of the present invention;

[0032] Figure 2 This is a diagram showing the clustering analysis results of this invention;

[0033] Figure 3 This is a comparison chart of the number of start-up and shutdown switching times of the two sets of PEM electrolyzers in this invention. Detailed Implementation

[0034] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0035] To optimize the scheduling strategy of PEM electrolyzers, increase hydrogen production and power absorption rate, and reduce the number of start-up and shutdown switching, thereby improving the efficiency and stability of the PEM water electrolysis hydrogen production system, this invention provides the following technical solution:

[0036] An optimized method for multi-cell hybrid operation of PEM water electrolysis for hydrogen production includes the following steps:

[0037] S1. Acquire photovoltaic power generation data, separate the data into daytime data and nighttime data, and perform data preprocessing. Data preprocessing includes extracting temperature, humidity and solar radiation characteristics from the acquired photovoltaic power generation daytime data, performing normalization processing, and eliminating the influence of different feature dimensions and ranges.

[0038] S2. The K-means clustering algorithm is used to divide the daytime photovoltaic power generation data into sunny and non-sunny days according to the weather pattern. The number of categories can be adjusted as needed to calculate the average power under different weather conditions. By using the average power and characteristics of different clusters, the start and stop of the PEM electrolyzer can be dynamically adjusted to improve the system's adaptability and optimization capability to different weather conditions.

[0039] The data acquisition, preprocessing, and K-means clustering algorithm are shown below:

[0040]

[0041]

[0042] K-means clustering analysis was used to group daytime photovoltaic data. The principle is to divide the data into multiple groups based on the similarity of data features, making data points within the same group as similar as possible, while maximizing the differences between data points in different clusters. This allows for the identification and analysis of photovoltaic power generation characteristics under different weather conditions. Meteorological variables (such as solar irradiance, temperature, and humidity) were extracted, and the data was divided into daytime and nighttime based on irradiance. Multiple meteorological features were selected and normalized to eliminate dimensional differences. Then, the K-means algorithm was applied to divide the daytime data into two categories (sunny days and non-sunny days). The average meteorological features of each category were analyzed, and the clustering results were visualized using a scatter plot.

[0043] S3. Construct scheduling strategy;

[0044] S301. Set the parameters of the PEM electrolyzer, including the rated power of each PEM electrolyzer and the total number of PEM electrolyzers in the system (this can be set according to specific circumstances; the data can be modified in the code).

[0045] S302, Calculate the power threshold and the number of activated PEM electrolyzers;

[0046] Based on the average photovoltaic power of each weather pattern and the safety factor, the applicable power threshold and the number of PEM electrolyzers that can be activated for each weather pattern are calculated to ensure stable operation under different weather conditions.

[0047] The scheduling strategy algorithm is constructed as follows:

[0048]

[0049]

[0050] S4. Use the improved Particle Swarm Optimization (PSO) algorithm for global search and optimization to find the optimal PEM scheduling strategy.

[0051] S401. Set a multi-objective optimization function as a weighted sum of hydrogen production, number of switching operations, and power absorption rate. The multi-objectives include maximizing hydrogen production, improving power absorption rate, and minimizing the number of switching operations. The weights of hydrogen production, number of switching operations, and power absorption rate in the multi-objective optimization function are preset according to requirements. The multi-objective optimization function is shown below:

[0052]

[0053] in, S is the hydrogen production, S is the number of switching cycles, and PAR is the power absorption rate, which represents the proportion of power absorbed by the electrolyzer to the total photovoltaic power generation. The weights k1, k2, and k3 control the relative importance of these three factors, and these weights are usually adjusted through experiments or demand.

[0054]

[0055] Where N represents the number of electrolytic cells, PEMstate(i) represents the state of the i-th electrolytic cell (0 for off, 1 for on), P is the operating power of the electrolytic cell, and Δt is the time step. This is a coefficient representing the hydrogen production per megawatt per hour.

[0056]

[0057] Where PEMstate(i) represents the state of the i-th electrolytic cell at the previous moment, and prestate(i) represents the state of the i-th electrolytic cell at the next moment.

[0058]

[0059] Where CurP is the photovoltaic power generation at the current moment.

[0060] S402. Initialize the particle swarm optimization algorithm parameters, including initializing the number of particles, the number of iterations, and the inertia weight. Introduce mutation operations, dynamically adjust parameters, and add a local search mechanism. The local search mechanism is to perform a small-scale search and adjustment in the vicinity of the generated PEM electrolyzer scheduling strategy to obtain the optimal PEM electrolyzer scheduling strategy, thereby realizing the global search and optimization of the improved particle swarm optimization algorithm PSO.

[0061] S403. By iteratively updating the velocity and position of particles, find the optimal PEM electrolyzer scheduling strategy, including start-stop thresholds and PEM electrolyzer standby duration, to achieve multi-objective balance optimization.

[0062] The particle swarm optimization algorithm is shown below:

[0063]

[0064]

[0065]

[0066] The Particle Swarm Optimization (PSO) algorithm is used to optimize the scheduling strategy of PEM electrolyzers, with the optimization objectives being the PEM start-up threshold and standby time. The algorithm works by simulating the motion of multiple particles, representing each particle (vector) as a scheduling strategy, and calculating a fitness function (objective function value) to evaluate the performance of each particle in each iteration. In the example code, the fitness function is a weighted sum of hydrogen production, switching frequency, and power absorption rate. Particles continuously update their position and velocity in the search space based on their own experience and guidance from the globally optimal particle, ultimately finding the optimal scheduling strategy. In each iteration, the algorithm dynamically adjusts the particle velocity and further improves the accuracy of the global optimum through local search strategies, finally outputting the optimal scheduling parameters and corresponding fitness values.

[0067] S5. Record the operating time and start / stop switching count of each PEM electrolyzer, and visually output the operating status of the PEM electrolyzer. Evaluate the generated PEM scheduling strategy by calculating hydrogen production and power absorption rate. The algorithm is shown below:

[0068]

[0069]

[0070]

[0071] S6. Analysis of Operation Results

[0072] Basic parameter settings are shown in Table 1:

[0073] Table 1

[0074]

[0075] like Figure 2 As shown in the figure, the two color sets of points represent different categories after clustering. K-means clustering yielded two clustering results as shown in Table 2-1 and Table 2-2.

[0076] Table 2-1

[0077]

[0078] Table 2-2

[0079]

[0080] Category 1 averages show higher irradiance, especially direct normal irradiance and total irradiance, which usually indicates a sunny or highly irradiated weather condition with moderate air temperature and low relative humidity, suggesting that it may be a dry and sunny day. Category 2 averages show lower irradiance, especially direct normal irradiance and total irradiance, which may indicate a cloudy, hazy, or foggy day. The air temperature is closer to the category, but the relative humidity is higher, indicating that the air may be relatively humid.

[0081] The method provided by this invention has a significant effect on distinguishing weather patterns. By performing K-means clustering analysis on daytime photovoltaic data, it can effectively divide weather patterns into two categories, representing sunny weather and non-sunny weather, respectively. The visualization results in the figure clearly show the distribution of these two clusters. Category 1 (sunny weather) has high irradiance and low humidity, while Category 2 (non-sunny weather) shows low irradiance and high humidity. This result verifies the effectiveness and accuracy of the proposed method in weather pattern recognition.

[0082] Specifically, in practical applications, the start-up and shutdown decisions of PEM electrolyzers are mainly based on the power generation of the photovoltaic panels at every moment. The electrolyzers need to obtain the power generation of the photovoltaic panels at any time to start and stop, which leads to different switching times for different electrolyzers. In addition, there are also issues of energy efficiency and cost considerations. In some scheduling strategies, the start-up and shutdown status of each electrolyzer may be dynamically adjusted according to the efficiency of the electrolyzers or fluctuations in electricity prices, which will also lead to uneven switching times.

[0083] Based on actual photovoltaic power generation data, the electrolyzer startup threshold was reasonably set. Under sunny weather conditions, two electrolyzers were required to complete the coupling. The results showed that the total hydrogen production was 103,955.00 kg, the number of switching times for PEM electrolyzer No. 1 was 7,524, the number of switching times for PEM electrolyzer No. 2 was 2,692, and the total number of switching times was 10,216. The total operating time of PEM electrolyzer No. 1 was 5,299.50 hours, the total operating time of PEM electrolyzer No. 2 was 5,096.50 hours, and the total power absorption under the scheduling optimization strategy was 13,291.25 MW·h.

[0084] At some point, due to an increase in hydrogen demand or a change in power demand, PEM electrolyzer No. 1 may be activated, changing its state from off (0) to on (1). If its operating state does not change in subsequent time steps, its switching count will increase by 1. Assume that in the same time step, PEM electrolyzer No. 2 remains in standby mode (not activated), or even if activated, it continues to operate in subsequent time steps without switching states. In this case, the switching count for No. 2 remains unchanged.

[0085] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A PEM electrolytic water hydrogen production multi-tank hybrid optimization method, characterized in that, The method comprises the following steps: S1, obtaining photovoltaic power generation data, separating the data into daytime data and nighttime data, and performing data preprocessing; S2, using a K-means clustering algorithm to divide the photovoltaic power generation daytime data into sunny days and non-sunny days according to weather patterns, and calculating the average power of different weather patterns; S3, constructing a scheduling strategy; S301, setting PEM electrolyzer parameters, including the rated power of each PEM electrolyzer and the total number of PEM electrolyzers in the system; S302, calculating the power threshold and the number of enabled PEM electrolyzers; According to the average photovoltaic power of each weather pattern combined with the safety factor, the power threshold and the number of enabled PEM electrolyzers suitable for each weather pattern are calculated; S4, using an improved particle swarm optimization (PSO) algorithm for global search and optimization to find the optimal PEM scheduling strategy; S401, setting a multi-objective optimization function, including maximizing hydrogen production, improving power absorption rate, and minimizing switching times; S402, initializing particle swarm optimization algorithm parameters, including initializing the number of particles, the number of iterations, and the inertia weight, introducing mutation operation, dynamic adjustment of parameters, and increasing local search mechanism; S403, updating the speed and position of the particles through iteration to find the optimal PEM electrolyzer scheduling strategy, including the start-stop threshold and the standby duration of the PEM electrolyzer; S5, recording the running time of each PEM electrolyzer and the number of state switching times, and evaluating the generated PEM scheduling strategy; The generated PEM scheduling strategy is evaluated by calculating the hydrogen production and power absorption rate.

2. The PEM water electrolysis hydrogen production multi-tank hybrid optimization method according to claim 1, characterized in that, The data preprocessing includes extracting temperature, humidity and light radiation features from the obtained photovoltaic power generation daytime data and performing normalization processing.

3. The PEM water electrolysis hydrogen production multi-tank hybrid optimization method according to claim 1, characterized in that, The multi-objective optimization function is the weighted sum of hydrogen production, switching times and power absorption rate, and the weights of hydrogen production, switching times and power absorption rate are preset according to requirements.

4. The PEM water electrolysis hydrogen production multi-tank hybrid optimization method according to claim 1, characterized in that, The local search mechanism is a small range search and adjustment around the generated PEM electrolyzer scheduling strategy to obtain the optimal PEM electrolyzer scheduling strategy.

Images