A method for determining the optimal energy storage capacity of industrial parks based on load time-series characteristics
By using a method based on load time-series characteristics, and employing Monte Carlo methods and regression models to optimize energy storage capacity configuration, the problem of low computational efficiency in existing technologies is solved, achieving more accurate and efficient energy storage capacity configuration, which is suitable for industrial parks of different sizes and types.
Patent Information
- Application Number
- CN202611131666.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-29
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2046-07-29
AI Technical Summary
Existing technologies suffer from low computational efficiency and difficulty in adapting to rapid iterative optimization across multiple scenarios when configuring energy storage capacity. In particular, the randomness and volatility of production scheduling in industrial parks increase the difficulty of configuring energy storage capacity, and traditional methods struggle to balance the randomness of production scheduling with the efficiency of optimization solutions.
A method based on load time-series characteristics is adopted. Energy consumption curves under different scheduling scenarios are sampled using the Monte Carlo method to construct a regression model, extract the relationship between load time-series characteristics and optimal energy storage capacity, and optimize it using mixed integer linear programming (MILP) and combine it with a random forest model for fast calculation.
It achieves more accurate energy storage capacity configuration, improves the rationality and reliability of configuration schemes, significantly reduces computing resource consumption and time costs, is suitable for rapid iterative design in the multi-scheme comparison and planning stages, and has versatility and engineering promotion value.
Smart Images

Figure CN122656288B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for determining energy storage capacity, specifically a method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics, belonging to the field of power load forecasting and energy storage capacity optimization technology. Background Technology
[0002] Currently, distributed photovoltaic (PV) power generation in industrial parks is being promoted and applied on a large scale. However, there is often a significant time mismatch between PV power generation and the energy load of industrial park production. Therefore, configuring energy storage devices for power dispatch has become a key research topic in building renewable energy.
[0003] In current research on energy storage capacity configuration in industrial parks, mixed integer linear programming (MILP) is the mainstream solution method, which has advantages such as high solution accuracy and strong model interpretability. However, for energy storage optimization configuration scenarios with high-resolution load curves of 8760 hours per year and 15 minutes per day, the scale of decision variables in MILP models can reach hundreds of thousands, resulting in technical drawbacks such as excessively long solution time, extremely low computational efficiency, and difficulty in adapting to rapid iterative optimization in multiple scenarios.
[0004] Meanwhile, production in industrial parks is characterized by significant randomness and volatility. Enterprise scheduling is affected by various factors such as orders and production capacity, and the energy consumption schedule varies from week to week, further increasing the difficulty of energy storage capacity configuration. Existing methods are unable to balance the randomness of production scheduling with the efficiency of optimization solutions.
[0005] Therefore, in order to solve the above-mentioned technical problems, it is indeed necessary to provide an innovative method for determining the optimal energy storage capacity of industrial parks based on load time-series characteristics, so as to overcome the defects in the existing technology. Summary of the Invention
[0006] The purpose of this invention is to provide a method for determining the optimal energy storage capacity of industrial parks based on load time-series characteristics. This method starts from the time-series characteristics of energy consumption, samples energy consumption curves under different scheduling scenarios using the Monte Carlo method, and predicts the energy storage capacity through the regression relationship between the time-series characteristics and the optimal energy storage capacity, thereby providing strong guidance for the configuration of energy storage capacity in industrial parks.
[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics, comprising the following process steps: 1) Obtain basic data of the industrial park, including the rated power and fixed operating schedule of electrical equipment with fixed operating time, as well as the types of electrical equipment with variable operating time, rated power and corresponding operating rules; 2) Based on Monte Carlo scheduling sampling, construct a random sampling model for the production operation schedule of the industrial park; 3) Generate the electricity load curve for the industrial park based on Openstudio building energy consumption simulation software; 4) Generate the photovoltaic power generation curve of the industrial park based on the planned photovoltaic installed capacity; 5) Load time series feature extraction: Based on the load curve generated in step 3) and the photovoltaic power generation curve in step 4), calculate the load time series features of the park for k time periods, repeat the calculation until the features of all n load curves are completed; 6) Independently optimize and solve each set of photovoltaic power generation curves and industrial park electricity load curves, and solve the optimal energy storage capacity corresponding to each scenario in turn, and complete the calculation of all n scenarios; 7) Based on the load time sequence characteristics of step 5) and the optimal energy storage capacity of step 6), a regression fitting model is constructed to obtain a fast calculation model for energy storage capacity.
[0008] The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics of the present invention is further described as follows: In step 1), the basic data of the industrial park includes meteorological parameters of the industrial park location, geometric parameters of buildings within the park, thermal parameters of the building envelope, including parameters such as air conditioning cooling and heating set temperature, personnel density, personnel activity level and building fresh air volume, photovoltaic installed capacity, purchase and operation and maintenance costs of photovoltaic and energy storage equipment, and grid electricity purchase costs.
[0009] The method for determining the optimal energy storage capacity of an industrial park based on load time sequence characteristics of the present invention is further described as follows: In step 1), the operating rules for electrical equipment with variable operating time are as follows: Based on historical operating data, production plan data, equipment control records or energy consumption monitoring data, the operating time, operating duration, start-stop status, operating frequency, load rate or start-up coefficient of the equipment are statistically extracted, as well as the probability distribution parameters of these operating characteristics, and the probability of occurrence of various operating scenarios under different production organization modes, operating conditions or scheduling strategies.
[0010] The method for determining the optimal energy storage capacity of industrial parks based on load time sequence characteristics of the present invention is further described as follows: Step 2) specifically involves: based on the operating rules of the power-consuming equipment with variable operating time in step 1), randomly sampling the equipment activation coefficient and constraining its value within the range of [0,1]; simultaneously, discrete sampling is used to determine the weekly shift schedule, including four modes: one-shift, two-shift, three-shift, and three-shift with full-day weekend overtime; n sets of samples are generated through the above sampling method, each set of samples covering all weeks of the year, including the equipment activation coefficient and shift time type corresponding to each week.
[0011] The method for determining the optimal energy storage capacity of industrial parks based on load time-series characteristics of the present invention is further described in step 3): 3-1) Based on the building geometry parameters in step 1), complete the three-dimensional visualization modeling of the park buildings in the Openstudio building energy consumption simulation software, and input the thermal parameters of the building envelope into the model; 3-2) Input meteorological parameters, the rated power of electrical equipment with fixed running time and fixed operating time, and the rated power of electrical equipment with variable running time into the model of step 3-1) to construct a simulation model of the park's power load; 3-3), Substitute the kth group of annual random operating schedules into the park's electricity load simulation model in step 3-2) to perform energy consumption simulation, and output the annual electricity load curve with a time resolution of 8760 hours and 15 minutes. (3-4) Iterate through k = 1,2,...,n to complete n sets of load curve simulations, where n is the total number of timetables that need to be sampled.
[0012] The method for determining the optimal energy storage capacity of an industrial park based on load time sequence characteristics of the present invention is further described as follows: In step 4), the photovoltaic power generation at any time t = photovoltaic capacity × solar radiation at time t × module efficiency; wherein, photovoltaic capacity is the photovoltaic installed capacity of the park, solar radiation is the real-time effective solar radiation data in the meteorological parameters of the park site at the corresponding time, and module efficiency is the comprehensive working efficiency of the photovoltaic module. Based on the above formula, the photovoltaic power generation curve with a time resolution of 8760 hours and 15 minutes throughout the year is obtained, where time t is any time in the high time resolution time series data of 8760 hours and 15 minutes throughout the year.
[0013] The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics of the present invention is further described in step 5), where the load time-series characteristics of the park specifically include ten characteristics: annual maximum load, annual average slope, peak and off-peak season difference coefficient, weekend load ratio, average daily skewness, average daily peak, average daily Shannon entropy, nighttime load ratio, photovoltaic direct absorption rate, and maximum photovoltaic surplus.
[0014] The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics of the present invention is further described in step 6): Specifically, with minimizing the average electricity cost of the park as the optimization objective, the purchase and maintenance costs of photovoltaic and energy storage equipment, and the electricity purchase cost of the power grid are input. The mixed integer linear programming (MILP) method is used to independently optimize and solve each set of photovoltaic power generation curves and industrial park electricity load curves, and the optimal energy storage capacity corresponding to each set of scenarios is solved in turn, and the calculation of all n sets of scenarios is completed.
[0015] The method for determining the optimal energy storage capacity of industrial parks based on load time-series characteristics of the present invention further includes: step 7) specifically: 7-1), the load time series characteristics of each load group obtained in step 5) are used as input, and the optimal energy storage capacity obtained in step 6) is used as a one-to-one paired sample dataset of tags. 7-2) Based on the constructed paired sample dataset, a random forest model is used as the fitting model to fit the correlation between load time series characteristics and optimal energy storage capacity. The intrinsic mapping relationship between load time series fluctuation characteristics, time series distribution law and optimal energy storage capacity is quantified, and finally a fast energy storage capacity calculation model is obtained.
[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention uses probabilistic modeling and Monte Carlo sampling of various scheduling modes in industrial parks to fully consider the engineering reality of non-fixed scheduling and temporal randomness of load in actual production. It overcomes the configuration deviation problem caused by traditional energy storage configuration methods that only use fixed typical day or fixed year load curves. It can more realistically reflect the potential energy consumption scenarios of the park in the future, making the energy storage capacity configuration results more in line with actual operation needs and improving the rationality and reliability of the configuration scheme.
[0017] 2. To address the issues of large variable scale and excessively long solution time in traditional MILP time series optimization with a resolution of 15 minutes throughout the year, this invention adopts a technical approach of first sampling optimization and then feature fitting. This transforms a large number of repetitive high-precision optimization calculations into a single offline training and rapid online prediction, significantly reducing the number of times the MILP solver is repeatedly called, significantly reducing computational resource consumption and time costs, enabling rapid decision-making on energy storage capacity, and is suitable for multi-scheme comparison, sensitivity analysis, and rapid iterative design in the planning stage.
[0018] 3. This invention does not rely on historical energy consumption data. It only needs to be based on building parameters, equipment parameters, and scheduling probability distribution to complete energy consumption simulation and energy storage configuration. It can be used for energy storage transformation and operation optimization of existing industrial parks, as well as for preliminary scheme design of undeveloped industrial parks in the planning stage. At the same time, the method is not limited to specific building forms and industry types. It can be adapted to industrial parks of different production processes and sizes, and has strong versatility and engineering promotion value.
[0019] 4. This invention adopts a modular design approach, where time-series feature extraction, objective function optimization, and regression fitting models can all be flexibly replaced and extended according to actual needs: load time-series features can be added or removed based on the scenario; different regression algorithms or optimization solvers can be used instead. The overall method is robust, highly scalable, and adaptable to different energy market mechanisms and engineering requirements, possessing broad applicability.
[0020] 5. This invention can not only provide park owners with the optimal energy storage planning scheme for future scheduling uncertainties, but also analyze the optimal capacity and electricity cost corresponding to different scheduling modes, provide scheduling strategy optimization guidance for industrial parks with existing energy storage systems, and realize two-way collaborative optimization of "energy storage configuration guiding production scheduling and production scheduling feeding back into energy storage operation", thereby improving the overall economic efficiency of the park. Attached Figure Description
[0021] Figure 1 This is a flowchart of the method for determining the optimal energy storage capacity of industrial parks based on load time sequence characteristics, as proposed in this invention.
[0022] Figure 2 This is a graph showing the daily power generation of the photovoltaic power generation curve obtained in step 4) of this invention.
[0023] Figure 3 This is the regression fitting graph constructed based on load time sequence characteristics and optimal energy storage capacity in step 7 of the present invention. Detailed Implementation
[0024] This invention provides a method for determining the optimal energy storage capacity of industrial parks based on load time-series characteristics, comprising the following process steps: 1) Obtain basic data of the industrial park, including the rated power and fixed operating schedule of electrical equipment with fixed operating time, as well as the types of electrical equipment with variable operating time, rated power, and corresponding operating rules.
[0025] Specifically, the basic data for the industrial park includes meteorological parameters of the park's location, geometric parameters of buildings within the park, thermal parameters of the building envelope, including parameters such as air conditioning cooling and heating setpoints, personnel density, personnel activity levels, and building fresh air volume, photovoltaic installed capacity, purchase and maintenance costs of photovoltaic and energy storage equipment, and electricity purchase costs from the grid. The geometric parameters of the buildings within the park include building floor plans and the height of each building to facilitate the generation of 3D building models. The thermal parameters of the building envelope are calculated based on the material composition and thickness of the building envelope in the construction drawings for use in building thermal simulation. Parameters such as air conditioning cooling and heating setpoints, personnel density, personnel activity levels, and building fresh air volume are used to support building energy consumption simulation.
[0026] This embodiment takes an industrial park as an example. The thermal coefficient of the building envelope is based on the "Unified Standard for Energy-Saving Design of Industrial Buildings" GB 51245-2017. The building area includes 10% cleanroom, 50% general production room, 15% ambient temperature warehouse, 5% cool warehouse, 5% cold storage, and 15% office building.
[0027] The basic parameters for simulating the energy consumption of various types of buildings are shown in Table 1.
[0028] Table 1 Basic parameters for energy consumption simulation of various types of buildings
[0029] Among them, the ambient temperature warehouse, cool warehouse, cold warehouse and office building use a fixed schedule. The air conditioning in each warehouse runs from 0:00 to 24:00 every day throughout the year. The office building only runs from Monday to Friday. Its workday schedule is shown in Table 2.
[0030] Table 2 Fixed Schedule Settings
[0031] Furthermore, the operating rules for the variable-running-time electrical equipment are as follows: based on historical operating data, production plan data, equipment control records or energy consumption monitoring data, statistically extract the equipment running time, running duration, start-stop status, running frequency, load rate or start-up coefficient, as well as the probability distribution parameters of these operating characteristics, and the probability of occurrence of various operating scenarios under different production organization modes, operating conditions or scheduling strategies.
[0032] In this embodiment, the equipment with variable operating schedules is the equipment in the clean room and general production room. Actual operating coefficients were obtained through surveys and statistical calculations were performed. The sample data fitting results of the weekly equipment activation coefficients approximately follow a Beta (4.2, 5.8) distribution. Therefore, during the simulation process, a large number of samples were taken based on this distribution to generate multiple sets of differentiated operating schedules to simulate actual operating conditions.
[0033] Based on a survey of the park's actual work schedule, weekly work hours are divided into four categories: single shift: workers work only from 8:00 to 16:00 on weekdays; two shifts: workers work in two time slots on weekdays, 8:00-16:00 and 16:00-0:00; three shifts: workers work in all three time slots, 8:00-16:00, 16:00-0:00, and 0:00-8:00, i.e., working all day on weekdays; and full-week work: workers work in all three time slots, 8:00-16:00, 16:00-0:00, and 0:00-8:00, on both weekdays and weekends.
[0034] Meanwhile, the value of the equipment activation coefficient has certain limitations on the extraction of working time: when the weekly equipment activation coefficient is <0.3, a single shift is mandatory; when it is 0.3~0.6, a single or two-shift system is allowed; when it is >0.6, a two-shift or three-shift system is allowed. If a three-shift system is selected, it is further determined whether it is a full-week work.
[0035] 2) Based on Monte Carlo scheduling sampling, a random sampling model for the production operation schedule of the industrial park is constructed.
[0036] This step specifically involves: based on the variable operating time rules of electrical equipment from step 1), randomly sampling the equipment activation coefficients and constraining their values within the range [0,1]; simultaneously, discrete sampling is used to determine the weekly shift schedule, including four modes: one-shift, two-shift, three-shift, and three-shift with full-day weekend overtime. n sets of samples are generated through the above sampling method, each covering all weeks of the year, including the corresponding equipment activation coefficients and shift schedule types for each week.
[0037] This embodiment uses the week as the basic time unit and performs independent sampling for each of the 52 weeks of the year. This sampling method generates 100 sample groups, each covering all weeks of the year and including the corresponding equipment activation coefficient and shift schedule type for each week.
[0038] 3) Based on OpenStudio building energy consumption simulation software, generate the electricity load curve for the industrial park, which includes the following sub-steps: 3-1) Based on the building geometry parameters in step 1), complete the three-dimensional visualization model of the park buildings in the Openstudio building energy consumption simulation software, and input the thermal parameters of the building envelope into the model.
[0039] 3-2) Further input meteorological parameters, the rated power and fixed operating schedule of electrical equipment with fixed operating time, and the rated power of electrical equipment with variable operating time into the model of step 3-1) to construct a simulation model of the park's power load. Among them, the meteorological parameters are typical annual meteorological data from CSWD.
[0040] 3-3), Substitute the kth group of the annual random operating schedule into the park's electricity load simulation model in step 3-2) to perform energy consumption simulation, and output the annual electricity load curve with a time resolution of 8760 hours and 15 minutes.
[0041] 3-4), iterate through k = 1,2,...,n to complete n sets of load curve simulations, where n is the total number of time schedules to be sampled, which can be determined according to the actual accuracy requirements and computational load. In this embodiment, it is taken as 1000.
[0042] 4) Generate the photovoltaic power generation curve of the industrial park based on the planned photovoltaic installed capacity.
[0043] In this step, the photovoltaic power generation at any time t = photovoltaic capacity × solar radiation at time t × module efficiency; where photovoltaic capacity is the photovoltaic installed capacity of the park, solar radiation is the real-time effective solar radiation data in the meteorological parameters of the park at the corresponding time, and module efficiency is the overall working efficiency of the photovoltaic module.
[0044] Based on the above formula, the photovoltaic power generation curve with a time resolution of 8760 hours and 15 minutes throughout the year is obtained, where time t is any time in the high time resolution time series data of 8760 hours and 15 minutes throughout the year.
[0045] In this embodiment, the module efficiency is taken as 0.215, which is the overall operating efficiency of the photovoltaic module. The daily power generation of the simulated photovoltaic power generation curve is shown in the attached figure. Figure 2 As shown.
[0046] Furthermore, this embodiment can introduce a production-to-consumption ratio constraint, namely the ratio of the annual power generation of the photovoltaic system to the total annual electricity load of the buildings in the park, to limit the matching ratio between the photovoltaic power generation volume and the electricity demand in the park.
[0047] The utilization ratio of the photovoltaic system in the park is set at 0.4 to reverse-calculate the actual installed capacity that the photovoltaic system in the park can adapt to. The specific calculation logic is as follows: Based on the electricity load curves corresponding to the operating schedules obtained by the aforementioned n sets of random sampling, the total annual electricity load of the park under multiple operating conditions is statistically obtained; combined with the hourly output calculation results of photovoltaics, the annual power generation corresponding to the unit photovoltaic installed capacity is calculated; finally, based on the preset utilization ratio of 0.4, the actual installed capacity of the photovoltaic system adapted to the electricity consumption characteristics of the park is calculated in reverse iteration through the coupled calculation of the total annual electricity load of the kth electricity load curve, the annual power generation per unit capacity, and the utilization ratio, thereby achieving the matching configuration of photovoltaic installed capacity and park load characteristics.
[0048] 5) Load time series feature extraction: Based on the load curve generated in step 3) and the photovoltaic power generation curve in step 4), calculate the load time series features of the park for k time periods, repeat the calculation until the features of all n load curves are completed.
[0049] The specific characteristics of the park's load time series include ten features: annual maximum load, annual average slope, peak and off-peak season difference coefficient, weekend load ratio, average daily skewness, average daily kurtosis, average daily Shannon entropy, nighttime load ratio, photovoltaic direct absorption rate, and maximum photovoltaic surplus.
[0050] The calculation formulas for these ten features are shown in Formulas 1-10 of Table 3.
[0051] Table 3 Feature Calculation Method
[0052] in, , is the time period number in the energy consumption curve for 8760 hours per year with a 15-minute resolution; T is 35040, which is the sum of the number of time periods throughout the year. This is the time step, with a value of 15 minutes. The net energy consumption during time period t, i.e., max(0, - ), and Let be the energy consumed at time t and the photovoltaic power generation, respectively, in kWh; Refers to the total annual net energy consumption, i.e. The sum, kWh; Maximum load of the year, kWh; It is the annual average slope, kWh / (15 min); It is the coefficient of difference between peak and off-peak seasons; M refers to the month number. M is 12; S high S refers to the set of the three months with the highest energy consumption. low Refers to the three months with the highest energy consumption; C m D is the total energy consumption in month m. m This represents the number of days in the m-th month; It is a time period In days Net energy consumption, D is the number of days, T d It refers to the number of time slots per day. ; It is the average daily skewness. It is the standard deviation of net energy consumption over d days; It is the average daily kurtosis; It is the average daily entropy of the Xiangnong (a type of rice noodle); pro d Let B be the discrete probability density function on day d. d Q is composed of several intervals. i Represents the i-th percentile; It is the weekend load ratio; It is the nighttime load ratio; T weekend Refers to all weekend times included in T; T night T refers to all nighttime times included in T. In this embodiment, nighttime is defined as 19:00 to 5:00 the next day. It refers to the direct absorption rate of photovoltaic power. It represents the largest surplus in photovoltaic power.
[0053] 6) Independently optimize and solve the photovoltaic power generation curve and the industrial park's electricity load curve for each group, and solve the optimal energy storage capacity corresponding to each scenario in turn, and complete the calculation of all n scenarios.
[0054] Specifically, with the goal of minimizing the average electricity cost of the industrial park, the purchase and maintenance costs of photovoltaic and energy storage equipment, as well as the electricity purchase cost from the power grid, are input. The mixed integer linear programming (MILP) method is used to independently optimize and solve each set of photovoltaic power generation curves and industrial park electricity load curves. The optimal energy storage capacity corresponding to each set of scenarios is solved in turn, and the calculation of all n sets of scenarios is completed.
[0055] In this embodiment, the average electricity cost of the park is calculated as shown in Formula 11: (Formula 11) in, This is the electricity purchase cost from the power grid, calculated as shown in Formula 12.
[0056] (Formula 12) The power purchased by the power grid at time t, in kWh. Let t be the time-of-use electricity price at time t; It is the equivalent carbon price of carbon emissions generated by grid power generation, calculated as shown in Formula 13.
[0057] (Formula 13) The total electricity purchased by the power grid that year, in kWh; This is the carbon emission factor of the power grid, taken here as 570.3 gCO2 / kWh; It is the market trading price per unit of carbon emissions; This is the annual investment and operation and maintenance cost of photovoltaic power, as shown in Formula 14.
[0058] (Formula 14) This is the annual investment and operation and maintenance cost of energy storage, as shown in Formula 15.
[0059] (Formula 15) This refers to the installed capacity of photovoltaic power, in kW; It refers to the installed capacity of energy storage, in kWh; , and These are the initial investment costs per unit of photovoltaic capacity / energy storage capacity / energy storage power; CRF is the capital recovery factor. and The annual operation and maintenance cost rates for photovoltaics and energy storage are 2.2% and 3%, respectively. This is the rated power-capacity ratio; in this study, we take 0.5 h. -1 This is equivalent to a 2-hour energy storage system. The return on investment is 6%. This refers to the service life of photovoltaic equipment, which is set at 25 years. This refers to the service life of the energy storage equipment, which is set to 10 years.
[0060] The constraints for optimizing the energy storage capacity in this embodiment are shown in Equations 16-24 of Table 4.
[0061] Table 4 Energy Storage Optimization Constraints
[0062] in, This represents the photovoltaic power generation at time t. This refers to the portion of photovoltaic power generation used on the energy consumption side at time t. This refers to the portion of photovoltaic power generated at time t that is used to charge the energy storage battery. It represents the portion of photovoltaic power generation that is wasted at time t, in kWh. It is the electricity consumption at time t. It is the battery discharge amount at time t. The battery charge at time t, in kWh; It is the portion of the electricity purchased by the power grid at time t that is used on the energy consumption side. It is the portion of the electricity purchased by the grid at time t that is used for battery charging, in kWh; The battery charge at time t, in kWh; The battery charge / discharge efficiency is 94%. This is a charging indicator; the value is 1 if the device is charging and 0 if it is not charging. It is a discharge detection function; the value is 1 if the discharge is in progress, and 0 if the discharge is not progressing. and It is an indicator of rechargeable (dischargeable) capacity, measured in kWh.
[0063] 7) Based on the load time sequence characteristics of step 5) and the optimal energy storage capacity of step 6), a regression fitting model is constructed to obtain a fast calculation model for energy storage capacity.
[0064] This step is specifically as follows: 7-1), the load time sequence characteristics of each load group obtained in step 5) are used as input, and the optimal energy storage capacity obtained in step 6) is used as a one-to-one paired sample dataset of tags.
[0065] 7-2) Based on the constructed paired sample dataset, a random forest model is used as the fitting model to train the fitting of the correlation between load time series characteristics and optimal energy storage capacity (the fitting results are shown in the appendix of the instruction manual). Figure 3 As shown in the figure, the intrinsic mapping relationship between the time-series fluctuation characteristics and time-series distribution patterns of the load and the optimal energy storage capacity is quantified, and finally a fast energy storage capacity calculation model is obtained.
[0066] The model constructed by this invention can directly use the load time-series characteristics of any scenario to be configured as the model input, and quickly and accurately output the corresponding optimal energy storage capacity prediction results through a preset regression fitting formula. It does not require repeated iterative optimization calculations for new load scenarios, effectively simplifying the solution process of optimal energy storage capacity and improving the configuration calculation efficiency.
[0067] When changes occur in the production plan and work scheduling patterns of an industrial park, leading to alterations in load timing characteristics, this model can be used to quickly match and calculate the optimal energy storage capacity to adapt to the new production scenario, enabling rapid planning and configuration of energy storage capacity. Simultaneously, this model can perform reverse engineering, quantitatively analyzing the optimal energy storage capacity and associated electricity costs under different production scheduling patterns through model fitting. For industrial parks with existing energy storage systems, it provides precise data support and technical guidance for iterative optimization of their production scheduling strategies. Ultimately, this forms a two-way collaborative optimization mechanism where "energy storage configuration standards guide production scheduling optimization, and production scheduling adjustments in turn support efficient energy storage operation," effectively improving the overall energy economy of the industrial park and reducing its energy consumption costs.
[0068] The above-described specific embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics, characterized in that: The process includes the following steps: 1) Obtain basic data of the industrial park, including the rated power and fixed operating schedule of electrical equipment with fixed operating time, as well as the types of electrical equipment with variable operating time, rated power and corresponding operating rules; 2) Based on Monte Carlo scheduling sampling, construct a random sampling model for the production operation schedule of the industrial park; Specifically, based on the variable operating time of the electrical equipment in step 1), the equipment activation coefficient is randomly sampled and its value is constrained within the range of [0,1]. At the same time, discrete sampling is used to determine the weekly shift mode, including four modes: one shift, two shifts, three shifts, and three shifts with full-day weekend overtime. n sets of samples are generated through the above sampling method, and each set of samples covers all weeks of the year, including the equipment activation coefficient and shift time type corresponding to each week. 3) Generate the electricity load curve for the industrial park based on Openstudio building energy consumption simulation software; 4) Generate the photovoltaic power generation curve of the industrial park based on the planned photovoltaic installed capacity; 5) Load time series feature extraction: Based on the load curve generated in step 3) and the photovoltaic power generation curve in step 4), calculate the load time series features of the park for k time periods, repeat the calculation until the features of all n load curves are completed; Among them, the time-series characteristics of the park's load are specifically ten characteristics: annual maximum load, annual average slope, peak and off-peak season difference coefficient, weekend load ratio, average daily skewness, average daily kurtosis, average daily Shannon entropy, nighttime load ratio, photovoltaic direct absorption rate, and maximum photovoltaic surplus. 6) Independently optimize and solve each set of photovoltaic power generation curves and industrial park electricity load curves, and solve the optimal energy storage capacity corresponding to each scenario in turn, and complete the calculation of all n scenarios; 7) Based on the load time-series characteristics from step 5) and the optimal energy storage capacity from step 6), a regression fitting model is constructed to obtain a fast energy storage capacity calculation model; specifically: 7-1), the load time series characteristics of each load group obtained in step 5) are used as input, and the optimal energy storage capacity obtained in step 6) is used as a one-to-one paired sample dataset of tags. 7-2) Based on the constructed paired sample dataset, a random forest model is used as the fitting model to fit the correlation between load time series characteristics and optimal energy storage capacity. The intrinsic mapping relationship between load time series fluctuation characteristics, time series distribution law and optimal energy storage capacity is quantified, and finally a fast energy storage capacity calculation model is obtained.
2. The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics as described in claim 1, characterized in that: In step 1), the basic data of the industrial park includes meteorological parameters of the industrial park location, geometric parameters of buildings in the park, thermal parameters of the building envelope, including parameters such as air conditioning cooling and heating set temperature, personnel density, personnel activity level and building fresh air volume, photovoltaic installed capacity, purchase and operation and maintenance costs of photovoltaic and energy storage equipment, and grid electricity purchase costs.
3. The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics as described in claim 1, characterized in that: In step 1), the operating rules for electrical equipment with variable operating time are as follows: based on historical operating data, production plan data, equipment control records or energy consumption monitoring data, statistically extract the equipment operating period, operating duration, start-stop status, operating frequency, load rate or start-up coefficient, as well as the probability distribution parameters of these operating characteristics, and the probability of occurrence of various operating scenarios under different production organization modes, operating conditions or scheduling strategies.
4. The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics as described in claim 1, characterized in that: Step 3) specifically refers to: 3-1) Based on the building geometry parameters in step 1), complete the three-dimensional visualization modeling of the park buildings in the Openstudio building energy consumption simulation software, and input the thermal parameters of the building envelope into the model; 3-2) Input meteorological parameters, the rated power of electrical equipment with fixed running time and fixed operating time, and the rated power of electrical equipment with variable running time into the model of step 3-1) to construct a simulation model of the park's power load; 3-3), Substitute the kth group of annual random operating schedules into the park's electricity load simulation model in step 3-2) to perform energy consumption simulation, and output the annual electricity load curve with a time resolution of 8760 hours and 15 minutes. (3-4) Iterate through k = 1,2,...,n to complete n sets of load curve simulations, where n is the total number of timetables that need to be sampled.
5. The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics as described in claim 1, characterized in that: In step 4), the photovoltaic power generation at any time t = photovoltaic capacity × solar radiation at time t × module efficiency; where photovoltaic capacity is the photovoltaic installed capacity of the park, solar radiation is the real-time effective solar radiation data in the meteorological parameters of the park at the corresponding time, and module efficiency is the comprehensive working efficiency of the photovoltaic module. Based on the above formula, the photovoltaic power generation curve with a time resolution of 8760 hours and 15 minutes throughout the year is obtained, where time t is any time in the high time resolution time series data of 8760 hours and 15 minutes throughout the year.
6. The method for determining the optimal energy storage capacity of an industrial park based on load time-series characteristics as described in claim 1, characterized in that: Step 6) specifically involves: taking minimizing the average electricity cost of the park as the optimization objective, inputting the purchase and maintenance costs of photovoltaic and energy storage equipment, and the electricity purchase cost from the power grid, and using the mixed integer linear programming (MILP) method to independently optimize and solve each set of photovoltaic power generation curves and industrial park electricity load curves, and sequentially solving for the optimal energy storage capacity corresponding to each set of scenarios, thus completing the calculation of all n sets of scenarios.
Citation Information
Patent Citations
Energy storage capacity optimal configuration method based on production operation simulation
CN114722615A
Optical storage optimization method based on production time sequence simulation
CN115860245A
Method and system for optimizing and guiding park new energy consumption balance
CN118646064A
Industrial park building load rapid prediction method based on energy utilization partition
CN120527891A
