Smart city resource flow simulation and planning system based on city metabolism model

Abstract

The invention relates to the technical field of resource planning, and discloses a smart city resource flow simulation and planning system based on a city metabolism model, and the system comprises a parameter collection module which determines a time boundary parameter and a scale parameter; the scale parameterization module is used for calculating metabolic integral scale elasticity; the effective budget generation module is used for calculating to obtain an elastic weighted effective budget; the resource allocation solving module is used for solving the allocation amount of each type of source; the scheduling curve mapping module is used for calculating to obtain a scheduling curve; and the planning result output module outputs the allocation list. According to the method, the robust budget is constructed through the scale parameterization model and the metabolic integral scale elasticity, and the planning deviation under different statistical granularities is reduced. The objective function is fused with the energy cost, the source cost and the scale robust item, the balance of cost control and scale adaptability is realized, and the execution deviation caused by purely pursuing the optimal cost is avoided.

Description

Smart City Resource Flow Simulation and Planning System Based on Urban Metabolism Model Technical Field

[0001] This invention relates to the field of resource planning technology, and more specifically, to a smart city resource flow simulation and planning system based on an urban metabolism model. Background Technology

[0002] In the process of building smart cities, resource flow planning, as a key link in ensuring the stable supply of core resources such as water and energy, needs to adapt to the multi-scenario requirements of macro-budget formulation and micro-equipment scheduling. In current resource flow planning technologies, the setting of the time aggregation window (i.e., statistical caliber) directly affects the effectiveness of the planning results, but traditional methods generally have significant technical bottlenecks.

[0003] Traditional planning processes typically begin by fixing a single time window and then building a forecasting and optimization model based on that caliber. However, different business scenarios have varying requirements for statistical granularity. Planning may use daily or monthly caliber windows, while execution requires hourly or even minute-level windows. This fixed-caliber modeling approach makes planning results highly sensitive to statistical caliber. When the statistical caliber is adjusted, the predicted net stock change within the interval will fluctuate significantly, resulting in budget drift due to caliber shifts, severely impacting the consistency and reliability of the planning scheme.

[0004] Meanwhile, traditional optimization objectives focus solely on minimizing energy and source costs, neglecting scale sensitivity. This leads to cost-optimized allocation schemes that, while cost-effective, are prone to implementation deviations under different statistical calibers. Furthermore, the generation of scheduling curves often relies on manually set allocation rules or independent time-series models, resulting in a disconnect from prior forecasts and an inability to accurately reflect the actual dynamic changes in resource flows, further reducing the feasibility of planning schemes. These combined problems make traditional resource flow planning incapable of simultaneously addressing cost control, scale robustness, and implementation adaptability, failing to meet the demands of refined smart city management. Summary of the Invention

[0005] This invention provides a smart city resource flow simulation and planning system based on an urban metabolism model, which solves the technical problems mentioned in the background.

[0006] This invention provides a smart city resource flow simulation and planning system based on an urban metabolic model, comprising: a parameter acquisition module, which sets the planning period, scale domain, scale variables, planning scope, and reference scope, and determines time boundary parameters and scale parameters; a scale parameterization module, which uses a scale parameterization model to generate instantaneous net flux prediction and interval net stock change based on exogenous feature vectors and scale variables, and calculates the metabolic integral scale elasticity of interval net stock change relative to scale variables; an effective budget generation module, which uses the metabolic integral scale elasticity to construct a weight kernel, performs weighted aggregation calculation on interval net stock change, and obtains an elastically weighted effective budget; and a resource allocation solution module, which establishes a system including energy costs. The objective function for source costs and scale robustness terms is as follows: The scale robustness term is determined by the difference between the flexible weighted effective budget and the interval net stock change after superimposing the total allocation. The solution is to obtain the allocation quantities of various sources and the total allocation quantity that satisfy the source constraints. The scheduling curve mapping module constructs a shape function based on the instantaneous net flux prediction under the reference caliber of the scale parameterized model, decomposes the total allocation quantity into scheduling curves according to the shape function, and maps them to the equipment power time series. The planning result output module calculates the operating cost based on the price parameters, power time series, and source allocation quantity, superimposes the total allocation quantity onto the interval net stock change to obtain the post-planned scale curve, and outputs the allocation list, scheduling curve, cost indicators, and post-planned scale curve.

[0007] The beneficial effects of this invention are as follows: It constructs a robust budget through a scale-parametric model and metabolic integral scale elasticity, reducing planning bias at different statistical granularities. Its objective function integrates energy costs, source costs, and scale robustness terms, achieving a balance between cost control and scale adaptability, avoiding execution bias caused by simply pursuing cost optimization. The scheduling curve is constructed based on prediction results under a reference caliber, closely reflecting the actual dynamics of resource flow and improving feasibility. The output allocation list, scheduling curve, cost indicators, and post-planning scale curve provide unified data support for macro-budgeting and micro-scheduling, thereby meeting the requirements of refined and robust resource planning in smart cities. Attached Figure Description

[0008] Figure 1 is a schematic diagram of the smart city resource flow simulation and planning system based on the urban metabolism model of the present invention.

[0009] In the diagram: Parameter acquisition module 101, Scale parameterization module 102, Effective budget generation module 103, Resource allocation solution module 104, Scheduling curve mapping module 105, Planning result output module 106. Detailed Implementation

[0010] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0011] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" indicate that the element or object preceding the term encompasses the elements or objects listed following the term and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0012] As shown in Figure 1, the smart city resource flow simulation and planning system based on the urban metabolism model includes: a parameter acquisition module 101, which is used to set the planning period, scale domain, scale variables, planning scope and reference scope, and determine time boundary parameters and scale parameters; a scale parameterization module 102, which uses a scale parameterization model to generate instantaneous net flux prediction and interval net stock change based on exogenous feature vectors and scale variables, and calculates the metabolic integral scale elasticity of interval net stock change relative to scale variables; an effective budget generation module 103, which uses the metabolic integral scale elasticity to construct a weight kernel, performs weighted aggregation calculation on interval net stock change, and obtains an elastically weighted effective budget; and a resource allocation solution module 104, which is used to establish... The objective function includes energy costs, source costs, and a scale robustness term. The scale robustness term is determined by the difference between the flexible weighted effective budget and the interval net stock change after superimposing the total allocation. It solves for the various source allocations and the total allocation that satisfy the source constraints. The scheduling curve mapping module 105 is used to construct a shape function based on the instantaneous net flux prediction under the reference caliber of the scale parameterized model. It decomposes the total allocation into scheduling curves according to the shape function and maps them to the equipment power time series. The planning result output module 106 is used to calculate the operating cost based on the price parameters, power time series, and source allocation. It superimposes the total allocation onto the interval net stock change to obtain the post-planned scale curve and outputs the allocation list, scheduling curve, cost indicators, and post-planned scale curve.

[0013] In one embodiment of the present invention, the planning period, scale domain, scale variable, planning scope, and reference scope are set, and the time boundary parameters and scale parameters are determined, including: determining the start time t0 and the end time t1, defining the planning period T=[t0,t1], calculating the duration L=t1-t0, where t0 is the start time, t1 is the end time, T is the planning period, and L is the total duration of the planning period; defining a time aggregation window Δt, calculating the scale variable s=ln(Δt) under the condition that Δt>0, establishing the inverse mapping Δt=exp(s), where Δt is the time aggregation window, s is the dimensionless scale variable, ln(·) is the natural logarithm function, and exp(·) is the exponential function; and setting the minimum time aggregation window Δt. min With the maximum time aggregation window Δt max Satisfying 0 < Δt min ≤Δt max ≤L, lower bound of the computational scale domain s min =ln(Δt min ) and the upper bound of the scale domain s max =ln(Δt max ), where Δt min For the minimum time aggregation window, Δt max For the maximum time aggregation window, s min As the lower bound of the scale domain, s maxThe upper bound of the scale domain is used; the time aggregation window Δt corresponding to the planning caliber is selected. p Time aggregation window Δt corresponding to the reference aperture ref Satisfying Δt min ≤Δt p ≤Δt max And Δt min ≤Δt ref ≤Δt max Calculate the planning caliber s p =ln(Δt p ) and reference caliber s ref =ln(Δt ref ), where Δt p For the time aggregation window corresponding to the planning caliber, Δt ref For the time aggregation window corresponding to the reference caliber, s p For the purpose of planning the scope, s ref For reference purposes, the planning period T, start time t0, end time t1, and total planning period L are aggregated into time boundary parameters, and the scale variable s and the lower bound of the scale domain s are also included. min Scale domain upper bound s max Planning scope p Compared with reference caliber s ref It is aggregated into scale parameters.

[0014] It should be noted that the planning period represents the complete time interval for resource flow simulation and planning. The duration of the planning period represents the time span from the start time to the end time of the planning. The time aggregation window represents the time interval for merging and statistically analyzing time-series data. The dimensionless scaling variable represents the unitless variable obtained by transforming the natural logarithm of the time aggregation window. The inverse mapping relationship of the exponential function represents the restoration of the dimensionless scaling variable to the time aggregation window. The lower bound of the scale domain represents the minimum value of the scale domain, and the upper bound of the scale domain represents the maximum value of the scale domain. The scale domain represents the range of values ​​for the dimensionless scaling variable. The time aggregation window corresponding to the planning caliber represents the data aggregation time interval used for budget planning, reflecting the data statistical granularity at the planning level. The time aggregation window corresponding to the reference caliber represents the data aggregation time interval used for execution scheduling, reflecting the data statistical granularity at the execution level. The planning caliber represents the dimensionless parameter obtained by transforming the natural logarithm of the time aggregation window corresponding to the planning caliber. The reference caliber represents the dimensionless parameter obtained by transforming the natural logarithm of the time aggregation window corresponding to the reference caliber. The time boundary parameter represents the set of time parameters related to the planning period that have been aggregated. The scale parameter represents the set of scale-related variables and relationships in the aggregation.

[0015] It should be noted that time aggregation windows themselves have time intervals with units, such as 1 hour or 24 hours. Directly using them for model calculations may affect computational stability due to unit differences or excessively large numerical ranges. Converting them to dimensionless scaling variables using the natural logarithm allows time aggregation windows of different magnitudes to be mapped to similar numerical intervals. For example, the natural logarithm for 1 hour is approximately 0, for 24 hours it is approximately 3.18, and for 720 hours it is approximately 6.58, making scale-related calculations more efficient and stable, while providing a unified dimensionless benchmark for subsequent scale sensitivity analysis. The inverse exponential function mapping is used when it is necessary to restore dimensionless scaling variables to specific time aggregation windows. For example, if it is necessary to determine the actual data statistical interval based on the calculated dimensionless scaling variable, this inverse mapping can be used. The operation is as follows: given the value of the dimensionless scaling variable, calculate the result of the exponential function for that value; this is the corresponding time aggregation window. For example, when the dimensionless scaling variable has a value of 3.18, its exponential function result is approximately 24, and the corresponding time aggregation window is 24 hours.

[0016] It's important to note that the selection of the minimum and maximum time aggregation windows should be based on actual business needs and data availability. For example, if the planning scenario involves urban water supply scheduling, and the execution level requires hourly precision, then the minimum time aggregation window can be set to 1 hour. If the planning level requires monthly budgeting, then the maximum time aggregation window can be set to 720 hours (30 days). At the same time, strict constraints must be followed: the minimum time aggregation window must be greater than 0, the maximum time aggregation window must not exceed the duration of the planning period, and the minimum time aggregation window cannot be greater than the maximum time aggregation window. For example, if the planning period lasts for 720 hours, then the maximum time aggregation window cannot exceed 720 hours. If the minimum time aggregation window is set to 1 hour, then the maximum time aggregation window can be selected between 1 hour and 720 hours.

[0017] It should be noted that the time aggregation window corresponding to the planning caliber needs to adapt to the needs of planning-related business such as budget formulation and resource allocation, and usually selects a larger time granularity, such as 24 hours per day or 720 hours per month, to facilitate macro-level decision-making. The time aggregation window corresponding to the reference caliber needs to adapt to the needs of execution-related business such as equipment scheduling and real-time monitoring, and usually selects a smaller time granularity, such as 1 hour per hour or 0.0167 minutes per minute, to facilitate execution. Both must fall between the minimum and maximum time aggregation windows. For example, if the minimum time aggregation window is 1 hour and the maximum time aggregation window is 720 hours, then the time aggregation window corresponding to the planning caliber can be 24 hours, and the time aggregation window corresponding to the reference caliber can be 1 hour. This invention defines the time boundary of the planning period, thus determining the time range and span of the plan; it achieves standardized representation and bidirectional conversion of time granularity by converting time aggregation windows into dimensionless scale variables and establishing inverse mapping; it defines the effective range and application scenarios of scale parameters by setting scale domains and time aggregation windows corresponding to two types of calibers; and finally, by aggregating time boundary parameters and scale parameters, it provides consistent and reusable basic parameters for subsequent modules such as prediction calculation, budget generation, and optimization solution, ensuring that each module works collaboratively under a unified parameter system, avoiding calculation deviations caused by inconsistent parameter definitions, and guaranteeing the overall accuracy and operability of the system.

[0018] In one embodiment of the present invention, a scale-parameterized model is employed to generate instantaneous net flux prediction and interval net stock change based on exogenous eigenvectors and scale variables, and to calculate the metabolic integral scale elasticity of interval net stock change relative to the scale variables, including: establishing a scale-parameterized model f within the planning period T and the scale domain. θ , exogenous feature vector Using a dimensionless scaling variable s as input to a scale parameterized model, instantaneous net flux predictions are calculated. ,in For instantaneous net flux prediction, f is an exogenous eigenvector. θ For the scale-parameterized model; the time aggregation window Δt is calculated based on the inverse mapping relationship of the exponential function Δt=exp(s), and the discrete time point sequence t is constructed. m =t0+mΔt and numerical integration weight w m (s)=Δt, performing a weighted cumulative calculation of the change in net stock over the time interval for the instantaneous net flux forecast. Where Δt is the time aggregation window, exp(·) is the exponential function, and t m For discrete time points, t0 is the start time, m is the positive integer index, and w m (s) represents the numerical integration weight. The change in net stock over the interval is represented by N(s), which is the largest integer satisfying the time constraint; the current scale point s is selected within the scale domain.k Smaller scale point s k-1 With larger scale point s k+1 Calculate the integral scale elasticity of the change in net stock over a period relative to the scaling variable. M-ISE θ (s k ) represents the metabolic integral scale elasticity, ln(·) is the natural logarithm function, |·| is the absolute value sign, and s k For the current scale point, s k-1 For a smaller scale point, s k+1 These are points on a larger scale.

[0019] It should be noted that the scale-parametric model represents a parameterized prediction model incorporating scale variables, used to generate instantaneous net flux predictions based on exogenous eigenvectors and scale variables. Exogenous eigenvectors represent the set of external features influencing resource flows, reflecting the impact of external factors such as weather, prices, and equipment status on resource flows. Instantaneous net flux prediction represents the predicted net flow of resources at a specific time and scale, reflecting the dynamic trend of resource flows across spatiotemporal scales. Discrete time point sequence represents an ordered set of time points divided by time aggregation windows within the planning period. Numerical integration weights represent the weight values ​​used for numerical integration calculations, reflecting the degree of integration contribution corresponding to discrete time points. Interval net stock change represents the change in net resource stock within the planning period. Current scale point represents the target scale point within the scale domain where the metabolic integral scale elasticity to be calculated is located. Metabolic integral scale elasticity represents the logarithmic elasticity of interval net stock change relative to scale variables, reflecting the sensitivity of resource stock changes to scale changes.

[0020] It should be noted that traditional prediction models typically only take temporally exogenous features as input and output prediction results at a fixed scale, failing to reflect the impact of different time aggregation windows on the prediction results. Scale-parametric models, by incorporating dimensionless scale variables into the input layer, enable the model to learn the correlation between exogenous features and net resource flux at different scales. For example, the same exogenous feature may have different weights on net water resource flux at hourly and daily scales. The model can adaptively capture this scale dependence, thereby outputting instantaneous net flux predictions at different scales, providing model support for subsequent scale sensitivity analysis. The composition of the exogenous feature vector needs to be determined in conjunction with the specific resource flow scenario, and the selection criteria are that it has a significant correlation with net resource flux and is temporally available. For example, in urban water resource flow scenarios, exogenous features may include meteorological features such as daily precipitation and daily evaporation, social activity features such as production plans of large industrial water users and peak water usage periods for residents, and equipment features such as the operating efficiency of water supply pump sets and the leakage rate of water transmission networks. In energy resource flow scenarios, these may include meteorological features such as daily average temperature and wind speed, market features such as the peak-valley period division of real-time electricity prices, and load features such as industrial electricity load and residential electricity load. During screening, correlation analysis methods such as Pearson correlation coefficient and mutual information entropy can be used to retain features with an absolute value of correlation coefficient greater than 0.3 with the instantaneous net flux, ensuring the effectiveness of the features and the generalization ability of the model.

[0021] It should be noted that the generation of the discrete time point sequence starts from the start time of the planning period and recursively generates each time point in turn, using time aggregation windows as intervals. That is, the first time point is the planning start time, the second time point is the planning start time plus one time aggregation window, and so on, until the generated time point does not exceed the planning end time. If the duration of the planning period is not an integer multiple of the time aggregation window, the last time point is the maximum time point not exceeding the planning end time. The remaining time period less than one time aggregation window is still included in the integration calculation of the last time point to ensure that all time within the planning period is covered. For example, if the planning period is 72 hours and the time aggregation window is 24 hours, then the discrete time point sequence is the planning start time, planning start time plus 24 hours, planning start time plus 48 hours, and planning start time plus 72 hours; if the planning period is 80 hours and the time aggregation window is 24 hours, then the discrete time point sequence is the planning start time, planning start time plus 24 hours, planning start time plus 48 hours, and planning start time plus 72 hours, with the remaining 8 hours included in the calculation of the last time point.

[0022] It should be noted that the selection of larger and smaller scale points must satisfy the following conditions: they must be adjacent to the current scale point within the scale domain, and the distance between the two points and the current scale point must be equal to ensure the accuracy of the elasticity calculation. The spacing setting needs to be combined with the range and accuracy requirements of the scale domain, and is usually selected as one percent to five percent of the scale domain length. For example, if the scale domain is 0 to 6.58, corresponding to 1 hour to 720 hours, and the current scale point is 3.18, corresponding to 24 hours, if one percent of the scale domain length, i.e., 0.0658, is selected as the spacing, then the smaller scale point is 3.18 minus 0.0658, which equals 3.1142, and the larger scale point is 3.18 plus 0.0658, which equals 3.2458. If the current scale point is close to the lower bound of the scale domain, the smaller scale point is taken as the lower bound of the scale domain, and the distance between the larger scale point and the current scale point is the difference between the current scale point and the lower bound; if the current scale point is close to the upper bound of the scale domain, the larger scale point is taken as the upper bound of the scale domain, and the distance between the smaller scale point and the current scale point is the difference between the upper bound and the current scale point, ensuring that there are always two valid reference points.

[0023] In one embodiment of the present invention, a weighting kernel is constructed using a metabolic integral scale elasticity, and a weighted aggregation calculation is performed on the interval net stock change to obtain an elastically weighted effective budget, including: setting a scale weighting kernel sensitivity coefficient α, in the scale domain [s min ,s max Internally constructed scale weight kernel , where K(s) p ,s) is the scale weight kernel, exp(·) is the exponential function, α is the scale weight kernel sensitivity coefficient, |·| is the absolute value sign, M-ISE θ (s) represents the metabolic integral scale elasticity, where s is the scaling variable, u is the integral variable, and s min As the lower bound of the scale domain, s max The upper bound of the scale domain is used; the scale weights are used to verify the changes in net stock over the interval, and a weighted aggregation calculation is performed to obtain a flexible weighted effective budget. B eff (s p ) represents a flexible weighted effective budget, s p For the purpose of planning, This represents the change in net stock over the period.

[0024] It should be noted that the scale weight kernel represents the non-negative normalized weights constructed based on the metabolic integral scale elasticity, reflecting the contribution weights of net stock changes in different scales to budget calculations. The normalization factor represents the definite integral of the negative exponential function value of the absolute value of the metabolic integral scale elasticity in the scale domain, reflecting the normalization calibration benchmark of the scale weight kernel. The elastic weighted effective budget represents the single budget quantity obtained after weighted aggregation by the scale weight kernel, reflecting a scale-robust budget value that fits the planning caliber.

[0025] It should be noted that traditional budget calculations typically employ equal weighting or subjective weighting, failing to consider the scale sensitivity differences in net stock changes across different scales. This makes budget results susceptible to changes in statistical methods. In this invention, the larger the absolute value of the metabolic integral scale elasticity, the more sensitive the net stock changes across the interval are to scale changes at that scale. The smaller the corresponding negative exponential function value, the smaller the scale weight kernel obtained after normalization, thus reducing the contribution of this sensitive scale during budget aggregation. Conversely, the smaller the absolute value of the metabolic integral scale elasticity, the larger the corresponding weight kernel, and the more the budget results rely on scale-robust data on net stock changes across the interval. For example, if the absolute value of the metabolic integral scale elasticity is 3 at one scale and 0.8 at another, and the coefficient in the negative exponential function is 1, the negative exponential value of the former is approximately 0.05, while that of the latter is approximately 0.449. After normalization, the weight of the former is significantly lower than that of the latter, achieving adaptive weight reduction for sensitive scales.

[0026] It should be noted that the coefficients in the negative exponential function are used to adjust the influence of the metabolic integral scale elasticity on the scale weight kernel. The values ​​should be chosen based on the specific resource flow scenario's requirements for budget scale robustness, typically ranging from 0.5 to 5. If the scenario has extremely high requirements for budget stability, the weighting effect on highly elastic and sensitive scales needs to be strengthened, and the coefficients can be selected from 3 to 5. If the scenario allows for a certain degree of scale fluctuation in the budget, the coefficients can be selected from 0.5 to 1. For example, urban core area water supply resource planning has stringent requirements for budget stability, so a coefficient of 4 can be used. In this case, the negative exponential value corresponding to a scale with an absolute elasticity of 2 is approximately 0.018, significantly reducing the weight ratio. Conversely, suburban agricultural irrigation water planning has relatively lenient stability requirements, so a coefficient of 0.8 can be used, and the negative exponential value corresponding to a scale with an absolute elasticity of 2 is approximately 0.202, resulting in a relatively moderate weight ratio. The coefficients need to be determined through validation in a small number of scenarios to ensure that the weight allocation matches business needs.

[0027] It should be noted that the adaptation of the flexible weighted effective budget to the planning scope is achieved through the natural correlation between the integration range of the scale weight kernel and the scale characteristics of the planning scope. During calculation, the integration range of the scale weight kernel covers the entire scale domain, but the scale sensitivity of net stock changes in the areas surrounding the scale point corresponding to the planning scope is usually lower, resulting in larger corresponding weight kernels. This naturally tilts the weighted aggregation result towards the statistical granularity of the planning scope. For example, if the scale point corresponding to the planning scope is 3.18 corresponding to 24 hours, the surrounding scales, such as 3.0 to 3.36 corresponding to 20 to 29 hours of metabolic integration, typically have lower scale elasticity and larger weights. These scales contribute more significantly during the integration process, ultimately allowing the flexible weighted effective budget to adapt to the business needs of the planning scope without additional limitations on the integration range.

[0028] It should be noted that this invention constructs a scale weight kernel flexibly using the metabolic integral scale to achieve adaptive weight allocation at different scales, allowing scale-robust interval net stock changes to receive higher weights. Then, through weighted aggregation of the weight kernel and interval net stock changes, the stock change information across all scales is compressed into a single, flexibly weighted effective budget. This avoids the information loss of traditional single-caliber budgets and solves the problem of budget results being sensitive to statistical calibers, providing an accurate and robust budget benchmark for subsequent resource allocation optimization and ensuring the consistency and reliability of resource flow planning schemes at different scales.

[0029] In one embodiment of the present invention, an objective function is established that includes energy cost, source cost, and a scale robustness term. The scale robustness term is determined by the difference between the elastically weighted effective budget and the interval net stock change after superimposing the total allocation. The solution is then used to determine the various source allocations and the total allocation that satisfy the source constraints, including: setting the local water source allocation V. L The source and allocation of reclaimed water V R The source allocation of water for inter-basin water transfer V T Calculate the total allocation V tot =V L +V R +V T V tot V represents the total allocation. L V represents the local water supply. R V represents the allocation of reclaimed water sources. T The total water allocation for inter-basin water transfer is calculated by adding the total allocation to the change in net stock between the two basins, thus obtaining the planned change in net stock between the two basins. , where ΔS plan (s) represents the net change in stock within the planned interval. Let s be the change in net stock over the interval, and s be the scaling variable. Scale robustness terms are calculated using the scaling weight kernel and the flexible weighted effective budget. ,in For the scale robustness term, s min As the lower bound of the scale domain, s max As the upper bound of the scale domain, K(s) p ,s) is the scale weight kernel, B eff (s p ) represents a flexible weighted effective budget, s p For planning purposes; calculate energy costs C E For energy costs, The energy cost coefficient corresponding to the unit ration; calculation of source costs. C T For source cost, p L p R p TThe unit cost parameters for local water, reclaimed water, and inter-basin water transfer are defined respectively; the tradeoff coefficient λ for the scale robustness term is set, and the objective function is constructed. Where J is the objective function; set an upper limit for the local water source allocation. Upper limit of reclaimed water source allocation Upper limit of water allocation from inter-basin water transfer sources Under constraints , and Next, find the local water source allocation V that minimizes the objective function J. L V, the amount of reclaimed water source and allocation R Inter-basin water transfer source allocation V T With total allocation V tot The value.

[0030] It should be noted that the allocation quantities from various sources represent the allocation quantities from different resource supply channels, reflecting the resource allocation scale of each supply channel. The total allocation quantity represents the sum of the allocation quantities from all resource supply channels. The change in net stock after planning represents the change in net resource stock after adding the total allocation quantity. The scale robustness term represents the weighted cumulative value of the deviation between the change in net stock after planning and the flexible weighted effective budget, reflecting the robustness of resource allocation to scale changes. The energy cost coefficient per unit allocation represents the average energy cost corresponding to a unit resource allocation quantity, reflecting the energy cost efficiency of resource allocation. The time-varying energy price parameter represents the energy price changing over time during the planning period. The equipment power time series represents the sequence of equipment operating power changes over time, reflecting the time-series distribution of equipment energy consumption. The source cost represents the sum of allocation costs from various resource supply channels. The objective function represents the optimization objective expression integrating energy cost, source cost, and scale robustness term, used to solve for the optimal allocation quantity. The trade-off coefficient represents the coefficient that adjusts the importance of the scale robustness term and the cost term, reflecting the balance between scale robustness and cost control. The upper limit parameters for the allocation of various sources represent the maximum allowable allocation quantity for each type of resource supply channel, reflecting the supply capacity constraints of each supply channel.

[0031] It should be noted that traditional resource allocation optimization typically focuses only on cost minimization, neglecting the robustness of allocation schemes to scale changes, leading to susceptibility to deviations due to adjustments in statistical methods. This invention introduces a scale-robust term into the objective function, incorporating scale sensitivity into the optimization constraints, and adjusting its weight relative to the cost term through a tradeoff coefficient. For example, when water resource allocation in urban core areas prioritizes stability, the tradeoff coefficient can be set to 5 to strengthen the constraint of the scale-robust term; when agricultural water allocation in suburban areas focuses more on cost control, the tradeoff coefficient can be set to 0.8 to weaken the impact of the scale-robust term, achieving a balance between objectives in different scenarios. The value of the tradeoff coefficient needs to be set in conjunction with the priority of scale robustness and cost control in the resource flow scenario, typically ranging from 0.1 to 10. If the scenario has extremely high requirements for scale robustness, such as the allocation of lifeline water supply in cities, the tradeoff coefficient can be set to 5 to 10, allowing the scale robustness term to have a higher proportion in the objective function. If the scenario focuses on cost control, such as the allocation of general industrial water, the tradeoff coefficient can be set to 0.1 to 1, highlighting the dominant role of the cost term. If both have equal priority, the tradeoff coefficient can be set to 1 to 3. The coefficients need to be verified through trial calculations to ensure that the optimization results meet the scale robustness requirements without causing excessive cost increases. For example, if a trial calculation for a certain scenario shows that a tradeoff coefficient of 2 meets the scale robustness requirements and the cost increases by no more than 5% compared to the baseline, then this is a reasonable value.

[0032] It should be noted that the upper limit parameters for the allocation of various sources need to be determined based on the actual supply capacity of the supply channels, with the core basis including resource reserves, production capacity, and transportation capacity. For example, the upper limit of local water allocation is determined by the combined factors of local reservoir capacity, exploitable groundwater, and water supply facility transmission capacity. Assuming the maximum available water supply from the local reservoir is 100,000 cubic meters, the exploitable groundwater is 50,000 cubic meters, and the maximum transmission capacity of the water pipeline is 120,000 cubic meters, then the upper limit of local water allocation is the minimum of these three factors, 100,000 cubic meters. The upper limit of reclaimed water allocation is determined by the product of the daily treatment capacity of the reclaimed water treatment plant and the planning period. If the daily treatment capacity of the treatment plant is 2,000 cubic meters and the planning period is 30 days, then the upper limit of reclaimed water allocation is 60,000 cubic meters. The upper limit of inter-basin water transfer allocation is determined by the maximum water transfer volume agreed upon in the water transfer agreement. If the agreement stipulates an annual water transfer volume of 360,000 cubic meters and the planning period is 30 days, then the upper limit of water transfer allocation is 30,000 cubic meters.

[0033] It should be noted that time-varying energy price parameters must be obtained from authoritative channels such as local energy management departments and power companies, prioritizing officially released time-of-use pricing and peak-valley pricing data. The granularity of the time series division must match the planning period and dispatch cycle. If the planning period is monthly and the dispatch cycle is hourly, the time-varying energy price parameters are divided by hour, with each day divided into peak, flat, and valley periods. For example, the peak period price from 8:00 to 22:00 is 1.2 yuan per kilowatt-hour, and the valley period price from 22:00 to 8:00 the next day is 0.5 yuan per kilowatt-hour. If the planning period is annual and the dispatch cycle is daily, the time-varying energy price parameters are divided by day, distinguishing between weekday and holiday prices. If authoritative time series data is unavailable, the monthly average electricity price can be used, broken down by peak-valley ratio. For example, if the monthly average electricity price is 0.8 yuan per kilowatt-hour and the peak-valley ratio is 2:1, then the peak price is 1.07 yuan per kilowatt-hour, and the valley price is 0.53 yuan per kilowatt-hour.

[0034] In one embodiment of the present invention, a shape function is constructed based on the instantaneous net flux prediction under a reference caliber using a scale-parameterized model. The total allocation is then decomposed into scheduling curves according to the shape function and mapped to equipment power timing. This includes: calling the reference caliber s. ref Given a planning period T=[t0,t1], obtain the instantaneous net flux forecast under the reference caliber. Construct shape function Where h(t) is the shape function, and max{·} is the maximum value operation. For the instantaneous net flux prediction under the reference caliber, t is time, t0 is the start time, t1 is the end time, and u is the integration variable; obtain the total allocation V. tot Calculate the scheduling curve , where q plan (t) is the scheduling curve, V tot Given the total allocation; establish the equipment energy consumption function e(·) and calculate the equipment power timing. , where P plan (t) represents the device power timing sequence, and e(·) represents the device energy consumption function.

[0035] It should be noted that non-negative flux represents the portion of instantaneous net flux forecast that is not less than zero, reflecting the positive flow scale of resource flows. The normalization factor represents the definite integral result of non-negative flux over the planning period. The shape function represents the weighting function describing the temporal distribution of resource scheduling, reflecting the temporal allocation pattern of resources during the planning period. The scheduling curve represents the time-series curve of the total allocation, reflecting the instantaneous scheduling scale of resources during the planning period. The equipment energy consumption function represents the mapping relationship between the scheduling curve and equipment power, used to convert resource scheduling quantities into equipment operating power.

[0036] It should be noted that traditional resource scheduling curve generation often relies on manually set allocation rules or independent time-series models, which can easily become disconnected from the prediction results, leading to a mismatch between scheduling and actual demand. This invention directly reuses the instantaneous net flux prediction of a scale-parameterized model under a reference caliber. By taking the larger value between the predicted value and zero, non-negative flux is obtained, ensuring that the shape function only reflects the temporal pattern of positive resource flow. Then, through definite integral normalization within the planning period, the shape function satisfies the constraint that the integral value is one. Thus, multiplying the total allocation by the shape function not only restores the total scale but also closely matches the predicted temporal distribution. For example, if the instantaneous net flux prediction under the reference caliber peaks between 8:00 AM and 10:00 AM, the shape function will have a higher weight for that time period, and the scheduling curve will naturally reflect the scheduling peak during that period, achieving temporal coordination between prediction and scheduling.

[0037] It should be noted that the purpose of nonnegation is to ensure the nonnegativity of the shape function, avoiding negative scheduling quantities (i.e., resource backflow) in the scheduling curve, and conforming to the constraints of actual resource scheduling. In smart city resource flow scenarios, such as water and power supply scheduling, equipment operation can only achieve positive resource supply and cannot recover and redistribute resources in reverse; negative scheduling quantities have no practical significance. For example, if the instantaneous net flux prediction is negative due to model error, taking the larger value of the two (the one above zero) results in zero nonnegative flux for that period, zero weight for the corresponding shape function, and no scheduling quantity in the scheduling curve for that period, thus avoiding unreasonable negative scheduling instructions and ensuring the executability of the scheduling scheme. The form of the equipment energy consumption function needs to be determined in conjunction with the resource type and the operating characteristics of the equipment. The core is to establish a quantitative mapping relationship between the scheduling curve (resource flow / load) and the equipment power. Common forms include linear functions, quadratic functions, and piecewise functions. For example, in a water supply scenario, the energy consumption function of a water pump, when the flow rate is within the rated range, has an approximate quadratic relationship with the flow rate. This can be expressed as: equipment power equals coefficient a multiplied by the square of the flow rate plus coefficient b multiplied by the flow rate plus a constant c. Here, coefficients a, b, and c are obtained by fitting the pump's performance parameter table or measured data. In a power supply scenario, the energy consumption function of a transformer is approximately linear within a load range of 70% to 100%. This can be expressed as: equipment power equals coefficient k multiplied by the load, where coefficient k is the transformer's load loss coefficient. When selecting parameters, the performance parameters provided by the equipment manufacturer should be prioritized. If no explicit parameters are available, fitting can be performed using 3 to 5 sets of measured data to ensure the function mapping accuracy meets engineering requirements.

[0038] It should be noted that the scaling parameterization model uses an LSTM (Long Short-Term Memory) network enhanced with a temporal attention mechanism as its basic structure. The model is divided into four modules: input layer, feature fusion layer, temporal encoding layer, and output layer. 1. Input layer: Receives two types of input data: first, the exogenous feature sequence within the planning period, with a dimension of (number of time steps × exogenous feature dimension), where the number of time steps is determined by the planning period duration and the reference time aggregation window, and the exogenous feature dimension is the total number of filtered features; second, the scale variable, with a dimension of 1, participates in the model calculation as a global feature. 2. Feature fusion layer: Maps the scale variable to a vector with the same dimension as the exogenous feature sequence through a fully connected layer, and then fuses the exogenous features and scale variable through element-wise addition, outputting a fused feature sequence with the same dimension as the input exogenous feature sequence. 3. Temporal Encoding Layer: This layer consists of two bidirectional LSTM networks and an attention mechanism layer. The first LSTM network has an output dimension of 128, and the second LSTM network has an output dimension of 64. The attention mechanism layer assigns weights to the output of the second LSTM network, strengthening the feature representation of key time steps and outputting the encoded temporal feature vector. 4. Output Layer: This layer maps the temporal feature vector to the instantaneous net flux prediction value through a fully connected layer. The output dimension is (number of time steps × 1), consistent with the number of time steps in the input exogenous feature sequence, corresponding to the instantaneous net flux prediction result at each time point within the planning period.

[0039] It should be noted that the training samples for the scale parameterization model can be selected from historical data covering multiple complete planning cycles over the past 3 to 5 years and including all time aggregation windows within the preset scale domain as the data source. The data is divided into multiple sample segments according to the length of the target planning period. For the exogenous features of each segment, missing values ​​are filled by linear interpolation, outliers are removed by the 3σ criterion, and then normalized to the [0,1] interval by Min-Max. The time aggregation windows within the scale domain are uniformly sampled and transformed by natural logarithm to obtain scale variables. Each standardized exogenous feature sequence is combined with all sampled scale variables to form sample pairs. Based on historical measured resource flow data, the net flux true value at each time point is calculated according to the time aggregation window corresponding to the sample as the label sequence. Finally, all samples composed of standardized exogenous feature sequences, scale variables, and label sequences are randomly divided into training set, validation set, and test set in a 7:2:1 ratio to ensure that the time distribution and scale distribution of the three datasets are consistent. Furthermore, mean squared error was used as the loss function, the Adam optimizer was selected, and the initial learning rate was set to 0.001. The learning rate was halved when the validation set loss did not decrease for 5 consecutive rounds. The training rounds were set to 100 rounds, the batch size to 32, and an early stopping strategy was adopted, in which training was stopped when the validation set loss did not decrease for 10 consecutive rounds. The training set samples were input into the model in batches, the predicted values ​​were calculated through forward propagation, the bias was calculated using the loss function, and the model parameters were updated through backpropagation. The model performance was evaluated and the loss was recorded after each training round using the validation set. After training, the generalization ability was evaluated using the test set, ensuring that the mean squared error of the test set did not exceed 1.2 times the mean squared error of the training set. Finally, the model parameters with the minimum validation set loss were saved.

[0040] In one embodiment of the present invention, operating costs are calculated based on price parameters, power time series, and source allocation. The total allocation is superimposed on the change in net stock over the interval to obtain a post-planning scaling curve. The output includes an allocation list, a scheduling curve, cost indicators, and the post-planning scaling curve, including: calling the planning period T=[t0,t1] and the time-varying electricity price parameter c. e (t) and device power timing P plan (t), calculate energy cost ,in For energy cost, t0 is the start time, t1 is the end time, and c is the end time. e (t) represents the time-varying electricity price parameter, P plan (t) represents the equipment power timing; obtain the local water source allocation V. L V, the amount of reclaimed water source and allocation R Inter-basin water transfer source allocation V T Obtain the corresponding source unit cost parameter p L p R With p T Calculate source cost ,in For source cost, V L V R V T These represent the local water supply allocation, the reclaimed water supply allocation, and the inter-basin water transfer supply allocation, respectively. L p R p T Provide the corresponding source unit cost parameters; calculate the total operating cost. ,in The total operating cost includes energy costs. Source cost Total operating costs Aggregate into cost indicators; obtain the change in net inventory over the period. With total allocation V tot In the scale domain [s min ,s max ] Calculate the post-planning scale curve , where ΔS plan (s) represents the post-planning scaling curve. V represents the net change in stock over the period. tot Let s be the total allocation, and s be the scale variable. min As the lower bound of the scale domain, s max The upper bound of the scale domain is defined as the local water source allocation V. L V, the amount of reclaimed water source and allocation R Inter-basin water transfer source allocation V T The data is aggregated into a ration list, and the ration list and scheduling curve q are output. plan (t), cost indicators and post-planning scaling curve ΔS plan (s).

[0041] It should be noted that energy cost represents the total cost of energy consumed by equipment operation during the planning period. The unit cost parameter represents the unit allocation price of various resource supply channels, reflecting the cost differences between different supply channels. The total operating cost represents the sum of energy cost and source cost. The cost index represents the aggregated total of energy cost, source cost, and operating cost. The post-planning scale curve represents the change in net resource stock at different scales after superimposing the total allocation, reflecting the impact of the planning scheme on the stock at each scale. The allocation list represents the collection of allocations from various sources, reflecting the allocation results of resources through different supply channels.

[0042] It should be noted that traditional cost accounting often only calculates total energy cost without establishing a correlation with the allocation quantity, thus failing to measure the energy consumption efficiency per unit of resource. This invention uses a unit allocation energy cost coefficient to break down the total energy cost into the unit resource allocation quantity. For example, if the total allocation quantity is 1000 cubic meters and the energy cost is 5000 yuan, then the coefficient is 5 yuan per cubic meter. This coefficient can be used to compare different planning schemes. For instance, if the coefficient for scheme A is 4.5 yuan per cubic meter and that for scheme B is 5.2 yuan per cubic meter, it indicates that scheme A has better unit allocation energy efficiency, providing a precise basis for decision-making. The output format of the post-planning scale curve needs to be discretized data, outputting corresponding values ​​at uniform nodes within the scale domain. For example, if the scale domain is 0 to 6.58 (corresponding to 1 hour to 720 hours), a node is taken every 0.1, for a total of 66 nodes, outputting the net change in inventory for each node within the post-planning interval. Application scenarios include multi-caliber report verification, such as checking whether inventory changes meet budget requirements on a daily basis (scale 3.18) and assessing long-term inventory balance on a monthly basis (scale 6.58). It can also be used to verify the robustness of a solution; if the curves fluctuate little across different scales, it indicates that the solution is highly adaptable to scale changes. Furthermore, a zero energy cost coefficient per unit allocation is only applicable to scenarios where the total allocation is zero, i.e., when there is no resource scheduling requirement, the equipment does not need to operate, and the energy cost is zero. Setting the coefficient to zero avoids errors in division operations.

[0043] It should be noted that energy costs are calculated by integrating time-varying energy prices and equipment power time series data. Source costs are then obtained by combining source allocation quantities and unit costs, and compiled into a comprehensive cost indicator. The total allocation quantity is overlaid with the change in net stock over the interval to generate a post-planning scale curve reflecting the impact at multiple scales. Simultaneously, the allocation list and scheduling curves are compiled to form a complete output system. This achieves the transformation from process data to business indicators, presenting the impact of planning schemes on stock at various scales, providing decision-makers with comprehensive and directly applicable execution data, and ensuring the practicality and completeness of the output results.

[0044] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0045] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A smart city resource flow simulation and planning system based on an urban metabolism model, characterized in that, include: The parameter acquisition module allows setting the planning period, scale domain, scale variables, planning scope, and reference scope, and determining time boundary parameters and scale parameters. The scaling parameterization module uses a scaling parameterization model to generate instantaneous net flux predictions and interval net stock changes based on exogenous feature vectors and scaling variables, and calculates the metabolic integral scale elasticity of interval net stock changes relative to scaling variables. The effective budget generation module uses the metabolic integral scale elasticity to construct a weight kernel, performs weighted aggregation calculations on interval net stock changes, and obtains an elastically weighted effective budget. The resource allocation solution module establishes an objective function that includes energy costs, source costs, and scale robustness terms. The scale robustness term is determined by the difference between the elastically weighted effective budget and the interval net stock change after superimposing the total allocation, and solves for various source allocations and the total allocation that satisfy source constraints. The scheduling curve mapping module constructs a shape function based on the instantaneous net flux predictions under the reference caliber of the scaling parameterization model, decomposes the total allocation into scheduling curves according to the shape function, and maps them to equipment power time series. The planning results output module calculates the operating cost based on price parameters, power time series and source allocation, and superimposes the total allocation to the interval net stock change to obtain the post-planning scale curve, and outputs the allocation list, scheduling curve, cost indicators and post-planning scale curve.

2. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 1, characterized in that, Determine the start and end times of the plan, define the closed interval between the start and end times as the planning period, and calculate the duration of the planning period by subtracting the start time from the end time. Define a time aggregation window, calculate the natural logarithm of the time aggregation window to obtain a dimensionless scaling variable, and establish an inverse exponential function mapping relationship from the dimensionless scaling variable to the time aggregation window; Set a minimum time aggregation window and a maximum time aggregation window. The minimum time aggregation window is greater than zero, and the maximum time aggregation window is no greater than the duration of the planning period. The minimum time aggregation window is no greater than the maximum time aggregation window. Calculate the natural logarithm of the minimum time aggregation window to obtain the lower bound of the scale domain, and calculate the natural logarithm of the maximum time aggregation window to obtain the upper bound of the scale domain. The closed interval from the lower bound of the scale domain to the upper bound of the scale domain is defined as the scale domain.

3. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 2, characterized in that, Select the time aggregation window corresponding to the planning caliber and the time aggregation window corresponding to the reference caliber. The time aggregation window corresponding to the planning caliber and the time aggregation window corresponding to the reference caliber are both not less than the minimum time aggregation window and neither are greater than the maximum time aggregation window. Calculate the natural logarithm of the time aggregation window corresponding to the planning caliber to obtain the planning caliber. Calculate the natural logarithm of the time aggregation window corresponding to the reference caliber to obtain the reference caliber. The planning period, planning start time, planning end time, and duration of the planning period are grouped into time boundary parameters, while the dimensionless scale variables, lower bound of the scale domain, upper bound of the scale domain, planning scope, reference scope, natural logarithm calculation relationship, and inverse mapping relationship of exponential function are grouped into scale parameters.

4. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 1, characterized in that, Configure a scale parameterization model, input the exogenous feature vectors within the planning period and the dimensionless scale variables within the scale domain into the scale parameterization model, and output the instantaneous net flux prediction corresponding to the planning period and the scale domain; Based on the inverse mapping relationship of the exponential function from the dimensionless scaling variable to the time aggregation window, the time aggregation window corresponding to the dimensionless scaling variable is calculated. The discrete time point sequence is generated within the planning period with the time aggregation window as the interval, and the time aggregation window is set as the numerical integration weight. The instantaneous net flux prediction at discrete time points is multiplied by the numerical integral weight, and the product is accumulated over the planning period to obtain the interval net stock change.

5. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 4, characterized in that, Within the scale domain, select a larger scale point and a smaller scale point located on either side of the current scale point. Calculate the natural logarithm of the absolute value of the change in net stock within the interval corresponding to the larger scale point, calculate the natural logarithm of the absolute value of the change in net stock within the interval corresponding to the smaller scale point, calculate the difference between the two natural logarithms, calculate the difference between the larger scale point and the smaller scale point, and divide the difference between the two natural logarithms by the difference between the larger scale point and the smaller scale point to obtain the metabolic integral scale elasticity of the current scale point.

6. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 1, characterized in that, Calculate the negative exponential function value of the absolute value of the elasticity of the metabolic integral scale, perform definite integral calculation on the negative exponential function value in the scale domain to obtain the normalization factor, and divide the negative exponential function value by the normalization factor to obtain the non-negative normalized scale weight kernel. Multiply the scale weight kernel by the net change in stock in the interval, and perform definite integral calculation on the result in the scale domain to obtain the flexible weighted effective budget corresponding to the planning scope.

7. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 1, characterized in that, Define the allocation quantities for each source, calculate the sum of the allocation quantities for each source to obtain the total allocation quantity; perform addition calculation on the total allocation quantity and the change in net stock in the interval to obtain the change in net stock in the interval after planning; calculate the difference between the change in net stock in the interval after planning and the flexible weighted effective budget, perform squaring calculation on the difference, multiply the result of the squaring calculation with the scale weight kernel, and perform definite integral calculation on the result of the multiplication in the scale domain to obtain the scale robust term.

8. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 7, characterized in that, The energy cost is obtained by multiplying the total allocation by the energy cost coefficient per unit allocation. The energy cost coefficient per unit allocation is the ratio of the energy cost obtained by integrating the product of the time-varying energy price parameter and the equipment power time series over the planning period to the total allocation. The source cost is obtained by multiplying the allocation of each source by the corresponding unit cost parameter and summing the results. The objective function is established by multiplying the scale robustness term by the trade-off coefficient and adding the result to the energy cost and the source cost. Set upper limit parameters for the allocation of various sources. Under the constraint that the allocation of each source is not less than zero and not greater than the corresponding upper limit parameter, perform a minimization solution on the objective function to obtain the optimal values ​​of the allocation of each source and the total allocation.

9. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 1, characterized in that, The instantaneous net flux prediction under the reference caliber is called by the scale parameterization model. The larger value between the instantaneous net flux prediction and the zero value is selected to obtain the non-negative flux. The non-negative flux is calculated by definite integral over the planning period to obtain the normalization factor. The non-negative flux is divided by the normalization factor to obtain the shape function. The total allocation is multiplied by the shape function to obtain the scheduling curve. Based on the equipment energy consumption function, the scheduling curve is subjected to function mapping calculation to obtain the equipment power timing.

10. The smart city resource flow simulation and planning system based on the urban metabolism model according to claim 1, characterized in that, The time-varying energy price parameter and the equipment power time series are called, and the time-varying energy price parameter and the equipment power time series are multiplied. The product result is then integrated within the planning period to obtain the energy cost. The energy cost coefficient per unit of allocation is calculated by the ratio of energy cost to total allocation. When the total allocation is zero, the energy cost coefficient per unit of allocation is zero. The system retrieves the source unit cost parameters and various source allocation quantities, multiplies each source allocation quantity with its corresponding source unit cost parameter, and sums the products to obtain the source cost. It then adds the energy cost to the source cost to obtain the total operating cost, and aggregates the energy cost, source cost, and operating cost into a cost indicator. Finally, it retrieves the interval net inventory change and total allocation quantity, overlaying the total allocation quantity onto the interval net inventory change to obtain the post-planning scale curve. The system then aggregates various source allocation quantities into an allocation list and outputs the allocation list, scheduling curve, cost indicator, and post-planning scale curve.

Images