Electricity-hydrogen combined scheduling method and system for large-scale renewable energy source hydrogen production

Abstract

The invention discloses an electricity-hydrogen combined scheduling method and system for large-scale renewable energy source hydrogen production. The method comprises the steps of defining projects, equipment, units and systems, configuring topological relations, setting working conditions, grading loads and accessing data. Adjusting, optimizing and correcting ultra-short time power generation prediction, predicting hydrogen production allowance and identifying a current working condition; pre-judging future working conditions, power supply quantity and hydrogen production direct-current power; simulating balance, and solving an optimal scheduling strategy by multiple targets; and decomposing the scheduling strategy and issuing an asynchronous control instruction. According to the invention, the thought of mutual assistance of source and load is adopted, energy balance and hydrogen production balance are jointly dispatched, the purposes of power-hydrogen coupling as a main part, energy storage stabilization as an auxiliary part and off-grid or grid-connected non-grid are achieved, off-grid stability and 100% green power consumption of a renewable energy hydrogen production system are achieved, the policy requirements of green hydrogen production are met, the dependence on a large power grid is fundamentally reduced, and the green hydrogen production efficiency is improved. The electricity cost is reduced; and meanwhile, the support dependence on energy storage is reduced, and the use cost of energy storage is reduced.

Description

Technical Field

[0001] The present invention relates to a scheduling method and system for hydrogen production from renewable energy, and in particular to an electricity-hydrogen combined scheduling method and system for large-scale hydrogen production from renewable energy. Background Art

[0002] Renewable energy hydrogen production is a process that uses renewable energy such as solar energy and wind energy to produce hydrogen through water electrolysis technology. This technology not only helps reduce dependence on fossil fuels, but also promotes the use of clean energy, thereby reducing greenhouse gas emissions.

[0003] Green hydrogen is hydrogen produced by using renewable energy (such as solar energy, wind energy, nuclear energy, etc.). For example, hydrogen is produced by electrolyzing water through renewable energy. In the process of producing green hydrogen, all electricity comes from renewable energy, and no grid electricity is used, so there will be no carbon emissions at all.

[0004] Large-scale renewable energy hydrogen production systems consist of photovoltaic power generation units, wind power generation units, hydropower units, energy storage battery units, power distribution systems, grid-connected and off-grid switchers, water electrolysis hydrogen production units (including water treatment, water electrolyzers, separation and purification devices), hydrogen storage tank units, compressed air units, etc. Since the volatility of wind and solar power generation and the power supply stability requirements of the hydrogen production system are difficult to reconcile, the existing renewable energy hydrogen production management system mainly adopts the idea of ​​source following load, and ensures the stability of hydrogen production electricity through the stability of large power grids and energy storage batteries, and optimizes the scheduling of renewable energy hydrogen production systems with the maximum system economic benefits and the minimum environmental costs as the optimization goals.

[0005] Because the existing technology is based on the stability of large power grids and energy storage to achieve stable electricity consumption for hydrogen production, the electricity and energy storage costs are high. At the same time, the performance is average in the scenario of off-grid (off-grid) stable operation of renewable energy hydrogen production systems. Grid-connected operation cannot guarantee 100% use of renewable energy, and cannot fully meet the definition and policy requirements of green hydrogen production. The excessive use of energy storage makes the cost of hydrogen production high, and its value in real scenarios is not high. Summary of the invention

[0006] Purpose of the invention: The purpose of the present invention is to propose an electric-hydrogen joint scheduling method and system for large-scale renewable energy hydrogen production, so as to solve the problems of low off-grid operation stability, high off-grid electricity cost and energy storage cost caused by the high dependence of renewable energy hydrogen production system on large power grid and energy storage in the prior art.

[0007] Technical solution: The present invention includes: defining projects, equipment, units and systems, configuring topological relationships, setting operating conditions, load classification, and data access; optimizing and correcting ultra-short-term power generation predictions, predicting hydrogen production margins and identifying current operating conditions; predicting future operating conditions and power supply, and hydrogen production DC power; simulating balance and seeking optimal scheduling strategies with multiple objectives; decomposing scheduling strategies, and issuing asynchronous control instructions.

[0008] The optimization and correction of the ultra-short-term power generation forecast is specifically as follows: after receiving the result of the ultra-short-term power generation forecast, combining the historical power generation of renewable energy, the historical self-consumption power, and the historical microgrid power, the inertia calculation method is used to optimize the predicted microgrid power of renewable energy, and the predicted microgrid power is cut into h-minute granularity.

[0009] The optimization of the predicted power of renewable energy on the microgrid is specifically as follows: the predicted power generation from renewable energy, with a data frequency of g minutes; the historical power generation from on-site collection and processing, with a data frequency of h minutes; the power generation data of the day is fitted into a series (G1, G 24*60 / h ), where G1~G t Take the historical power generation; G t+1 ~G 24*60 / h , take the forecast power generation after optimization and correction; t is the nearest scale from the current moment forward, the data is updated every h minutes, and as the collected data is added, the historical data replaces the forecast data.

[0010] The tuning and correction method is:

[0011] When |G t+1 -G t |>M t When the limiting rate trend extrapolation method is used, G t+1 =G t *(1+Sa), where Sa is the maximum rate that can be carried in h minutes, and is calculated step by step to G t+1、 G t+2、 G t+3 ;

[0012] When |G t+1 -G t |≤M t When the interpolation method is used, according to [G t、 G t+3 ]Equally supplement G t+1、 G t+2 ;

[0013] This process is repeated for the next g-minute forecast data. At least six data are forecasted for each g-minute. The above method is repeated until the data is optimized. The resulting array is used as the ultra-short-term forecast of microgrid power.

[0014] Among them, Mt is the fluctuating power that the current system can carry during the g-minute period, n represents the number of hydrogen production units planned to be operated, a i Indicates the maximum fluctuation that a single group of electrolytic cells can withstand, b i represents the coefficient of variation of different working conditions of a single group of electrolytic cells; m represents the number of other controllable loads, c j Indicates the fluctuation margin of a single controllable load.

[0015] The specific method for predicting the hydrogen production margin is as follows:

[0016] Predicted hydrogen production margin = (total capacity of hydrogen storage tank * availability factor - current hydrogen storage - planned consumption) / (100% - storage loss);

[0017] Predicted maximum hydrogen production power = hydrogen production margin / hydrogen production electricity consumption; where, hydrogen production electricity consumption is the average value calculated based on the fluctuation characteristics of wind and solar power and statistically calculated by time period; hydrogen production electricity consumption = current hydrogen production electricity consumption / current hydrogen production.

[0018] The prediction and calculation method of the power supply and hydrogen production DC power is as follows:

[0019] The predicted microgrid power, predicted hydrogen production margin, and future operating plan derived from the renewable energy power generation forecast are used as the initial non-state variables, and the power system network topology, component parameters, fluctuation characteristics of wind / solar / storage under control logic, and load characteristics of various types of power-consuming equipment are used as parameters; the voltage, frequency, power, and current of the power supply point, power supply point, hydrogen production power supply point, and other similar load integration points are used as state variables, and the steady-state voltage and power of each node are calculated using power flow; the power supply power, hydrogen production DC power, power supply point power, and energy storage charging and discharging power are obtained;

[0020] Among them, considering the influence of hydrogen production margin,

[0021] When the predicted maximum hydrogen production power is greater than the predicted microgrid power during the current ultra-short period*(1-grid loss ratio), the predicted microgrid power during the current ultra-short period is taken as the initial variable;

[0022] When the predicted maximum hydrogen production power is ≤ the predicted microgrid power consumption during the current ultra-short period*(1-grid loss ratio), the predicted maximum hydrogen production power is taken as the initial variable;

[0023] According to the above method, the h-minute frequency data is extrapolated step by step to obtain the power dispatch plan data for the next 1-4 hours.

[0024] The simulation balance is to train the operation mode of each electrolytic cell through a machine learning algorithm.

[0025] The multi-objective optimal scheduling strategy is specifically as follows: Using the trained algorithm model, with the changing trend of the start and stop of the hydrogen production equipment as the target, and the minimum number of equipment start and stop times and the operation duration of the water electrolyzers within a single group approaching consistency as the constraint conditions, identify the power consumption data and operation status of each hydrogen production unit included in the total hydrogen production DC power consumption;

[0026] According to the power consumption data and operation status of each electrolyzer, substitute them into the hydrogen production system simulation, simulate the consumption of public auxiliary and materials, deduce the power change of the public auxiliary power equipment, and correct the power supply power and the energy storage charge and discharge power, and perform secondary optimization on the power balance;

[0027] Based on the above power distribution plan and the correction results of the power supply quantity and the power grid connection quantity data, determine the adjustment parameters for the next cycle.

[0028] The decomposition scheduling strategy, the asynchronous control instruction is issued, specifically as follows:

[0029] Set the strategy issuance period T of each unit / controller i ;

[0030] When Ti = h minutes, issue the target parameters or interpolation calculation of this unit as a parameter trend array;

[0031] When Ti < h minutes, in the interpolation calculation, divide the current operation of the unit participating in the strategy target into h / Ti equal parts and issue the decomposition steps hour by hour;

[0032] When Ti > h minutes and Ti can be divided evenly by h, merge the unit control strategies of Ti / h groups into trend parameters, or issue the strategy with the last target parameter;

[0033] Arrange the above decomposed / merged control instructions in the instruction issuance queue according to the issuance time. The instruction issuance queue issues the control instructions of the same batch according to the time plan through the message channel and receives the feedback message mark.

[0034] A renewable energy hydrogen production electro-hydrogen combined scheduling system includes: an information definition module, a data calculation module, a data storage module, an Internet of Things acquisition module, a data interface module, an energy balance module, a comprehensive scheduling module, and a human-computer interaction module, which are divided into a data layer, an acquisition layer, an application layer, and a display layer; among them, the information definition module, the data calculation module, and the data storage module are located in the data layer; the Internet of Things acquisition module and the data interface module are located in the acquisition layer; the energy balance module and the comprehensive scheduling module are located in the application layer; the human-computer interaction module is located in the display layer.

[0035] Beneficial effect: The present invention adopts the idea of ​​source-load mutual assistance, and jointly dispatches energy balance and hydrogen production balance to achieve the purpose of electricity-hydrogen coupling as the main, energy storage as the auxiliary, off-grid or grid-connected without grid, so as to achieve off-grid stability of renewable energy hydrogen production system, 100% green electricity consumption, meet the policy requirements of green hydrogen production, fundamentally reduce dependence on large power grids, and reduce electricity costs; at the same time, reduce dependence on energy storage support and reduce the cost of energy storage. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the system structure of the present invention;

[0037] Figure 2 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0038] The present invention will be further described below in conjunction with the accompanying drawings.

[0039] like Figure 1 As shown, the electricity-hydrogen joint scheduling system for hydrogen production from renewable energy of the present invention includes: information definition module, data calculation module, data storage module, Internet of Things acquisition module, data interface module, energy balance module, comprehensive scheduling module and human-computer interaction module, which is divided into four layers of structure: data layer, acquisition layer, application layer and display layer. Among them, the information definition module, data calculation module and data storage module are located in the data layer; the Internet of Things acquisition module and data interface module are located in the acquisition layer; the energy balance module and comprehensive scheduling module are located in the application layer; and the human-computer interaction module is located in the display layer.

[0040] The data layer, as the basic capability layer, provides the entire system with the ability to define, process, store, and analyze data.

[0041] Among them, the information definition module is used to define the digital attributes of equipment, units, systems, projects and other things, establish energy and production operation models, and determine the data acquisition method; the defined information models include: photovoltaic power generation model, wind power generation model, hydropower generation model, energy storage model, distribution system model, water electrolysis hydrogen production model, hydrogen storage model and other unit or system models; single group electrolyzer model, water treatment device model, air compressor model, hydrogen storage tank model and other equipment models; microgrid flow model, electricity-hydrogen balance model, hydrogen production control model and other system models. The unit model has built-in operating indicators and operating relationships of multiple equipment, simulation formulas, energy input / output indicators, production input / output indicators and dynamic electricity weights; equipment information definition includes equipment interpretation, data acquisition method, energy efficiency model, operation indicators, production indicators, and fault definition. The system information model definition includes system interpretation, system type and purpose, internal and external topological relationships, input / output indicators, simulation formulas, steady-state parameters, etc.

[0042] The data computing module is used to process, count, manage, analyze, and simulate multi-source data. It adopts multiple computing frameworks such as streaming computing, batch processing, and real-time computing. According to business requirements and computing rules, it customizes mixed computing expressions and logical expressions to realize the full-cycle processing and computing of data from collection, verification, cleaning, statistics, management, advanced analysis, freezing, etc.

[0043] The data storage module is used to store real-time data, relational data, and time series data; it includes various types of databases such as cache databases, relational databases, real-time databases, historical time series databases, etc., and builds data flows in conjunction with the data computing module according to business requirements.

[0044] The acquisition layer provides the system with the ability to acquire and send data. Depending on the type of external connection system, it is divided into two IoT acquisition modules and data interface modules.

[0045] The IoT acquisition module is used to collect various types of data from field equipment and process control systems; it follows the on-site data collection of equipment, units, and systems in the information definition module, including equipment nameplates, equipment operating parameters, set parameters, fault signals, unit supply and demand data, system status data, etc.

[0046] The data interface module is used to receive information data such as plans, operations, markets, and forecasts from the production side and the Internet side; including production plans and consumption plans from the production and operation system, products from the Internet, raw material price fluctuations, wind and solar power generation forecast data from third-party systems, and business plans from within the enterprise. It meets the interface specifications and data provision cycles of various external systems, decomposes and fits the data into minute-level trend data, and matches the scheduling method.

[0047] The application layer is the core capability layer for realizing the scheduling of hydrogen production from renewable energy. It follows the electricity-hydrogen joint scheduling method to realize the stable operation of the renewable energy hydrogen production system. It mainly includes energy balance module and comprehensive scheduling module.

[0048] The energy balance module is used for energy efficiency analysis, multi-energy balance analysis, electricity-hydrogen balance analysis and comprehensive energy management in real and simulated scenarios; it includes comprehensive monitoring and unit monitoring, energy efficiency and unit consumption analysis, operating condition analysis, load integration management, power quality analysis, microgrid flow calculation, source-load prediction, hydrogen production prediction, energy balance evaluation and other functions. It analyzes the energy balance and electricity-hydrogen balance under various operating conditions in multiple dimensions, and analyzes the characteristics of steady-state conditions for further optimization of algorithms and strategies.

[0049] The comprehensive dispatching module asynchronously controls and optimizes the operation of each unit / equipment based on renewable energy prediction, energy balance analysis, simulation calculation, multi-objective optimization, etc. It includes production and consumption plan analysis, energy allocation, control instruction queue, dispatching strategy model, solver, emergency plan, excellent dispatching examples and other functions. Through rationality analysis and calculation, it simulates and evaluates the future steady-state conditions and dispatching strategies, adopts asynchronous control ideas, and gradually issues control instructions to the process control devices of each unit / equipment. Considering the influence of multiple objective variables such as production capacity, predicted power, equipment failure, smoothing fluctuations, product quality, and power quality, the stable operation of large-scale renewable energy hydrogen production systems can be achieved.

[0050] The presentation layer is the interactive interface between the digital system and the user. It displays the comprehensive system status and scheduling strategy to the user. It can also receive decision instructions from the user and pass them down to the application layer for processing.

[0051] The human-computer interaction module is a human-computer interaction interface for user visual operation, viewing and analysis. It mainly includes visual cockpit, comprehensive monitoring and various unit monitoring, fault alarm and processing, analysis and report management functions, and dispatch management functions, providing users with a convenient, efficient and intuitive management interface.

[0052] The electricity-hydrogen joint scheduling method for producing hydrogen from renewable energy of the present invention comprises the following steps:

[0053] S1. Definition of projects, equipment, units, and systems, topology configuration, working condition setting, load classification, data access and other preparatory work, including:

[0054] Production project parameter configuration, including project name, city location, product and designed output, target product unit consumption, organizational personnel, etc.

[0055] Equipment information configuration, including equipment name, nameplate parameters, asset information such as user and operator, operation data collection items, IoT point relationships, etc. The main production-related equipment types include photovoltaic inverters, photovoltaic panels, wind turbines, wind power inverters, energy storage batteries, PCS, transformers, incoming circuit breakers, feeder switches, DC power supplies, water electrolyzers, water treatment devices, hydrogen storage tanks, gas flow meters, liquid flow meters, energy meters, UPS, edge gateways, and other equipment.

[0056] Unit definition includes the unit's energy type, unit name, equipment contained in the unit, unit operation model (including energy, production) related input, output and internal parameter settings, and unit alarm rules. Unit types mainly include photovoltaic power generation units, wind power units, energy storage battery units, power distribution system units, single-group hydrogen production units, hydrogen storage units, and if necessary, expandable hydrogen consumption units such as synthetic ammonia units, synthetic methanol units, and hydrogen refueling station units.

[0057] The system is considered to be a composite independent unit composed of units with the same or similar business functions in the entire production process, which may not be directly connected in terms of geographical location and physical connection. For example, the green hydrogen microgrid system (power generation + power supply and distribution system), hydrogen production system (multiple groups of hydrogen production units and connecting pipelines), hydrogen storage and transportation system, and green hydrogen system (overall) are defined including the name of the system, the business type of the system, the included units / equipment, system alarm rules, inputs and outputs of the operation model (including energy and production), and internal parameter settings.

[0058] Topological relationships, through the topological diagram to draw the internal and external topological relationships of each system, the external relationship through the input and output connection relationship, the internal relationship through the input and output connection relationship between units / equipment. The connection relationship line marks the medium (energy) type, transportation method, carrier characteristics (length, material, characteristics), transportation speed range, and rated voltage, pressure, temperature and other operating characteristics (can be associated through equipment or metering point information).

[0059] The operating condition refers to the operating status of the overall renewable energy hydrogen production system, including startup conditions, cold start conditions, hot start conditions, stable operation conditions, shutdown conditions, abnormal conditions, etc., and is divided into different conditions under grid connection / off-grid. The operating condition setting includes the name of the operating condition, entry conditions, exit conditions, operating status of each unit / equipment in the system, and load range. And maintain the one-way conversion relationship between the operating conditions.

[0060] Load classification is to set the load priority principle when allocating limited energy. Taking electricity as an example, the initial classification adopts the important load classification method, and the power equipment is classified into uninterruptible load (level 0), key production load (level 1), important load (level 2), and other load (level 3). For further classification, a matrix scoring system of equipment / units and working conditions is used to dynamically calculate the priority of each load.

[0061] Data access, through the Internet of Things collection module and data interface module, configure the data that needs to be collected and received, including equipment operation data, alarm signals, action signals, photovoltaic / wind power generation forecast data, consumption plan data, etc.

[0062] S2, inertia method optimization of ultra-short-term power generation prediction, prediction of hydrogen production margin, identification of current operating conditions, etc.

[0063] Adjust and correct the ultra-short-term power generation forecast. After receiving the results of the ultra-short-term power generation forecast, combine the historical power generation of renewable energy, historical self-consumption, and historical microgrid power (to the microgrid, not to the main grid), and use the inertia calculation method to optimize the predicted microgrid power of renewable energy, and cut the predicted microgrid power into 5-minute granularity.

[0064] The predicted power generation from renewable energy sources has a data frequency of g minutes; the historical power generation comes from on-site collection and processing, with a data frequency of h minutes; the power generation data of the day is fitted into a series (G1, G 24*60 / h ), where G1~G t Take the historical power generation; G t+1 ~G 24*60 / h , take the predicted power generation after adjustment and correction; t is the nearest scale from the current moment, and the data is updated every h minutes. As the collected data is added, the historical data replaces the predicted data. In this embodiment, g = 15 minutes, h = 5 minutes, that is:

[0065] The predicted power generation from renewable energy sources has a data frequency of 15 minutes. The historical power generation comes from on-site collection and processing, with a data frequency of 5 minutes. The power generation data of the day is fitted into a series (G1, G 288 ), where G1~G t Take the historical power generation; G t+1 ~G 288 , take the forecast power generation after optimization and correction; t is the nearest scale from the current moment forward, the data is updated every 5 minutes, and as the collected data is added, the historical data replaces the forecast data.

[0066] The tuning and correction methods are as follows:

[0067] When |G t+1 -G t |>M t When the limiting rate trend extrapolation method is used, G t+1 =G t *(1+Sa), Sa is the maximum loadable rate in 5 minutes, which is calculated step by step to G t+1、 G t+2、 G (t+3) ;

[0068] When |G t+1 -G t |≤M t When the interpolation method is used, according to [G t、 G t+3 ]Equally supplement G t+1、 G t+2 .

[0069] This is repeated for the next 15-minute forecast data. At least 6 data are forecasted every 15 minutes. In this embodiment, 6 data are forecasted every 15 minutes. The above method is repeated until the data is optimized after 4 hours to obtain the array (G t ,G t+48 ) as an ultra-short-term prediction of microgrid power consumption.

[0070] Among them, M tis the amount of fluctuating electricity that the current system can carry (15-minute period),

[0071] Where n represents the number of hydrogen production units planned to be operated, a i Indicates the maximum fluctuation that a single group of electrolytic cells can withstand, b i represents the coefficient of variation of different working conditions of a single group of electrolytic cells; m represents the number of other controllable loads, c j Indicates the fluctuation margin of a single controllable load.

[0072] The specific method for predicting hydrogen production margin is as follows:

[0073] Predicted hydrogen production margin = (total capacity of hydrogen storage tanks * availability factor - current hydrogen storage - planned consumption) / (100% - storage loss). Combined with ultra-short-term power generation forecast, optimize the hydrogen production electricity margin after 1 to 4 hours.

[0074] The predicted maximum power of hydrogen production = hydrogen production margin / hydrogen production electricity consumption. The hydrogen production electricity consumption is the average value calculated by time period based on the fluctuation characteristics of wind and solar power. Hydrogen production electricity consumption = current hydrogen production electricity consumption / current hydrogen production.

[0075] Identify the current working condition and determine the current system working condition based on the working condition settings and the current equipment / unit operating status. If the identification fails, analyze the current equipment operation status and modify or add working condition information in the working condition settings.

[0076] S3. Prediction of future operating conditions, power supply and hydrogen production DC power.

[0077] Predict future operating conditions. Combined with the results in step S2 (power margin for hydrogen production, predicted power consumption on the microgrid), predict the system operating conditions 1-4 hours later based on the operating condition settings and the principles of stability and economy. Identify the changing trends of equipment start-up and shutdown based on future operating condition changes (faulty or specially marked equipment is not included in the analysis).

[0078] Prediction and calculation method of power supply power and hydrogen production DC power:

[0079] The predicted microgrid power, predicted hydrogen production margin, and future operating plan (possible start-stop sequence of future equipment) derived from the power generation forecast of renewable energy are used as initial non-state variables, and the power system network topology, component parameters, fluctuation characteristics of wind / solar / storage under control logic, and load characteristics of various types of power-consuming equipment are used as parameters; the voltage, frequency, power, and current of power supply points, power supply points, hydrogen production power supply points, and other similar load integration points are used as state variables, and power flow calculation (including the Gossip-Seidel method, Newton-Raphson method, PQ decomposition method, or AI algorithms such as convolutional neural networks (CNN)) are used to obtain the steady-state voltage and power of each node. The power supply power, hydrogen production DC power, power supply point power, and energy storage charging and discharging power are obtained.

[0080] Among them, considering the influence of hydrogen production margin,

[0081] When the predicted maximum hydrogen production power is greater than the predicted microgrid power during the current ultra-short period*(1-grid loss ratio), the predicted microgrid power during the current ultra-short period is taken as the initial variable.

[0082] When the predicted maximum hydrogen production power is ≤ the predicted microgrid power consumption during the current ultra-short period*(1-grid loss ratio), the predicted maximum hydrogen production power is taken as the initial variable.

[0083] According to the above method, the 5-minute frequency data is extrapolated step by step to obtain the power supply for the next 1-4 hours, load power curve, power curve of direct current electricity for hydrogen production, executable power curve of microgrid, energy storage charging and discharging curve and other power dispatch plan data.

[0084] S4, simulation balance, multi-objective optimal scheduling strategy

[0085] According to the power curve of the direct current electricity consumption for hydrogen production in step S3, the data of the most recent 5 minutes are taken as the total direct current electricity consumption load for hydrogen production solved in the most recent planning and scheduling.

[0086] Load feature extraction

[0087] According to the working characteristics of the water electrolyzer, the change point identification method is used to identify the time point when the water electrolyzer changes from cold start, hot start, stable operation, shutdown, etc. as an event, and identify the conditions and load changes for each event.

[0088] The operation mode of each electrolytic cell is trained through machine learning algorithms (Adaboost algorithm, K nearest neighbor algorithm, sparse coding technology, support vector machine, artificial neural network method).

[0089] Using the trained algorithm model, taking the changing trend of the start and stop of the hydrogen production equipment in step S2 as the target, and taking the minimum number of equipment start and stop times and the consistent operating time of the water electrolyzers in a single group as the constraints, the power consumption data and operating status of each hydrogen production unit included in the total hydrogen production DC power consumption are identified.

[0090] According to the electricity consumption data and operating status of each electrolyzer, it is substituted into the hydrogen production system simulation to simulate the consumption of public auxiliaries and materials such as water, heat, alkali solution, and compressed air, and deduce the power changes of public auxiliary power equipment such as pure water system, water replenishment system, heating system, and air compressor. And make corrections to the power supply power and energy storage charging and discharging power. Make secondary adjustments to the power balance.

[0091] Based on the above power distribution scheme and the correction results of the power supply and microgrid power data in step S3, the adjustment parameters of photovoltaic, wind power, grid connection point, voltage source, hydrogen production system, public auxiliary equipment, energy storage battery, and other controllable loads in the next cycle (5-minute frequency) are determined.

[0092] The above method performs an optimization operation every 5 minutes based on the latest collected operating conditions and power consumption of each unit, as well as the predicted correction data. The comprehensive optimization goals in the above steps include: the minimum number of equipment starts and stops, the approximate operation time of a single group of water electrolyzers, the voltage drop at key nodes of the power grid meets the standard, and the rate of change of electric power at key nodes does not exceed the limit value.

[0093] S5. Decompose the scheduling strategy and issue asynchronous control instructions

[0094] According to the scheduling strategy of each independent control unit, combined with the current operating status, output / power change rate, and control method of each unit, the scheduling strategy of each unit is decomposed / merged.

[0095] The method of decomposing / merging the scheduling strategies of each unit is as follows:

[0096] Set the policy delivery cycle T for each unit / controller i ;

[0097] When Ti=5 minutes, the target parameter or interpolation value of the unit is calculated as a parameter trend array (depending on the requirements of the lower controller) and issued.

[0098] When Ti<5 minutes, use interpolation calculation to divide the unit's current operation participation and strategy target into 5 / Ti parts, and issue the decomposition steps hour by hour.

[0099] When Ti>5 minutes and Ti is divisible by 5, the unit control strategies of the Ti / 5 group are merged into trend parameters, or the strategy is issued with the last target parameter (depending on the requirements of the lower-level controller).

[0100] The above decomposed / merged control instructions are arranged in the instruction delivery queue according to the delivery time. The instruction delivery queue delivers the same batch of control instructions according to the time plan through the message channel and receives the feedback message mark. The message that failed to be delivered is immediately delivered again. If the delivery fails three times, an alarm is triggered. After the new batch of control instructions enters the instruction delivery queue, the unsent instructions become invalid and an alarm is triggered. The decomposed same batch strategy is marked with a batch identifier to identify whether the strategies executed by each unit are consistent in batches.

[0101] After receiving the control command, the process control system of each unit will feedback the successful command reception message. At regular intervals, the system collects the operation changes of each unit to evaluate whether the scheduling expectations are met. If not, the control strategy is fine-tuned according to the deviation.

Claims

1. A method for the combined dispatch of electricity and hydrogen for large-scale renewable energy hydrogen production, characterized in that: include: Define projects, equipment, units and systems, configure topological relationships, set operating conditions, load classification, and data access; optimize and correct ultra-short-term power generation forecasts, predict hydrogen production margins and identify current operating conditions; predict future operating conditions and power supply, hydrogen production DC power; simulate balance and multi-objective optimal scheduling strategies; Decompose the scheduling strategy and issue asynchronous control instructions.

2. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 1, characterized in that: The optimization and correction of the ultra-short-term power generation forecast is specifically as follows: after receiving the result of the ultra-short-term power generation forecast, combining the historical power generation of renewable energy, the historical self-consumption power, and the historical microgrid power, the inertia calculation method is used to optimize the predicted microgrid power of renewable energy, and the predicted microgrid power is cut into h-minute granularity.

3. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 2, characterized in that: The optimization of the predicted power of renewable energy on the microgrid is specifically as follows: the predicted power generation from renewable energy, with a data frequency of g minutes; the historical power generation from on-site collection and processing, with a data frequency of h minutes; the power generation data of the day is fitted into a series (G1, G 24*60 / h ), where G1~G t Take the historical power generation; G t+1 ~G 24*60 / h , take the forecast power generation after optimization and correction; t is the nearest scale from the current moment forward, the data is updated every h minutes, and as the collected data is added, the historical data replaces the forecast data.

4. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 3, characterized in that: The tuning and correction method is: When |G t+1 -G t |>M t When the limiting rate trend extrapolation method is used, G t+1 =G t *(1+Sa), Sa is the maximum rate that can be carried in h minutes, which is calculated step by step to G t+1、 G t+2、 G t+3 ; When |G t+1 -G t |≤M t When the interpolation method is used, according to [G t、 G t+3 ]Equally supplement G t+1、 G t+2 ; This process is repeated for the next g-minute forecast data. At least six data are forecasted for each g-minute. The above method is repeated until the data is optimized. The resulting array is used as the ultra-short-term forecast of microgrid power. Among them, M t is the fluctuating power that the current system can carry during the g-minute period, n represents the number of hydrogen production units planned to be operated, a i Indicates the maximum fluctuation that a single group of electrolytic cells can withstand, b i represents the coefficient of variation of a single group of electrolytic cells under different operating conditions; m represents the number of other controllable loads, c j Indicates the fluctuation margin of a single controllable load.

5. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 1, characterized in that: The specific method for predicting the hydrogen production margin is as follows: Predicted hydrogen production margin = (total capacity of hydrogen storage tank * availability factor - current hydrogen storage - planned consumption) / (100% - storage loss); Predicted maximum hydrogen production power = hydrogen production margin / hydrogen production electricity consumption; where, hydrogen production electricity consumption is the average value calculated based on the fluctuation characteristics of wind and solar power and statistically calculated by time period; hydrogen production electricity consumption = current hydrogen production electricity consumption / current hydrogen production.

6. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 5, characterized in that: The prediction and calculation method of the power supply and hydrogen production DC power is as follows: The predicted microgrid power, predicted hydrogen production margin, and future operating plan derived from the renewable energy power generation forecast are used as initial non-state variables, and the power system network topology, component parameters, fluctuation characteristics of wind / solar / storage under control logic, and load characteristics of various power-consuming equipment are used as parameters; The voltage, frequency, power and current of the power supply point, power supply point, hydrogen production power supply point and other similar load integration points are used as state variables, and the steady-state voltage and power of each node are calculated using power flow; the power supply power, hydrogen production DC power, power supply point power and energy storage charging and discharging power are obtained; Among them, considering the influence of hydrogen production margin, When the predicted maximum hydrogen production power is greater than the predicted microgrid power during the current ultra-short period*(1-grid loss ratio), the predicted microgrid power during the current ultra-short period is taken as the initial variable; When the predicted maximum hydrogen production power is ≤ the predicted microgrid power consumption during the current ultra-short period*(1-grid loss ratio), the predicted maximum hydrogen production power is taken as the initial variable; According to the above method, the h-minute frequency data is extrapolated step by step to obtain the power dispatch plan data for the next 1-4 hours.

7. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 1, characterized in that: The simulation balance is to train the operation mode of each electrolytic cell through a machine learning algorithm.

8. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 7, characterized in that: The multi-objective optimal scheduling strategy is specifically as follows: using a trained algorithm model, taking the change trend of the start and stop of the above-mentioned hydrogen production equipment as the target, taking the minimum number of equipment start and stop times and the consistency of the running time of the water electrolyzer in a single group as the constraint conditions, identifying the power consumption data and operating status of each hydrogen production unit included in the total hydrogen production DC power consumption; According to the power consumption data and operating status of each electrolytic cell, substitute them into the hydrogen production system simulation, simulate the consumption of public utilities and materials, deduce the power change of public utility electrical equipment, correct the power supply power and energy storage charge and discharge power, and perform secondary optimization on the power balance; Based on the above power distribution plan and the correction results of the power supply quantity and the power quantity on the microgrid, determine the adjustment parameters for the next cycle.

9. The method for combined electricity-hydrogen dispatching for large-scale renewable energy hydrogen production according to claim 8, characterized in that: For the decomposition scheduling strategy, asynchronous control instructions are issued, specifically: Set the policy delivery cycle T for each unit / controller i ; When Ti = h minutes, the target parameters or interpolation calculations of the unit are issued as a parameter trend array; When Ti < h minutes, during interpolation calculation, the current operating participation strategy target of the unit is equally divided into h / Ti parts, and the decomposition steps are issued hourly; When Ti > h minutes and Ti is divisible by h, the unit control strategies of Ti / h groups are merged into trend parameters, or the strategy is issued with the last target parameter; Arrange the above decomposed / merged control instructions into the instruction issuance queue according to the issuance time. The instruction issuance queue issues the control instructions of the same batch according to the time plan through the message channel and receives the feedback message mark.

10. An electricity-hydrogen joint dispatching system for hydrogen production from renewable energy, characterized in that: Including: An information definition module, a data calculation module, a data storage module, an IoT acquisition module, a data interface module, an energy balance module, a comprehensive scheduling module, and a human-machine interaction module, which are divided into a data layer, an acquisition layer, an application layer, and a display layer; among them, the information definition module, the data calculation module, and the data storage module are located in the data layer; the IoT acquisition module and the data interface module are located in the acquisition layer; the energy balance module and the comprehensive scheduling module are located in the application layer; the human-machine interaction module is located in the display layer.

Images