Cross-day two-stage random scheduling method for industrial park integrated energy system

Abstract

The invention provides a cross-day two-stage random scheduling method for an industrial park integrated energy system, and the method comprises the steps: fitting the output fluctuation characteristics of new energy power generation equipment in a cross-day time scale based on historical data, and constructing an uncertainty scene set containing a new energy prediction error; establishing a cross-day two-stage stochastic programming model; according to the two-stage model, an optimal scheduling scheme is solved so as to minimize the overall operation cost, and the operation cost comprises the demand electric charge, the electricity purchase electric charge, the new energy power abandoning cost, the unit start-stop cost and the standby penalty cost; and on the basis of tie line power constraint and dynamic response characteristics of the multi-energy coupling equipment, feasibility verification and rolling optimization adjustment are performed on the scheduling scheme. According to the method, the new energy consumption capability of the industrial park integrated energy system under the cross-day time scale can be effectively improved, the total operation cost is reduced, and collaborative scheduling of demand cost optimization and spot market participation is realized.

Description

Technical Field

[0001] This invention relates to the field of cross-day stochastic optimization scheduling of integrated energy systems, and in particular to a cross-day two-stage stochastic scheduling method for integrated energy systems in industrial parks. Background Technology

[0002] Industrial parks, as a significant source of energy consumption and carbon emissions in my country, are widely used in various fields such as industrial production, commercial operation, and regional energy supply. With the deepening implementation of the "dual-carbon" policy, industrial parks urgently need to construct integrated energy systems (IES) that coordinate multiple energy sources, including wind, solar, energy storage, and hydrogen, to achieve low-carbon and economical operation goals. Among related technologies, a multi-timescale optimized scheduling system has been constructed through the coordinated operation of new energy forecasting, energy storage scheduling, and market trading. Specifically, this system covers the entire process from day-ahead to real-time, including key aspects such as unit start-up and shutdown planning, power allocation, energy balance, and tie-line power control. Demand and cost optimization, new energy consumption, and spot market response have become the current research focus.

[0003] However, existing dispatching methods directly employ daily optimization strategies without fully considering the characteristic that renewable energy output fluctuations can last from half a day to a full day. This may result in renewable energy not being able to be consumed locally during certain periods, leading to power curtailment losses, or discrepancies between demand charge declarations and actual operation, thus affecting the system's economics and stability. Furthermore, traditional methods typically fail to effectively couple weekly demand charges with day-ahead / real-time spot market transactions, making it difficult to cope with the dual uncertainties of electricity price fluctuations and renewable energy forecasting errors, thus limiting the park's operational flexibility and revenue potential in a power market environment. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the related art.

[0005] This invention proposes a two-stage stochastic scheduling method for integrated energy systems in industrial parks across days, which effectively improves the renewable energy absorption capacity of integrated energy systems in industrial parks on a cross-day time scale, reduces total operating costs, and achieves coordinated scheduling with demand-cost optimization and spot market participation.

[0006] Another objective of this invention is to propose a cross-day two-stage random scheduling device for an integrated energy system in an industrial park.

[0007] To achieve the above objectives, this invention proposes a two-stage stochastic scheduling method for an integrated energy system in an industrial park, comprising: S1, based on historical data, fits the output fluctuation characteristics of new energy power generation equipment in the cross-day time scale, and constructs an uncertainty scenario set including new energy prediction errors; S2. Establish a two-stage stochastic programming model across days. The decision variables in the first stage include the weekly demand declaration value and the start-up and shutdown plan of slow equipment. The decision variables in the second stage include the power purchased and sold in the real-time market during the day and the charging and discharging power of energy storage equipment. S3. Based on the two-stage model, solve for the optimal scheduling scheme to minimize the overall operating cost, which includes demand electricity cost, electricity purchase cost, renewable energy curtailment cost, unit start-up and shutdown cost, and reserve penalty cost. S4. Based on tie-line power constraints and the dynamic response characteristics of multi-energy coupled devices, the scheduling scheme is subjected to feasibility verification and rolling optimization adjustment.

[0008] The cross-day two-stage stochastic scheduling method for the integrated energy system of industrial parks according to embodiments of the present invention may also have the following additional technical features: In one embodiment of the present invention, S1 includes: S11, a probability distribution function is used to fit the historical new energy output data to generate a joint probability distribution model of wind power and photovoltaic power; S12, Based on the Monte Carlo sampling method, generate multiple uncertainty scenarios from the joint probability distribution model, and assign corresponding occurrence probability weights to each scenario.

[0009] In one embodiment of the present invention, S2 includes: S21, the slow-speed equipment includes H-CHP units and electro-hydrogen production equipment, whose start-up and shutdown schedules are fixed on a weekly scale and remain unchanged during daily scheduling; S22, the power purchased and sold is modeled in the day-ahead market and the real-time market respectively, and dynamic adjustment is allowed in the real-time market based on the actual output of new energy and electricity price fluctuations.

[0010] In one embodiment of the present invention, S3 includes: S31, A penalty coefficient k is introduced into the demand electricity charge calculation. When the actual power peak exceeds the pre-declared maximum demand, the excess part is charged at k times the unit demand price. S32, the backup penalty cost is used to perform punitive optimization on scheduling schemes that fail to meet backup requirements by introducing a small positive penalty coefficient.

[0011] In one embodiment of the present invention, it further includes: S5. Based on the power constraints of the tie line and the dynamic response characteristics of the multi-energy coupling equipment, the scheduling scheme is continuously optimized and adjusted. At the end of each scheduling cycle, the model parameters are updated based on the latest new energy output data and load forecast, and the optimization problem for the subsequent scheduling cycle is solved again.

[0012] To achieve the above objectives, another aspect of the present invention proposes a two-stage stochastic scheduling device for an integrated energy system in an industrial park, comprising: The new energy fluctuation characteristic modeling module is used to fit the output fluctuation characteristics of new energy power generation equipment on a cross-day time scale based on historical data, and to construct an uncertainty scenario set that includes new energy prediction errors. The two-stage model building module is used to establish a cross-day two-stage stochastic programming model. The decision variables in the first stage include the weekly demand declaration value and the start-up and shutdown plan of slow equipment. The decision variables in the second stage include the power purchased and sold in the real-time market during the day and the charging and discharging power of energy storage equipment. The scheduling optimization solution module is used to solve the optimal scheduling scheme based on the two-stage model to minimize the overall operating cost, which includes demand electricity cost, electricity purchase cost, renewable energy curtailment cost, unit start-up and shutdown cost, and reserve penalty cost. The scheduling scheme verification and optimization module is used to perform feasibility verification and rolling optimization adjustments on the scheduling scheme based on tie-line power constraints and the dynamic response characteristics of multi-energy coupled devices.

[0013] The two-stage random dispatch method and apparatus for integrated energy systems in industrial parks across days according to embodiments of the present invention can effectively improve the renewable energy absorption capacity of integrated energy systems in industrial parks on a cross-day time scale, reduce demand electricity costs and overall operating costs, and achieve synergistic optimization of economic efficiency and low carbon emissions.

[0014] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0015] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a two-stage stochastic scheduling method for an integrated energy system in an industrial park across days, according to an embodiment of the present invention. Figure 2 This is a structural diagram of a cross-day two-stage random scheduling device for an integrated energy system in an industrial park, according to an embodiment of the present invention. Detailed Implementation

[0016] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0017] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0018] The following describes, with reference to the accompanying drawings, a two-stage stochastic scheduling method and apparatus for an integrated energy system in an industrial park across days, according to an embodiment of the present invention.

[0019] Example 1 Figure 1 This is a flowchart of a two-stage stochastic scheduling method for an integrated energy system in an industrial park across days, according to an embodiment of the present invention. Figure 1 As shown, it includes: S1, based on historical data, fits the output fluctuation characteristics of new energy power generation equipment in the trans-day time scale, and constructs an uncertainty scenario set including new energy prediction errors.

[0020] Specifically, this step aims to fit the output fluctuation characteristics of new energy power generation equipment over a trans-diurnal timescale based on historical data and construct an uncertainty scenario set including new energy prediction errors. This is a crucial prerequisite for realizing the trans-diurnal two-stage stochastic scheduling method for integrated energy systems. At the technical implementation level, this step first collects actual output data of new energy equipment such as wind power and photovoltaics over historical cycles (e.g., the past year or multiple typical meteorological cycles), and combines this data with meteorological data (e.g., wind speed, irradiance, temperature) to perform time series modeling. Optionally, methods such as Autoregressive Moving Average (ARMA), Long Short-Term Memory (LSTM), or Support Vector Regression (SVR) are used to fit the new energy output to extract its fluctuation patterns and periodic characteristics over a trans-diurnal scale (e.g., 24 hours to 72 hours). Further, statistical analysis (e.g., mean squared error, coefficient of variation, probability density function fitting) quantifies the distribution characteristics of prediction errors, constructing a multidimensional probability distribution model, such as a multivariate normal distribution, Copula function, or kernel density estimation (KDE) model, to reflect the correlation and uncertainty of new energy output at different time points.

[0021] At the parameter level, the prediction error of new energy sources is usually quantified as relative error (RE) or root mean square error (RMSE). For example, the RMSE of wind power prediction error is generally between 10% and 20%, while that of photovoltaic power may be higher, reaching 15% to 30%. During the construction of the scenario set, key parameters such as the number of scenarios (e.g., 50-100 scenarios), confidence level (e.g., 95%), and time resolution (e.g., 1 hour per scheduling period) need to be set to ensure a balance between computational complexity and uncertainty coverage. Simultaneously, the spatiotemporal correlation of new energy output needs to be considered, and Cholesky decomposition or Monte Carlo sampling methods should be used to generate multi-period output scenarios with actual physical meaning.

[0022] At the application level, this step is applicable to cross-day dispatch scenarios where integrated energy systems in industrial parks participate in the electricity spot market. By constructing a set of uncertain scenarios, the dispatch model can formulate robust start-up and shutdown plans in the day-ahead phase and make rolling adjustments based on specific scenarios in the real-time phase, thereby improving the system's ability to cope with fluctuations in renewable energy sources and reducing the risk of power curtailment and operating costs.

[0023] In terms of technical effects, this step effectively improves the accuracy of the scheduling model in characterizing the uncertainty of new energy sources, provides a reliable input basis for two-stage stochastic programming, and enhances the system's economic efficiency and low-carbon operation capability on a trans-diurnal scale, which has significant engineering practical value and innovative significance.

[0024] Furthermore, S1 includes: S11 uses a probability distribution function to fit historical renewable energy output data to generate a joint probability distribution model for wind power and photovoltaic power. Specifically, this step, "fitting historical renewable energy output data with a probability distribution function to generate a joint probability distribution model for wind and solar power," is a key uncertainty modeling step in constructing a two-stage stochastic scheduling method for integrated energy systems. Its technical implementation principle is based on statistical modeling and stochastic optimization theory, aiming to accurately characterize the joint stochastic characteristics of wind and solar power output in time and space, providing a reliable probabilistic basis for subsequent scheduling decisions.

[0025] In some implementations, this step first extracts hourly or 15-minute power output samples for wind and solar power from historical meteorological data and actual power output data. The sample length is usually no less than 3 years to ensure statistical significance. Subsequently, non-parametric or parametric methods are used to fit the marginal distributions of wind and solar power output. For wind power output, a Weibull distribution or a Beta distribution is usually used for fitting, and its shape parameter (k) and scale parameter (λ) are determined by maximum likelihood estimation (MLE) or least squares (LS). For solar power output, since it is significantly affected by solar radiation intensity, a Beta distribution or a mixed normal distribution is often used for modeling to reflect its asymmetric characteristics of high output during the day and zero output at night.

[0026] Furthermore, to construct a joint probability distribution model for wind and solar power, a Copula function needs to be introduced to model the correlation between the two. In practical applications, this step is typically performed during the preprocessing stage of inter-day scheduling, and is suitable for multi-energy coupled systems of wind, solar, storage, hydrogen, and heat within industrial parks. Through the joint probability distribution model, the scheduling system can generate probabilistic scenario-based optimization decisions during the day-ahead phase and make rolling corrections based on actual output during the real-time phase, thereby improving scheduling robustness and economy.

[0027] The technical effect of this step is that, through high-precision probabilistic modeling, the uncertainty of new energy output can be effectively quantified, providing a reliable basis for scenario generation for two-stage stochastic planning, thereby improving the scheduling efficiency and low-carbon operation level of the park's integrated energy system in the context of participating in the spot market.

[0028] S12, based on the Monte Carlo sampling method, generates multiple uncertainty scenarios from the joint probability distribution model and assigns corresponding occurrence probability weights to each scenario.

[0029] Specifically, this step, based on the Monte Carlo sampling method, generates multiple uncertainty scenarios from the joint probability distribution model and assigns corresponding probability weights to each scenario. This is a crucial step in realizing the cross-day two-stage stochastic scheduling of the park's integrated energy system. In some implementations, this step first constructs a joint probability distribution model of new energy output and load power. Typically, a Copula function or a multivariate Gaussian mixture model (GMM) is used to statistically model historical data to capture the nonlinear correlations between different variables. For example, based on wind speed, solar intensity, and load data from the past year, the data can be preprocessed using the K-means clustering algorithm to extract typical operating patterns, and then the joint probability distribution function can be fitted using maximum likelihood estimation or Bayesian methods.

[0030] In the Monte Carlo sampling process, efficient sampling techniques such as Latin hypercube sampling (LHS) or importance sampling can be selectively employed to improve the representativeness and computational efficiency of scenarios with a limited sample size. Typically, 50 to 200 uncertainty scenarios are generated, each representing a possible combination of renewable energy output and load. Furthermore, the probability weight of each scenario can be calculated using the integral value of the probability density function (PDF) over the corresponding region or the sampling frequency, ensuring that scenarios in high-probability regions have a greater decision-making influence in the optimization process.

[0031] In practical applications, this step is typically deployed during the day-ahead scheduling phase to generate a set of uncertainties covering the next 24 to 168 hours, supporting the robustness and economic optimization of cross-day scheduling plans. From a technical perspective, by introducing uncertainties with probability weights, the impact of renewable energy fluctuations and load changes on scheduling strategies can be effectively quantified, improving the adaptability and risk control level of scheduling schemes in the spot market environment, thereby achieving the low-carbon and economical operation goals of the park's integrated energy system.

[0032] S2 establishes a two-stage stochastic programming model spanning the day. The decision variables in the first stage include the weekly demand declaration value and the start-up and shutdown plan of slow equipment. The decision variables in the second stage include the power purchased and sold in the real-time market during the day and the charging and discharging power of energy storage equipment.

[0033] Specifically, this step involves establishing a two-stage stochastic programming model spanning two days for the optimal dispatch of an Integrated Energy System (IES) participating in the electricity spot market. This model improves the economic efficiency and low-carbon nature of system operation by dividing the decision-making process into two stages, addressing long-term (weekly) and short-term (intraday real-time) uncertainties respectively.

[0034] At the technical implementation level, the first-stage decision variables include weekly demand declarations and start-up / shutdown plans for slow-moving equipment (such as H-CHP and electrohydrogen generation). Determining demand declarations requires combining park load forecasts, historical electricity consumption data, and market demand cost structures, typically optimized on a weekly cycle to minimize capacity or demand costs. Start-up / shutdown plans for slow-moving equipment must consider minimum start-up / shutdown times, ramp rate limits (such as maximum upward / downward ramp rates), and start-up / shutdown costs (such as startup and shutdown costs) to ensure continuous and economical operation. The second-stage decision variables include the power purchased and sold in the intraday real-time market, as well as the charging and discharging power of energy storage devices (electric energy storage, hydrogen energy storage, and thermal energy storage). These variables need to be dynamically adjusted based on the first-stage decision results and actual renewable energy output and load fluctuations to minimize power purchase costs, curtailment costs, and reserve costs.

[0035] At the parameter level, the model incorporates several key parameters, such as the predicted value of new energy power generation, the predicted value of load power, the day-ahead and real-time market purchase and sale prices of electricity, the demand cost threshold (usually taken as 1.05), the curtailment penalty coefficient, equipment efficiency (such as the efficiency of electro-hydrogen production and the H-CHP electro-thermal ratio), the upper and lower limits of the SOC of energy storage equipment, and the charge and discharge efficiency. These parameters need to be reasonably set based on the actual equipment parameters of the industrial park, market rules, and historical operating data to enhance the practicality and accuracy of the model.

[0036] At the application level, this model is suitable for integrated energy systems with multi-energy coupling characteristics and participation in the electricity spot market, such as industrial parks and microgrids. Through cross-day scheduling, the system can optimize demand reporting and slow-speed equipment operation plans on a weekly scale, while addressing fluctuations in renewable energy output and load uncertainties on an intraday scale, achieving multi-time-scale collaborative optimization.

[0037] The technical effect of this step is that by introducing a two-stage stochastic programming method, it can effectively address the uncertainties caused by errors in new energy and load forecasting, improve the economic operation capability of the park's IES in the spot market environment and the local consumption level of new energy, and provide strong support for achieving the "dual carbon" goal.

[0038] Furthermore, S2 includes: S21, the slow-speed equipment includes H-CHP units and electro-hydrogen production equipment, whose start-up and shutdown schedules are fixed on a weekly scale and remain unchanged during daily scheduling.

[0039] Specifically, slow-operation equipment includes H-CHP (hydrogen fuel cell combined heat and power) units and electro-hydrogen production equipment, whose start-up and shutdown schedules are fixed on a weekly scale and remain unchanged during daily scheduling. The technical principle behind this step is based on the matching between the physical characteristics of the equipment and the scheduling cycle, aiming to reduce scheduling complexity and improve overall operational economy and stability.

[0040] In some implementations, H-CHP units and electrohydrogen production facilities are classified as slow-response equipment in weekly-scale scheduling due to their long start-up and shutdown response times (typically between 30 minutes and several hours), and the significant increase in equipment wear and operating costs caused by frequent start-ups and shutdowns. Their start-up and shutdown plans need to be predetermined during the inter-day scheduling phase to ensure that frequent adjustments are not needed during intraday scheduling, thereby reducing the impact of uncertainty on the scheduling results. Specifically, during the inter-day scheduling phase, a two-stage stochastic programming model is constructed based on the load forecast, renewable energy output forecast, and market electricity price information from the previous week. The optimal start-up and shutdown sequence is solved using optimization algorithms (such as Benders decomposition, scenario generation, and reduction techniques), and this sequence is used as a fixed input for intraday scheduling.

[0041] In practical industrial park integrated energy systems (IES), this step is suitable for parks with high renewable energy penetration and hydrogen storage capabilities. By scheduling the start-up and shutdown of fixed H-CHP and electro-hydrogen production equipment, the inter-day scheduling of multiple energy flows such as electricity, heat, and hydrogen can be effectively coordinated, improving the local consumption capacity of renewable energy and reducing the economic risks caused by demand-based electricity charges.

[0042] This step significantly improves the solution efficiency and robustness of the scheduling model, reduces the computational burden of intraday scheduling, and enhances the system's adaptability to capacity fees and new energy fluctuations through weekly-scale optimization decisions, providing key technical support for industrial parks to achieve low-carbon and economical operation.

[0043] S22, the power purchased and sold is modeled separately in the day-ahead market and the real-time market, and dynamic adjustments are allowed in the real-time market based on the actual output of new energy sources and fluctuations in electricity prices.

[0044] Specifically, this step involves the integrated energy system modeling the power purchased and sold in both the day-ahead and real-time markets when participating in the spot market, and dynamically adjusting the power output in the real-time market based on actual renewable energy output and electricity price fluctuations. The technology is implemented using a two-stage stochastic programming framework, aiming to improve the economic efficiency and renewable energy absorption capacity of the industrial park across multiple time scales.

[0045] At the technical implementation level, the day-ahead market modeling is based on forecast data, including load forecasts, renewable energy output forecasts, and market electricity price forecasts. The day-ahead power purchase and sale plan is determined through model optimization. The real-time market, based on the day-ahead plan, dynamically adjusts the plan according to actual output deviations and real-time electricity price fluctuations. Specifically, adjustments to the real-time market power purchase and sale must meet tie-line power constraints (Equation 26-34) and introduce 0-1 variables to distinguish between power purchase and sale states, thereby enabling flexible switching of power direction. In the real-time phase, if renewable energy output exceeds the day-ahead forecast, the system can sell excess electricity back to the grid; if output is insufficient, it is supplemented through energy storage or grid purchase.

[0046] In application scenarios, this step is suitable for industrial parks participating in the electricity spot market, especially given the high penetration rate of renewable energy, large load fluctuations, and the need to consider capacity fees, where it has significant dispatch optimization value. Through day-ahead to real-time phased modeling, the system can meet grid dispatch requirements while achieving efficient utilization of renewable energy and minimizing operating costs.

[0047] By dynamically adjusting the power purchased and sold, the uncertainty of new energy output and electricity price fluctuations can be effectively addressed, improving the flexibility and economy of system operation. At the same time, it enhances the park's responsiveness and profitability in the spot market, which is a key supporting link for realizing the cross-day two-stage random dispatch method.

[0048] S3. Based on the two-stage model, solve for the optimal scheduling scheme to minimize the overall operating cost, which includes demand electricity cost, electricity purchase cost, renewable energy curtailment cost, unit start-up and shutdown cost, and reserve penalty cost.

[0049] Specifically, in some implementations, the optimal scheduling scheme is solved based on the two-stage stochastic scheduling model, aiming to minimize the overall operating cost of the integrated energy system during the inter-day operating cycle. This operating cost comprehensively considers demand electricity costs, electricity purchase costs, renewable energy curtailment costs, unit start-up and shutdown costs, and reserve penalty costs, thereby achieving synergistic optimization of economic efficiency and low carbon emissions. At the technical implementation level, this step typically employs stochastic programming methods, combining day-ahead scheduling (the first stage) with real-time scheduling (the second stage) to address the uncertainties in renewable energy output and load power. The decision variables in the first stage include unit start-up and shutdown plans and day-ahead market electricity purchase and sale, which need to be predetermined within the inter-day scheduling cycle. The second stage adjusts the real-time market electricity purchase and sale, energy storage charging and discharging strategies, and reserve resource scheduling based on actual output and load deviations to reduce operational risks and cost fluctuations.

[0050] At the parameter level, demand-based electricity charges are calculated by comparing the actual peak power of the industrial park with the pre-declared maximum demand. If the actual peak power exceeds k times the contractually agreed amount (k is usually taken as 1.05), a penalty price will be applied. Purchase electricity charges are calculated by weighting the day-ahead and real-time market time-of-use prices. The price data must comply with spot market trading rules, such as the real-time price standards published by the China Power Exchange Center. The cost of curtailed renewable energy is calculated by multiplying the curtailed power by a unit curtailment penalty coefficient, which is usually set according to local policies, such as 0.2~0.5 yuan / kWh. Unit start-up and shutdown costs are determined based on the equipment type and start-up / shutdown energy consumption model. For example, the start-up cost of an electrohydrogen production unit may include preheating energy consumption and losses due to system response delays.

[0051] At the application level, this step is suitable for the cross-day scheduling optimization of integrated energy systems (IES) in industrial parks, and has significant advantages, especially in scenarios with high renewable energy penetration, sufficient energy storage, and participation in the electricity spot market. Through cross-day scheduling, the system can rationally arrange unit start-up and shutdown on a weekly timescale, reducing losses caused by frequent start-ups and shutdowns, while improving the local consumption capacity of renewable energy and reducing dependence on purchased electricity.

[0052] In terms of technical effectiveness, this step, by introducing a two-stage stochastic programming model, effectively addresses the uncertainties caused by errors in renewable energy output and load forecasting, reducing operational risks and economic losses due to forecast deviations. Simultaneously, by comprehensively optimizing demand, power purchase, curtailment, start-up, shutdown, and reserve costs, it enhances the overall economic efficiency and stability of the park's operation, providing key technical support for building a low-carbon, efficient, and flexible integrated energy system.

[0053] Furthermore, S3 includes: S31 introduces a penalty coefficient k in the calculation of demand electricity charges. When the actual peak power exceeds the pre-declared maximum demand, the excess part is charged at k times the unit demand price.

[0054] Specifically, a penalty coefficient is introduced into the calculation of demand-based electricity charges. This is an important economic modeling method used in the present invention to optimize the operating costs of the integrated energy system in a park. The core technical principle of this step is to set a punitive pricing mechanism in the demand-based electricity pricing model, imposing additional economic penalties on users whose actual peak power exceeds their pre-declared maximum demand. This guides users to declare their demand reasonably, optimizes the overall power dispatch strategy of the park, and reduces operating costs caused by demand deviations.

[0055] Furthermore, this step is applicable in practical scenarios where industrial parks participate in the spot market, especially in cross-day scheduling on a weekly timescale, where demand electricity costs, as a fixed cost item, have a significant impact on total operating costs. This is addressed by introducing a penalty coefficient. During the optimization process, the system can effectively avoid high penalty costs caused by users underestimating demand, thereby improving the robustness and economy of the scheduling strategy.

[0056] This penalty mechanism not only enhances the accuracy of user-reported demand, but also promotes refined management of load forecasting and renewable energy output in the park during the day-ahead dispatch phase, thereby improving overall energy utilization efficiency and market responsiveness, and providing key support for achieving low-carbon and economical integrated energy system operation.

[0057] S32, the backup penalty cost is used to perform punitive optimization on scheduling schemes that fail to meet backup requirements by introducing a small positive penalty coefficient.

[0058] Specifically, in the method of this invention, the introduction of reserve penalty cost is achieved by setting a small positive penalty coefficient in the objective function to penalize scheduling schemes that fail to meet reserve requirements, thereby enhancing the robustness and reliability of system operation in the two-stage stochastic scheduling model spanning two days. The technical implementation principle of this step is based on stochastic optimization theory. By introducing a penalty term for reserve capacity in the first-stage decision-making, the optimization model is guided to reserve sufficient reserve resources in the day-ahead scheduling stage to cope with the uncertainties caused by errors in new energy output and load forecasting in the second stage.

[0059] In practical terms, the reserve penalty cost is typically added to the objective function as a penalty coefficient multiplied by the reserve gap. This penalty coefficient is generally between 0.01 and 0.1, and the specific value can be adjusted based on the system's sensitivity to reserve reliability. For example, in industrial parks with high renewable energy penetration, to improve the system's tolerance to output fluctuations, the penalty coefficient can be appropriately increased to 0.05 to 0.1 to strengthen the penalty for insufficient reserves. The reserve gap is defined as the difference between the actual reserve capacity required by the system during the real-time scheduling phase and the reserve capacity reserved for day-ahead scheduling. Its calculation needs to consider the new energy forecasting error, load fluctuation range, and the adjustment capability of the energy storage system.

[0060] This step plays a crucial role in the inter-day scheduling model, particularly in handling the intermittency of renewable energy sources and load uncertainty. Through a penalized optimization mechanism, the system can pre-configure start-up and shutdown plans for multi-energy coupling devices such as electro-hydrogen production and H-CHP during the day-ahead scheduling phase, ensuring sufficient regulation capacity in real-time. In practical applications, this method is suitable for integrated energy systems in industrial parks participating in the spot market, especially in two-part tariff environments with high demand tariffs, helping to reduce additional costs caused by excessively high power peaks or insufficient reserves.

[0061] From a technical perspective, the setting of backup penalty costs effectively improves the robustness of the scheduling scheme, reduces operational risks caused by prediction bias, and achieves a better balance between economy and low carbon emissions. This step demonstrates the innovation and practicality of this invention in multi-timescale collaborative optimization and uncertainty handling.

[0062] S4. Based on tie-line power constraints and the dynamic response characteristics of multi-energy coupled devices, the scheduling scheme is subjected to feasibility verification and rolling optimization adjustment.

[0063] Specifically, this step, "based on tie-line power constraints and the dynamic response characteristics of multi-energy coupling equipment, performs feasibility verification and rolling optimization adjustment of the scheduling scheme," is a key link in realizing the cross-day two-stage stochastic scheduling method for the integrated energy system (IES) of the park. Its technical implementation principle integrates power system operation constraint analysis and multi-energy system dynamic response modeling, aiming to improve the robustness and economy of the scheduling scheme in actual operation.

[0064] At the application level, this step is suitable for cross-day dispatch scenarios in which industrial parks participate in the electricity spot market, especially in industrial and commercial parks with high renewable energy penetration, complex energy storage configurations, and demand-price sensitivity, where it has significant optimization value. By monitoring tie-line power and equipment operating status in real time, the system can continuously adjust dispatch strategies within the day, improving the local consumption rate of renewable energy and the system's economic efficiency.

[0065] The technical benefits of this step are that it effectively addresses the errors in new energy output and load forecasting, ensures the feasibility of the dispatching scheme under both physical and market constraints, and reduces the cost of curtailment and unit start-up and shutdown through dynamic optimization, thereby improving the stability and economy of the park's IES operation and providing technical support for achieving low-carbon, efficient, and flexible energy dispatching.

[0066] Also includes: S5. Based on the power constraints of the tie line and the dynamic response characteristics of the multi-energy coupling equipment, the scheduling scheme is continuously optimized and adjusted. At the end of each scheduling cycle, the model parameters are updated based on the latest new energy output data and load forecast, and the optimization problem for the subsequent scheduling cycle is solved again.

[0067] Specifically, this step involves a rolling optimization scheduling mechanism based on tie-line power constraints and the dynamic response characteristics of multi-energy coupled equipment. It is a core component for realizing cross-day two-stage stochastic scheduling of the integrated energy system (IES). At the end of each scheduling cycle, the system dynamically updates key parameters in the optimization model by collecting the latest renewable energy output data (such as actual wind and solar power generation) and load forecast information. These parameters include renewable energy output forecasts, load demand curves, and the SOC status of energy storage devices, thereby enabling a re-solution of the optimization problem for subsequent scheduling cycles.

[0068] At the technical implementation level, this rolling optimization process adopts a two-stage stochastic programming framework. The first stage decision variable is the pre-determined unit start-up and shutdown plan during cross-day scheduling. The second stage involves intraday rolling adjustments based on real-time uncertainties (such as renewable energy output deviations and load fluctuations). Specifically, at the end of each scheduling cycle (e.g., 1 hour), the system updates the scenario probability distribution of renewable energy output based on real-time data and, combined with the dynamic response models of multi-energy coupled equipment (such as hydrogen production by electricity, H-CHP, and thermal power units) (e.g., ramp-up rate, start-up / shutdown delays, efficiency changes, etc.), reconstructs the constraints and objective function of the optimization problem. For example, the response time of hydrogen production by electricity is typically 10-30 minutes, and its efficiency exhibits non-linear characteristics with load changes, requiring the introduction of an efficiency function or piecewise linear approximation into the model.

[0069] At the application level, this step is suitable for cross-day dispatch scenarios in which industrial parks participate in the electricity spot market, especially given the high penetration rate of renewable energy, complex energy storage configurations, and demand-price sensitivity. Through rolling optimization, the system can dynamically respond to fluctuations in renewable energy output and load changes, improving energy utilization efficiency and market responsiveness.

[0070] The technical benefits of this step are that it significantly improves the robustness and economy of the scheduling scheme. Through real-time data-driven model parameter updates and dynamic response modeling of multi-energy devices, the system can achieve reasonable allocation of tie-line power under uncertain environments, reduce wind and solar curtailment losses, lower demand-based electricity costs, and improve the overall optimization level of operating costs.

[0071] The two-stage stochastic scheduling method for integrated energy systems in industrial parks, as described in this invention, effectively reduces the operating costs and carbon emissions of integrated energy systems in industrial parks, enhances the capacity for cross-day absorption of new energy sources, and enables collaborative optimization scheduling of multi-energy equipment in a spot market environment.

[0072] Example 2 The cross-day two-stage random scheduling method for the integrated energy system of industrial parks of this invention can be specifically achieved by designing the following function to realize the objective of this invention.

[0073] The objective function is as follows: The objective function for the park's cross-day optimization is to minimize the overall operating cost over a period of time (such as one week). When considering uncertainty, it is modeled as a two-stage stochastic programming problem, where the uncertain variables include renewable energy generation, load power, etc., as shown in equation (1): (1) In the formula: For demand-based electricity pricing, To purchase electricity, For the cost of curtailing renewable energy, Costs for unit start-up and shutdown; The set consists of all scheduling moments, and the specific expressions for each part are as follows. Depending on different market rules and operational requirements, the specific expressions for each part in equation (1) may vary slightly: (2) (3) (4) (5) In the formula: superscript t Represents the current scheduling time. , These are the current spot market electricity purchase and sales prices, respectively. , These are the purchase and sale prices of electricity in the real-time spot market, respectively. For parks that do not participate in the spot market and purchase electricity only based on time-of-use pricing or a single electricity price, the real-time power output can be considered to be 0. , , , These refer to the purchased and sold electricity volumes in the day-ahead and real-time markets, respectively. , These are the unit demand prices within and beyond the contract, respectively. k The threshold for imposing punitive prices is typically set at 1.05; This represents the actual peak power. The maximum pre-declared demand is considered a constant during the cross-day scheduling phase; The penalty coefficient for curtailment of renewable energy. This refers to the power curtailment of renewable energy sources. It is a collection of all equipment (i.e., electric hydrogen production, H-CHP, and electric heating units). , respectively equipment i Startup and shutdown costs , respectively equipment i Startup / shutdown variables; To minimize the small positive penalty coefficient added to the reserve, the first-stage decision variables are those that need to be predetermined on a weekly timescale, including unit start-up and shutdown variables for which start-up and shutdown plans need to be determined on a pre-week timescale. Cross-day scheduling yields start-up and shutdown plans for slow-speed units.

[0074] The constraints are as follows: The constraints of the cross-day optimal scheduling problem include: Constraints on new energy output: (7) (8) In the formula: This refers to the collection of all wind turbines and photovoltaic units. For new energy power generation equipment i For predicted output, when the prediction accuracy is not high, it can also be based on historical data fitting for cross-day scheduling on a weekly time scale.

[0075] Multi-energy coupling device constraints: (9) (10) (11) (12) (13) (14) (15) In the formula: To indicate the unit i time t A 0-1 variable representing the start / stop state; , Representing the equipment i Maximum upward and downward ramp rates during normal operation; These represent devices i At any moment t The 0-1 variables of whether it is in the process of starting or shutting down are associated with the unit start-up and shutdown variables in equations (39)-(40); , These are the minimum and maximum electrical power of the unit, respectively. , , The efficiencies of the electric hydrogen production, electric heat production, and H-CHP units are respectively: This refers to the calorific value of hydrogen. This refers to the mass flow rate of hydrogen at the outlet of the electrogenerated hydrogen production plant. The heating capacity of the electric heating unit. , , The power supply, heating capacity, and inlet hydrogen mass flow rate of the H-CHP unit; The feasible range of electro-thermal power for H-CHP units can be modeled as a constant heat-to-power ratio, depending on the actual unit type.

[0076] Constraints of energy storage devices: (16) (17) (18) (19) In the formula: , These are the discharge and charge power, respectively. For a moment t The state of charge (SOC). , These are the maximum and minimum values ​​of SOC, respectively. , These represent charging and discharging efficiencies, respectively. The time interval between adjacent scheduling times is given by equations (16)-(22). These equations represent the operational constraints of the energy storage system.

[0077] Constraints of hydrogen energy storage devices:

[0078]

[0079]

[0080]

[0081] In the formula, , These represent the standard volume flow rates of hydrogen at time t, respectively. For a moment t Standard volume of hydrogen in a hydrogen storage device , These are the minimum and maximum permissible standard volumes for hydrogen storage devices. This represents the time interval between adjacent scheduling moments.

[0082] Constraints of thermal energy storage equipment:

[0083]

[0084]

[0085]

[0086] In the formula, , These represent the heat storage and heat release power of the heat storage tube at time t, respectively. Let the heat stored at time t be... , These represent the maximum and minimum values ​​of heat storage, respectively. This represents the time interval between adjacent scheduling moments.

[0087] a) Energy balance constraint: (twenty three)

[0088] (25) In the formula: , The respective park load times t The electrical and thermal power consumed, equations (23)-(25) describe the energy balance of electricity, heat and hydrogen, respectively.

[0089] Tie line power-related constraints: (26) (27) (28) (29) (30) (31) (32) (33) (34) In the formula: , , They are respectively t Power at the time of day, market power, and real-time market power are all positively correlated with power flowing into the park; , The 0-1 variables are introduced to handle the different day-ahead and real-time electricity purchase and sales prices, and are used to distinguish whether the current time is in the electricity purchase state or the electricity sales state. M It is a very large positive number; , These are the lower and upper limits of the tie line power, respectively.

[0090] The cross-day two-stage random scheduling method for integrated energy systems in industrial parks according to embodiments of the present invention can effectively improve the renewable energy absorption capacity of integrated energy systems in industrial parks on a cross-day time scale, reduce demand electricity costs and overall operating costs, and achieve synergistic optimization of economic efficiency and low carbon emissions.

[0091] Example 3 To achieve the above embodiments, such as Figure 2 As shown, this embodiment also provides a cross-day two-stage random scheduling device 10 for the integrated energy system of an industrial park, including: The new energy fluctuation characteristic modeling module 100 is used to fit the output fluctuation characteristics of new energy power generation equipment on a cross-day time scale based on historical data, and to construct an uncertainty scenario set that includes new energy prediction errors. The two-stage model building module 200 is used to establish a cross-day two-stage stochastic programming model. The decision variables in the first stage include the weekly demand declaration value and the start-up and shutdown plan of slow equipment. The decision variables in the second stage include the power purchased and sold in the real-time market during the day and the charging and discharging power of energy storage equipment. The scheduling optimization solution module 300 is used to solve the optimal scheduling scheme based on the two-stage model to minimize the overall operating cost, which includes demand electricity cost, electricity purchase cost, renewable energy curtailment cost, unit start-up and shutdown cost, and reserve penalty cost. The scheduling scheme verification and optimization module 400 is used to perform feasibility verification and rolling optimization adjustment of the scheduling scheme based on tie line power constraints and the dynamic response characteristics of multi-energy coupled equipment.

[0092] Furthermore, the new energy fluctuation characteristic modeling module is also used for: A probability distribution function is used to fit historical renewable energy output data to generate a joint probability distribution model for wind power and photovoltaic power. Based on the Monte Carlo sampling method, multiple uncertainty scenarios are generated from the joint probability distribution model, and each scenario is assigned a corresponding probability weight.

[0093] Furthermore, the two-stage model building module is also used for: The slow-speed equipment includes H-CHP units and electro-hydrogen production equipment, whose start-up and shutdown schedules are fixed on a weekly scale and remain unchanged during daily scheduling. The power purchased and sold are modeled separately in the day-ahead market and the real-time market, and dynamic adjustments are allowed in the real-time market based on the actual output of new energy sources and electricity price fluctuations.

[0094] Furthermore, the scheduling optimization solution module is also used for: The demand charge calculation introduces a penalty coefficient k. When the actual peak power exceeds the pre-declared maximum demand, the excess portion is charged at k times the unit demand price. The backup penalty cost is used to penalize and optimize scheduling schemes that fail to meet backup requirements by introducing a small positive penalty coefficient.

[0095] Furthermore, it also includes: The rolling optimization adjustment module is used to perform rolling optimization adjustment on the scheduling scheme based on the tie line power constraints and the dynamic response characteristics of the multi-energy coupling equipment. At the end of each scheduling cycle, the model parameters are updated based on the latest new energy output data and load forecast, and the optimization problem for the subsequent scheduling cycle is solved again.

[0096] The two-stage random scheduling device for the integrated energy system of industrial parks according to embodiments of the present invention effectively reduces the operating costs and carbon emissions of the integrated energy system of industrial parks, enhances the capacity for cross-day absorption of new energy sources, and realizes the coordinated and optimized scheduling of multi-energy equipment in the spot market environment.

[0097] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0098] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A two-stage stochastic scheduling method for an integrated energy system in an industrial park, characterized in that, include: S1, based on historical data, fits the output fluctuation characteristics of new energy power generation equipment in the cross-day time scale, and constructs an uncertainty scenario set including new energy prediction errors; S2. Establish a two-stage stochastic programming model across days. The decision variables in the first stage include the weekly demand declaration value and the start-up and shutdown plan of slow equipment. The decision variables in the second stage include the power purchased and sold in the real-time market during the day and the charging and discharging power of energy storage equipment. S3. Based on the two-stage model, solve for the optimal scheduling scheme to minimize the overall operating cost, which includes demand electricity cost, electricity purchase cost, renewable energy curtailment cost, unit start-up and shutdown cost, and reserve penalty cost. S4. Based on tie-line power constraints and the dynamic response characteristics of multi-energy coupled devices, the scheduling scheme is subjected to feasibility verification and rolling optimization adjustment.

2. The method as described in claim 1, characterized in that, S1 includes: S11, a probability distribution function is used to fit the historical new energy output data to generate a joint probability distribution model of wind power and photovoltaic power; S12, Based on the Monte Carlo sampling method, generate multiple uncertainty scenarios from the joint probability distribution model, and assign corresponding occurrence probability weights to each scenario.

3. The method as described in claim 1, characterized in that, S2 includes: S21, the slow-speed equipment includes H-CHP units and electro-hydrogen production equipment, whose start-up and shutdown schedules are fixed on a weekly scale and remain unchanged during daily scheduling; S22, the power purchased and sold is modeled in the day-ahead market and the real-time market respectively, and dynamic adjustment is allowed in the real-time market based on the actual output of new energy and electricity price fluctuations.

4. The method as described in claim 1, characterized in that, The S3 includes: S31, A penalty coefficient k is introduced into the demand electricity charge calculation. When the actual power peak exceeds the pre-declared maximum demand, the excess part is charged at k times the unit demand price. S32, the backup penalty cost is used to perform punitive optimization on scheduling schemes that fail to meet backup requirements by introducing a small positive penalty coefficient.

5. The method as described in claim 1, characterized in that, Also includes: S5. Based on the power constraints of the tie line and the dynamic response characteristics of the multi-energy coupling equipment, the scheduling scheme is continuously optimized and adjusted. At the end of each scheduling cycle, the model parameters are updated based on the latest new energy output data and load forecast, and the optimization problem for the subsequent scheduling cycle is solved again.

6. A cross-day two-stage random scheduling device for an integrated energy system in an industrial park, characterized in that, include: The new energy fluctuation characteristic modeling module is used to fit the output fluctuation characteristics of new energy power generation equipment on a cross-day time scale based on historical data, and to construct an uncertainty scenario set that includes new energy prediction errors. The two-stage model building module is used to establish a cross-day two-stage stochastic programming model. The decision variables in the first stage include the weekly demand declaration value and the start-up and shutdown plan of slow equipment. The decision variables in the second stage include the power purchased and sold in the real-time market during the day and the charging and discharging power of energy storage equipment. The scheduling optimization solution module is used to solve the optimal scheduling scheme based on the two-stage model to minimize the overall operating cost, which includes demand electricity cost, electricity purchase cost, renewable energy curtailment cost, unit start-up and shutdown cost, and reserve penalty cost. The scheduling scheme verification and optimization module is used to perform feasibility verification and rolling optimization adjustments on the scheduling scheme based on tie-line power constraints and the dynamic response characteristics of multi-energy coupled devices.

7. The apparatus as claimed in claim 6, characterized in that, The new energy fluctuation characteristic modeling module is also used for: A probability distribution function is used to fit historical renewable energy output data to generate a joint probability distribution model for wind power and photovoltaic power. Based on the Monte Carlo sampling method, multiple uncertainty scenarios are generated from the joint probability distribution model, and each scenario is assigned a corresponding probability weight.

8. The apparatus as claimed in claim 6, characterized in that, The two-stage model building module is also used for: The slow-speed equipment includes H-CHP units and electro-hydrogen production equipment, whose start-up and shutdown schedules are fixed on a weekly scale and remain unchanged during daily scheduling. The power purchased and sold are modeled separately in the day-ahead market and the real-time market, and dynamic adjustments are allowed in the real-time market based on the actual output of new energy sources and electricity price fluctuations.

9. The apparatus as claimed in claim 6, characterized in that, The scheduling optimization solution module is also used for: The demand charge calculation introduces a penalty coefficient k. When the actual peak power exceeds the pre-declared maximum demand, the excess portion is charged at k times the unit demand price. The backup penalty cost is used to penalize and optimize scheduling schemes that fail to meet backup requirements by introducing a small positive penalty coefficient.

10. The apparatus as claimed in claim 6, characterized in that, Also includes: The rolling optimization adjustment module is used to perform rolling optimization adjustment on the scheduling scheme based on the tie line power constraints and the dynamic response characteristics of the multi-energy coupling equipment. At the end of each scheduling cycle, the model parameters are updated based on the latest new energy output data and load forecast, and the optimization problem for the subsequent scheduling cycle is solved again.

Images