An urban comprehensive flexible resource regulation system and method based on multi-spatiotemporal scale technology

Abstract

The present invention discloses an urban comprehensive flexible resource regulation system and method based on multi-temporal and spatial scale technology. The system includes a multi-temporal and spatial scale module, a flexible resource set integration module, an energy input module, a city module, and a cloud platform module. The multi-temporal and spatial scale module, the flexible resource set integration module, the energy input module, the city module, and the cloud platform module are sequentially connected to form a closed loop. The present invention optimizes the energy purchase cost and equipment operation cost through an optimized convolutional optimization algorithm (COA), and realizes the lowest total cost of the regulation strategy for the optimal supply mode of urban comprehensive flexible resources.

Description

Technical Field

[0001] The present invention relates to the technical fields of multi - spatio - temporal scales and cloud computing, and particularly relates to an urban comprehensive flexible resource regulation system and method based on multi - spatio - temporal scale technology. Background Art

[0002] In the electricity spot market, the operation and management of the urban regional integrated energy system face many challenges. First, with the rapid development of urbanization, the energy demand is continuously increasing, while the energy supply is restricted by various factors, such as limited resources and difficult transportation. Second, with the development of technology, the proportion of renewable energy in the energy structure is gradually increasing, but due to its randomness and volatility, it brings certain risks to the stable supply of energy. In addition, the traditional energy management mode has been difficult to meet the needs of modern cities, and a more refined and intelligent management method is required to improve the energy utilization efficiency and environmental protection level.

[0003] To address these challenges, multi - spatio - temporal scale technology has been proposed to construct an urban regional integrated energy flexible resource set. This technology can comprehensively consider the energy demand and supply situations at different time scales, and through optimizing and coordinating various energy resources, achieve efficient, safe and environmentally friendly energy supply. In the electricity spot market, multi - spatio - temporal scale technology can help market participants better predict and manage their own energy demand and supply, and avoid risks brought by market fluctuations. At the same time, multi - spatio - temporal scale technology can also promote the development of renewable energy, improve the energy utilization efficiency, and promote the sustainable development of cities. Summary of the Invention

[0004] Object of the Invention: The object of the present invention is to provide an urban comprehensive flexible resource regulation system and method based on multi - spatio - temporal scale technology. On the one hand, it comprehensively utilizes various types of flexible resources to achieve optimal regulation of resources; on the other hand, it finds an optimal supply mode to ensure the minimization of energy costs while meeting the urban energy demand.

[0005] Technical Solution: An urban comprehensive flexible resource regulation system based on multi - spatio - temporal scale technology according to the present invention includes a multi - spatio - temporal scale module, a flexible resource set integration module, an energy input module, a city module and a cloud platform module; the multi - spatio - temporal scale module, the flexible resource set integration module, the energy input module, the city module and the cloud platform module are connected in sequence to form a closed loop.

[0006] Furthermore, the multi - spatio - temporal scale module includes a time - period division module, a short - term prediction module, a medium - and long - term prediction module, and an optimal scheduling module. Among them, the time - period division module divides a day into different time periods according to the changing rules of energy demand and supply, and conducts separate optimization and coordination in different time periods. The short - term prediction module predicts the energy demand and supply conditions in the next one or several days, and uses prediction models and technical means to make a relatively accurate prediction of the future energy demand and supply conditions. The medium - and long - term prediction module makes a trend prediction of the energy demand and supply conditions in the future for a relatively long time by analyzing information such as historical trends and technological development trends. The optimal scheduling module formulates an optimal scheduling plan according to the energy demand and supply conditions in different time periods to ensure the efficient utilization and conservation of energy.

[0007] Furthermore, the flexible resource set integration module includes: a combined heat, power, and cooling module, an air - source heat pump module, an electric refrigeration and heating module, and a cold and heat energy storage module, which integrates the comprehensive flexible resources of the city. Among them, the combined heat, power, and cooling module uses the technology of generating thermal energy, steam, or electric energy from gas, and through reasonable configuration and optimization, realizes the cascade utilization of energy. At the same time, the combined heat, power, and cooling module can be flexibly adjusted according to demand to meet the energy needs of different users. The air - source heat pump module uses the heat energy in the air to generate hot water. At the same time, it can be flexibly adjusted according to demand to meet the hot water needs of different users. The electric refrigeration and heating module uses electric energy to generate cooling and heating effects, and through the circulation of the refrigerant, discharges the heat in the room. At the same time, it provides hot water or heating services through electric heating. The cold and heat energy storage module stores energy using substances and charges during nighttime or low - electricity - price periods; and releases energy during daytime or high - electricity - price periods.

[0008] Furthermore, the energy input module includes: a heat source generation module, a cold source generation module, and an energy storage device.

[0009] Furthermore, the cloud platform module collects and analyzes the urban energy demand data and conducts cloud computing. The cloud computing transfers the optimized structure to the multi - spatio - temporal scale module, and through the multi - spatio - temporal scale module, controls the flexible resource set integration module to execute the optimal regulation plan, driving the energy input module to supply heat and cooling to the urban module.

[0010] A method for regulating comprehensive flexible resources in a city based on multi - spatio - temporal scale technology according to the present invention uses the improved convolutional optimization algorithm COA for cloud computing, and includes the following steps:

[0011] (1) Construct an optimization objective function, that is, the minimum value of the energy purchase cost and the equipment operation cost;

[0012] (2) Initialize the population;

[0013] (3) Update the vertical convolution position;

[0014] (4) Update the horizontal convolution position and the regional convolution position;

[0015] (5) Update the comprehensive position and enhance the solution quality.

[0016] Furthermore, in step (1), the formula is as follows:

[0017] O min = C b + C f

[0018] where C b is the energy purchase cost, and C f is the equipment operation cost;

[0019] C b = C b.e + C b.g

[0020] C b.e = ∑P e.grid (t)μ e (t)△t

[0021] C b.g = ∑P g.grid (t)μ g (t)△t

[0022] where C b.e is the electricity purchase cost within the scheduling period, C b.g is the gas purchase cost within the scheduling period, P e.grid (t) is the electricity purchase power at time t, μ e (t) is the electricity purchase price at time t, △t is the scheduling time interval, P g.grid (t) is the gas purchase power at time t, μ g (t) is the gas purchase price at time t;

[0023] C f = C f.ess + C f.hss + C f.css + C f.e.c + C f.e.h + C f.bl

[0024] C f.ess = ∑|P ess.c (t) + P ess.d (t)|υ ess △t

[0025] C f.hss = ∑|P hss.c (t) + P hss.d (t)|υ hss △t

[0026] C f.css = ∑|P css.c (t) + P css.d (t)|υ css △t

[0027] C f.e.c = ∑|P e.c (t)|υ e.c △t

[0028] C f.e.h = ∑|P e.h (t)|υ e.h △t

[0029] C f.bl = ∑|P bl (t)|υ bl △t

[0030] Among them, C f.ess is the operation and maintenance cost of the electrical energy storage system during the scheduling period, C f.hss is the operation and maintenance cost of the thermal energy storage system during the scheduling period, C f.css is the operation and maintenance cost of the cold energy storage system during the scheduling period, C f.e.c is the operation and maintenance cost of the refrigeration unit during the scheduling period, C f.e.h is the operation and maintenance cost of the electric heating equipment during the scheduling period, C f.bl is the operation and maintenance cost of the gas boiler during the scheduling period, P ess.c (t) is the charging power of the electrical energy storage system at time t, P ess.d (t) is the discharging power of the electrical energy storage system at time t, υ ess is the operation and maintenance cost coefficient of the electrical energy storage system, P hss.c (t) is the heat charging power of the thermal energy storage system at time t, P hss.d (t) is the heat discharging power of the thermal energy storage system at time t, υ hss is the operation and maintenance cost coefficient of the thermal energy storage system, P css.c (t) is the energy charging power of the cold energy storage system at time t, P css.d (t) is the energy discharging power of the cold energy storage system at time t, υ css is the operation and maintenance cost coefficient of the cold energy storage system, P e.c (t) is the electric power of the refrigeration unit at time t, υ e.c is the operation and maintenance cost coefficient of the refrigeration unit, P e.h (t) is the electric power of the electric heating equipment at time t, υ e.h is the operation and maintenance cost coefficient of the electric heating equipment, P bl (t) is the power of the gas boiler at time t, υ bl is the operation and maintenance cost coefficient of the gas boiler.

[0031] Furthermore, step (2) is specifically as follows:

[0032] Let the position vector O of the population consist of n individuals with dimension d, and the mathematical expression is as follows:

[0033]

[0034] where the position vector O of the individual x (x = 1, 2, …, n) is a candidate solution to the optimization problem, and O is defined x for searching in the d-dimensional space, where d is the dimension of the decision variable;

[0035] The fitness of the position vector O of the population is:

[0036]

[0037] where f(O x ) represents the fitness function;

[0038] The position vector O of the initial population 0 is randomly generated in the d-dimensional search space, and the initialization of the position vector of each individual is defined as:

[0039]

[0040] where l x is a 1×d matrix, which is the lower limit of the x-th individual, and u x is a 1×d matrix, which is the upper limit of the x-th individual, and rand is a random number between [0, 1].

[0041] Furthermore, in step (3), the specific update of the vertical convolution position is as follows:

[0042] Define the vertical convolution kernel as:

[0043] K L = 2×rand(k, 1) - I L

[0044] where K L is a k×1 matrix, which is the vertical convolution kernel, where k is the height of the vertical convolution kernel and 1 is the width of the vertical convolution kernel, rand(k, 1) is a k×1 matrix, and each element is a random number between [0, 1], and I L is a k×1 matrix, and all elements are 1;

[0045] Define the vertical convolution as:

[0046]

[0047] where h is the current iteration number, Oh is an \(n\times d\) matrix, and is the position vector of the \(h\)-th generation population. is an \(n\times d\) matrix, and is the position vector of the population after the vertical convolution position update in the \(h\)-th generation;

[0048] Compare and \(O\) h the fitness values of each individual position in, and preferentially replace the individual positions in \(O\) h the individual positions in, then there is

[0049]

[0050] In the formula, is the \(p\)-th individual position of the \(h\)-th generation population, is the \(p\)-th individual position of the population after the vertical convolution position update in the \(h\)-th generation; is the local optimal solution after the Zining vertical position update, that is, the regulation strategy for the minimum value of the sum of the local energy purchase cost and the equipment operation cost.

[0051] Furthermore, in step (4), the update of the horizontal convolution position is specifically:

[0052] Define the horizontal convolution kernel as:

[0053] \(K\) T \(= 2\times rand(k, 1)-I\) T

[0054] In the formula, \(K\) T is a \(1\times k\) matrix, which is the horizontal convolution kernel, where 1 is the height of the horizontal convolution kernel and \(k\) is the width of the horizontal convolution kernel, \(rand(k, 1)\) is a \(1\times k\) matrix, and each element is a random number between \([0, 1]\), \(I\) T is a \(1\times k\) matrix, and all elements are 1;

[0055] Define the horizontal convolution as:

[0056]

[0057] In the formula, is an \(n\times d\) matrix, which is the position vector of the population after the horizontal convolution update;

[0058] Compare and \(O\) h the fitness values of each individual position in, and preferentially replace the individual positions in \(O\) h the individual positions in, then there is:

[0059]

[0060] In the formula, The position of the p-th individual in the population after the h-th generation of horizontal convolution position update; That is, after the horizontal convolution position update, the local optimal solution, that is, the regulation strategy of the minimum value of the sum of the local energy purchase cost and the equipment operation cost;

[0061] The specific update of the regional convolution position is as follows:

[0062] Define the regional convolution kernel:

[0063] K R = 2×rand(k,k) - I R

[0064] In the formula, K R is a k×k matrix, which is the regional convolution kernel, where k is the height and width of the regional convolution kernel, rand(k,k) is a k×k matrix, and each element is a random number between [0,1], and I R is a k×k matrix, and all elements are 1.

[0065] Define the regional convolution as:

[0066]

[0067] In the formula, is an n×d matrix, which is the position vector of the population after the regional convolution update.

[0068] Compare and O h the fitness values of each individual position in, and preferentially replace the individual positions in O h Then there are:

[0069]

[0070] In the formula, is the position of the p-th individual in the population after the h-th generation of regional convolution position update. That is, after the regional convolution update, the local optimal solution, that is, the regulation strategy of the minimum value of the sum of the local energy purchase cost and the equipment operation cost.

[0071] Furthermore, in step (4), the comprehensive position update is specifically: the position vector of the population after the h-th generation of vertical convolution update the position vector of the population after the h-th generation of horizontal convolution update and the position vector of the population after the h-th generation of regional convolution update are added using random weights or equal proportion weights, and for its mathematical expression is as follows:

[0072]

[0073] where r 1 , r 2 , r 3 are all random numbers between [0, 1]. Additionally, let r 1 = r 2 = r 3 , and perform equal - ratio weighted addition.

[0074] Compare and the fitness values of each individual position in O h . Selectively replace the individual positions in O h . Then we have:

[0075]

[0076] where is the p - th individual position in the population after the h - th generation comprehensive position update;

[0077] Finally, calculate the fitness values of all individual positions in O h , and sort them according to the magnitude of the fitness values to select the optimal solution Sort and screen the above - mentioned local optimal solutions according to the fitness, and select the most global optimal solution;

[0078] Enhance the solution quality as follows: perform Gaussian mutation with non - inertial weight on each dimension of the d - dimensional search space of the optimal solution , and perturb the optimal solution . The formula is as follows:

[0079]

[0080] where is an n×1 matrix, and is the position of the q - th (q = 1, 2, …, d) dimension in the d - dimensional search space of the optimal solution . In max , iher is the maximum number of iterations, randn is a random number that follows a standard normal distribution with a mean of 0 and a variance of 1, is an n×1 matrix, and

[0081] Let the individual position after Gaussian mutation with non - inertial weight on the q - th dimension be Compare and in terms of their fitness values, and selectively replace the individual position of . Then we have:

[0082] ​

[0083] The finally output is the optimal regulation scheme of the total operation cost in the iterative result.

[0084] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: The present invention uses an urban comprehensive flexible resource regulation system to improve the optimization and upgrading of the urban energy structure, providing important support and guarantee for the sustainable development of the city; Compared with the traditional energy management mode, the present invention combines multi-temporal and spatial scale technologies to help the urban regional integrated energy system achieve more refined and intelligent management and scheduling, improving the energy utilization efficiency and conservation level; The present invention also optimizes the energy purchase cost and equipment operation cost through the optimized convolutional optimization algorithm (COA), and realizes the lowest total cost of the optimal supply mode regulation strategy of urban comprehensive flexible resources. Description of the Drawings

[0085] Figure 1 is a schematic structural diagram of the present invention;

[0086] Figure 2 is a comparison diagram of the energy utilization efficiency before and after optimization of the present invention;

[0087] Figure 3 is a comparison diagram of the operation cost of the system before and after optimization of the present invention. Detailed Embodiments

[0088] The technical solution of the present invention will be further described below with reference to the drawings.

[0089] As Figure 1 shown, an embodiment of the present invention provides an urban comprehensive flexible resource regulation system based on multi-temporal and spatial scale technologies, including a multi-temporal and spatial scale module, a flexible resource set integration module, an energy input module, a city module, and a cloud platform module; the multi-temporal and spatial scale module, the flexible resource set integration module, the energy input module, the city module, and the cloud platform module are connected in sequence to form a closed loop.

[0090] Among them, the multi-temporal and spatial scale module includes a time period division module, a short-term prediction module, a medium- and long-term prediction module, and an optimization scheduling module; among them, the time period division divides a day into different time periods according to the change rules of energy demand and supply, and performs separate optimization and coordination in different time periods. The short-term prediction module predicts the energy demand and supply situation within the next day or several days, and uses prediction models and technical means to make a relatively accurate prediction of the future energy demand and supply situation; The medium- and long-term prediction module makes a trend prediction of the energy demand and supply situation within a relatively long time in the future by analyzing historical trends and technological development trends and other information; The optimization scheduling module formulates an optimization scheduling plan according to the energy demand and supply situation in different time periods to ensure the efficient utilization and conservation of energy.

[0091] The flexible resource set integration module includes: a combined heat and power module, an air source heat pump module, an electric refrigeration and heating module, and a cold and heat storage module, which integrate the urban comprehensive flexible resources. Among them, the combined heat and power module uses the technology of generating heat energy, steam or electric energy from gas, and through reasonable configuration and optimization, realizes the cascade utilization of energy. At the same time, the combined heat and power module is flexibly adjusted according to the demand to meet the energy needs of different users. The air source heat pump module uses the heat energy in the air to generate hot water. At the same time, it is flexibly adjusted according to the demand to meet the hot water needs of different users. The electric refrigeration and heating module uses electric energy to generate cold and heat effects, and discharges the indoor heat through the circulation of the refrigerant. At the same time, through the way of electric heating, it provides hot water or heating services. The cold and heat storage module stores energy by substances and charges during the night or low electricity price periods, and releases energy during the day or high electricity price periods.

[0092] The energy input module includes: a heat source generation module, a cold source generation module and an energy storage device.

[0093] The cloud platform module collects and analyzes the urban energy demand data and performs cloud computing. The cloud computing transfers the optimized structure to the multi-temporal and spatial scale module, and through the multi-temporal and spatial scale module, controls the flexible resource set integration module to execute the optimal operation regulation plan, and drives the energy input module to supply heat and cold to the urban module.

[0094] The embodiment of the present invention also provides a method for regulating urban comprehensive flexible resources based on multi-temporal and spatial scale technology, which uses the improved convolutional optimization algorithm COA for cloud computing, and includes the following steps:

[0095] (1) Construct an optimization objective function, that is, the minimum value of the energy purchase cost and the equipment operation cost; the formula is as follows:

[0096] O min =C b +C f

[0097] Wherein, C b is the energy purchase cost, and C f is the equipment operation cost;

[0098] C b =C b.e +C b.g

[0099] C b.e =∑P e.grid (t)μ e (t)△t

[0100] C b.g =∑P g.grid (t)μ g (t)△t

[0101] Among them, C b.e is the electricity purchase cost within the scheduling period, and C b.g is the gas purchase cost within the scheduling period. P e.grid (t) is the electricity purchase power at time t, and μ e (t) is the electricity purchase price at time t. △t is the scheduling time interval. P g.grid (t) is the gas purchase power at time t, and μ g (t) is the gas purchase price at time t;

[0102] C f = C f.ess + C f.hss + C f.css + C f.e.c + C f.e.h + C f.bl

[0103] C f.ess = ∑|P ess.c (t) + P ess.d (t)|υ ess △t

[0104] C f.hss = ∑|P hss.c (t) + P hss.d (t)|υ hss △t

[0105] C f.css = ∑|P css.c (t) + P css.d (t)|υ css △t

[0106] C f.e.c = ∑|P e.c (t)|υ e.c △t

[0107] C f.e.h = ∑|P e.h (t)|υ e.h △t

[0108] C f.bl = ∑|P bl (t)|υ bl △t

[0109] Among them, C f.ess is the operation and maintenance cost of the electric energy storage system within the scheduling period, C f.hss is the operation and maintenance cost of the thermal energy storage system within the scheduling period, C f.css is the operation and maintenance cost of the cold energy storage system within the scheduling period, C f.e.c is the operation and maintenance cost of the refrigeration unit within the scheduling period, C f.e.hThe operation and maintenance cost of the electric heating equipment within the scheduling period, C f.bl The operation and maintenance cost of the gas boiler within the scheduling period, P ess.c (t) is the charging power of the electric energy storage system at time t, P ess.d (t) is the discharging power of the electric energy storage system at time t, υ ess The operation and maintenance cost coefficient of the electric energy storage system, P hss.c (t) is the heat charging power of the thermal energy storage system at time t, P hss.d (t) is the heat discharging power of the thermal energy storage system at time t, υ hss The operation and maintenance cost coefficient of the thermal energy storage system, P css.c (t) is the energy charging power of the cold energy storage system at time t, P css.d (t) is the energy discharging power of the cold energy storage system at time t, υ css The operation and maintenance cost coefficient of the cold energy storage system, P e.c (t) is the electric power of the refrigeration unit at time t, υ e.c The operation and maintenance cost coefficient of the refrigeration unit, P e.h (t) is the electric power of the electric heating equipment at time t, υ e.h The operation and maintenance cost coefficient of the electric heating equipment, P bl (t) is the power of the gas boiler at time t, υ bl The operation and maintenance cost coefficient of the gas boiler.

[0110] (2) Initialize the population; assume that the position vector O of the population consists of n individuals with a dimension of d, and the mathematical expression is as follows:

[0111]

[0112] In the formula, the position vector O of the individual x (x = 1, 2,..., n) is the candidate solution of the optimization problem, and O is defined x For searching in the d-dimensional space, where d is the dimension of the decision variable; this process is used to initialize and generate various supply strategies for the total operating cost within the regulation period that meets the production needs of the industrial park, and O x (x = 1, 2,..., n) is equivalent to any regulation strategy that can be optimized.

[0113] The fitness of the position vector O of the population is:

[0114]

[0115] In the formula, f(O x ) represents the fitness function; the addition of the fitness function can screen out the better supply strategies from various supply strategies.

[0116] The position vector O of the initial population 0Randomly generated in the d-dimensional search space, the position vector of each individual is initialized as follows:

[0117]

[0118] where l x is a 1×d matrix, is the lower limit of the x-th individual, u x is a 1×d matrix, is the upper limit of the x-th individual, and rand is a random number between [0,1].

[0119] (3) Update the vertical convolution position; specifically as follows: The update of the vertical convolution position is specifically as follows:

[0120] Define the vertical convolution kernel as:

[0121] K L = 2×rand(k,1) - I L

[0122] where K L is a k×1 matrix, is the vertical convolution kernel, where k is the height of the vertical convolution kernel and 1 is the width of the vertical convolution kernel, rand(k,1) is a k×1 matrix, and each element is a random number between [0,1], and I L is a k×1 matrix, and all elements are 1;

[0123] Define the vertical convolution as:

[0124]

[0125] where h is the current iteration number, O h is an n×d matrix, is the position vector of the h-th generation population, is an n×d matrix, is the position vector of the population after the vertical convolution position update in the h-th generation; here, the vertical position update is performed after iterating the individual positions, that is, the optimal supply mode regulation strategy of urban comprehensive flexible resources.

[0126] Compare and O h in terms of the fitness values of each individual position, and preferentially replace the individual positions in O h with the better ones, then there is

[0127]

[0128] where is the p-th individual position in the h-th generation population, is the p-th individual position in the population after the vertical convolution position update in the h-th generation; That is, after the vertical position of Zining is updated, it is the local optimal solution, that is, the control strategy with the minimum sum of local energy purchase cost and equipment operation cost. That is, after the vertical position of Zining is updated, it is the local optimal solution, that is, the control strategy with the minimum sum of local energy purchase cost and equipment operation cost.

[0129] (4) Update the horizontal convolution position and the regional convolution position; The specific update of the horizontal convolution position is as follows:

[0130] Define the horizontal convolution kernel as:

[0131] K T = 2×rand(k,1) - I T

[0132] In the formula, K T is a 1×k matrix, which is the horizontal convolution kernel, where 1 is the height of the horizontal convolution kernel and k is the width of the horizontal convolution kernel, rand(k,1) is a 1×k matrix, and each element is a random number between [0,1], and I T is a 1×k matrix, and all elements are 1;

[0133] Define the horizontal convolution as:

[0134]

[0135] In the formula, is an n×d matrix, which is the position vector of the population after the horizontal convolution update;

[0136] Compare and O h the fitness values of each individual position in, and preferentially replace the individual positions in O h with the individual positions in, then there is:

[0137]

[0138] In the formula, the p-th individual position of the population after the horizontal convolution position update in the h-th generation; That is, after the horizontal convolution position is updated, it is the local optimal solution, that is, the control strategy with the minimum sum of local energy purchase cost and equipment operation cost; That is, after the horizontal convolution position is updated, it is the local optimal solution, that is, the control strategy with the minimum sum of local energy purchase cost and equipment operation cost.

[0139] The specific update of the regional convolution position is as follows:

[0140] Define the regional convolution kernel:

[0141] K R= 2 × rand(k, k) - I R

[0142] Where K R is a k×k matrix, which is the regional convolution kernel, where k is the height and width of the regional convolution kernel, rand(k, k) is a k×k matrix, and each element is a random number between [0, 1], and I R is a k×k matrix, and all elements are 1.

[0143] Define the regional convolution as:

[0144]

[0145] Where is an n×d matrix, which is the position vector of the population after the regional convolution update.

[0146] Compare and O h for the fitness value of each individual position, and preferentially replace the individual positions in O h Then there is:

[0147]

[0148] Where is the p-th individual position of the population after the h-th generation regional convolution position update. is the local optimal solution after the regional convolution update, that is, the regulation strategy for the minimum value of the sum of the local energy purchase cost and the equipment operation cost. is the local optimal solution after the regional convolution update, that is, the regulation strategy for the minimum value of the sum of the local energy purchase cost and the equipment operation cost.

[0149] (5) Update the comprehensive position and enhance the solution quality. The specific update of the comprehensive position is as follows: Combine the position vector of the population after the h-th generation vertical convolution update the position vector of the population after the h-th generation horizontal convolution update and the position vector of the population after the h-th generation regional convolution update using random weights or equal-proportion weights for addition, and for its mathematical expression is as follows:

[0150]

[0151] Where r 1 、r 2 、r 3 are all random numbers between [0, 1], and another r 1 = r 2 = r 3 , and perform addition with equal-proportion weights.

[0152] Compare with O h the fitness values of each individual position in, and preferentially replace O h in the individual positions, then there is:

[0153]

[0154] In the formula, is the p-th individual position of the population after the comprehensive position update in the h-th generation; here, through iteration, within a local range, the optimal regulation strategy after the comprehensive position update is sought.

[0155] Finally, calculate the fitness values of all individual positions in O h and sort them according to the magnitudes of the fitness values to select the optimal solution Sort and screen the above local optimal solutions according to the fitness, and select the most global optimal solution; sort and screen the above local optimal solutions according to the fitness, and select the most global optimal solution.

[0156] Enhance the solution quality as follows: perform Gaussian mutation with non-inertial weight on each dimension of the d-dimensional search space of the optimal solution and perturb the optimal solution ; the formula is as follows:

[0157]

[0158] In the formula, is an n×1 matrix, and is the position of the q-th (q = 1, 2,..., d) dimension in the d-dimensional search space of the optimal solution ; in iher max is the maximum number of iterations, randn is a random number that follows a standard normal distribution with a mean of 0 and a variance of 1, is an n×1 matrix, and is the q-th dimension position after performing Gaussian mutation with non-inertial weight on the q-th dimension of the optimal solution ; through the solution quality enhancement mechanism, further search for the positions around the optimal solution to seek whether there is a better regulation strategy, that is, a regulation strategy with a smaller sum of energy purchase cost and equipment operation cost.

[0159] Let the individual position after performing Gaussian mutation with non-inertial weight on the q-th dimension be Compare with the magnitudes of the fitness values of, and preferentially replace the individual position of then there is:

[0160]

[0161] The finally output is the optimal regulation plan for the total operation cost in the iterative results. That is, the sum of the energy purchase cost and the equipment operation cost is minimized, which is the globally optimal regulation strategy.

[0162] Figure 2 is a comparison chart of the energy utilization efficiency before and after optimization; the system can comprehensively consider the source, time, spatial distribution and demand of energy, and realize the optimal allocation and scheduling of energy. This can not only reduce the loss of energy in the transmission and conversion processes, but also improve the operation efficiency of the equipment, thus significantly enhancing the energy utilization efficiency. In Figure 2 , the energy utilization efficiency after optimization shows a stable upward trend and is generally high, demonstrating the effectiveness and superiority of the entire system.

[0163] Figure 3 is a comparison chart of the operation cost of the system before and after optimization: by adopting the optimization strategy, the system can reduce the energy and equipment maintenance costs and minimize the total cost. In Figure 3 , the energy purchase cost and equipment maintenance cost of the optimized system show a stable downward trend and are generally low, demonstrating the remarkable effect of the system in reducing the system operation cost.

Claims

1. An urban comprehensive flexible resource control system based on multi-temporal and spatial scale technology, characterized in that: It includes a multi-space-time scale module, a flexible resource set integration module, an energy input module, a city module and a cloud platform module; the multi-space-time scale module, the flexible resource set integration module, the energy input module, the city module and the cloud platform module are connected in sequence to form a closed loop; the multi-space-time scale module includes a time division module, a short-term prediction module, a medium- and long-term prediction module and an optimization scheduling module; among them, the time division divides a day into different time periods according to the changing law of energy demand and supply, and performs optimization and coordination separately in different time periods. The short-term prediction module predicts the energy demand and supply situation in the next day or a few days, and uses prediction models and technical means to make a more accurate prediction of the future energy demand and supply situation; the medium- and long-term prediction module makes a trend prediction of the energy demand and supply situation in the future for a long time by analyzing historical trends and technological development trends; the optimization scheduling module formulates an optimization scheduling plan according to the energy demand and supply situation in different time periods to ensure efficient utilization and conservation of energy; the flexible resource set integration module includes: trigeneration module, air source heat pump module, electric cooling and heating module, cold storage and heat storage module, which is used for comprehensive flexible resources in the city. The energy input module includes: a heating source module, a cooling source module and an energy storage device; the cloud platform module collects and analyzes the urban energy demand data and performs cloud computing, and the cloud computing transmits the optimized structure to the multi-time and space scale module, and controls the flexible resource set integration module to execute the optimal operation control scheme through the multi-time and space scale module, and drives the energy input module to heat and cool the urban module; wherein, the improved convolution optimization algorithm COA is used for cloud computing, including the following steps: (1) Constructing the optimization objective function, i.e., the minimum value of energy purchase cost and equipment operation cost; (2) Initialize the population; (3) Update the vertical convolution position; (4) Update the horizontal convolution position and the regional convolution position; (5) Update the comprehensive position and enhance the solution quality.

2. According to claim 1, a comprehensive urban flexible resource control system based on multi-temporal and spatial scale technology is characterized in that: In cloud computing using the improved convolution optimization algorithm COA, the formula for step (1) is as follows: O min =C b +C f Among them, C b is the energy purchase cost, C f Equipment operating costs; C b =C b.e +C b.g C b.e =∑P e.grid (t)μ e (t)△t C b.g =∑P g.grid (t)μ g (t)△t Among them, C b.e is the electricity purchase cost within the dispatch period, C b.g is the gas purchase cost within the dispatch period, P e.grid (t) is the power purchased at time t, μ e (t) is the electricity purchase price at time t, △t is the dispatching time interval, P g.grid (t) is the gas purchase power at time t, μ g (t) is the gas purchase price at time t; C f =C f.ess +C f.hss +C f.css +C f.e.c +C f.e.h +C f.bl C f.ess =∑|P ess.c (t)+P ess.d (t)|υ ess △t C f.hss =∑|P hss.c (t)+P hss.d (t)|υ hss △t C f.css =∑|P css.c (t)+P css.d (t)|υ css △t C f.e.c =∑|P e.c (t)|υ e.c △t C f.e.h =∑|P e.h (t)|υ e.h △t C f.bl =∑|P bl (t)|υ bl △t Among them, C f.ess is the operation and maintenance cost of the energy storage system during the dispatch period, C f.hss is the operation and maintenance cost of the thermal energy storage system during the dispatch period, C f.css is the operation and maintenance cost of the cold energy storage system during the dispatch period, C f.e.c is the operation and maintenance cost of the refrigeration unit during the scheduling period, C f.e.h is the operation and maintenance cost of the electric heating equipment during the scheduling period, C f.bl is the gas boiler operation and maintenance cost within the dispatching period, P ess.c (t) is the charging power of the energy storage system at time t, P ess.d (t) is the discharge power of the energy storage system at time t, υ ess is the operation and maintenance cost coefficient of the energy storage system, P hss.c (t) is the charging power of the thermal energy storage system at time t, P hss.d (t) is the energy release power of the thermal energy storage system at time t, υ hss is the operation and maintenance cost coefficient of the thermal energy storage system, P css.c (t) is the charging power of the cold energy storage system at time t, P css.d (t) is the energy discharging power of the cold energy storage system at time t, υ css is the operation and maintenance cost coefficient of the cold energy storage system, P e.c (t) is the electrical power of the refrigeration unit at time t, υ e.c is the operation and maintenance cost coefficient of the refrigeration unit, P e.h (t) is the electric power of the electric heating equipment at time t, υ e.h is the operation and maintenance cost coefficient of the electric heating equipment, P bl (t) is the power of the gas boiler at time t, υ bl is the gas boiler operation and maintenance cost coefficient.

3. The urban comprehensive flexible resource control system based on multi-temporal and spatial scale technology according to claim 1 is characterized in that: In cloud computing using the improved convolution optimization algorithm COA, step (2) is as follows: Assume that the position vector O of the population consists of n individuals with dimension d. The mathematical expression is as follows: In the formula, the position vector O of the individual x (x=1,2,…,n) is the candidate solution of the optimization problem, and O is defined x Used to search in d-dimensional space, where d is the dimension of the decision variable; The fitness of the position vector O of the population is: In the formula, f(O x ) represents the fitness function; The position vector O of the initial population 0 Randomly generated in the d-dimensional search space, the position vector of each individual The initialization is defined as: In the formula, l x is a 1×d-order matrix, which is the lower limit of the x-th individual, u x is a 1×d-order matrix, is the upper limit of the x-th individual, and rand is a random number between [0,1].

4. The urban comprehensive flexible resource control system based on multi-temporal and spatial scale technology according to claim 1 is characterized in that: In cloud computing using the improved convolution optimization algorithm COA, step (3) updates the vertical convolution position as follows: Define the vertical convolution kernel as: K L =2×rand(k,1)-I L In the formula, K L is a k×1 matrix, which is the vertical convolution kernel, where k is the height of the vertical convolution kernel and 1 is the width of the vertical convolution kernel. rand(k,1) is a k×1 matrix, and each element is a random number between [0,1]. L is a k×1 matrix with all elements being 1; Define the vertical convolution as: Where h is the current iteration number, O h is an n×d matrix, is the position vector of the h-th generation population, is an n×d-order matrix, which is the position vector of the population after the h-th generation longitudinal convolution position update; Compare and O h The fitness value of each individual position in is replaced by O h In the individual position, we have In the formula, is the position of the pth individual in the hth generation population, It is the p-th individual position of the population after the h-th generation longitudinal convolution position update.

5. The urban comprehensive flexible resource control system based on multi-temporal and spatial scale technology according to claim 1 is characterized in that: In cloud computing using the improved convolution optimization algorithm COA, step (4) updates the horizontal convolution position as follows: Define the horizontal convolution kernel as: K T =2×rand(k,1)-I T In the formula, K T is a 1×k-order matrix, which is the horizontal convolution kernel, where 1 is the height of the horizontal convolution kernel, k is the width of the horizontal convolution kernel, rand(k,1) is a 1×k-order matrix, each element is a random number between [0,1], I T is a 1×k-order matrix with all elements being 1; Define the horizontal convolution as: In the formula, is an n×d matrix, which is the position vector of the population after the horizontal convolution update; Compare and O h The fitness value of each individual position in is replaced by O h In the individual position, we have: In the formula, The pth individual position of the population after the hth generation of lateral convolution position update; The update area convolution position is specifically: Define the regional convolution kernel: K R =2×rand(k,k)-I R In the formula, K R is a k×k matrix, which is the regional convolution kernel, where k is the height and width of the regional convolution kernel, rand(k,k) is a k×k matrix, each element of which is a random number between [0,1], I R is a k×k matrix with all elements being 1; Define the regional convolution as: In the formula, is an n×d-order matrix, which is the position vector of the population after regional convolution update; Compare and O h The fitness value of each individual position in is replaced by O h In the individual position, we have: In the formula, It is the p-th individual position of the population after the h-generation regional convolution position update.

6. The urban comprehensive flexible resource control system based on multi-temporal and spatial scale technology according to claim 1 is characterized in that: In cloud computing using the improved convolution optimization algorithm COA, step (5) of the comprehensive position update is specifically: the position vector of the population after the h-th generation of longitudinal convolution update is The position vector of the population after the h-th generation of horizontal convolution update And the position vector of the population after the h-th generation regional convolution update Use random weights or equal proportion weights to add, and Its mathematical expression is as follows: In the formula, r1, r2, and r3 are all random numbers between [0,1], and r1 = r2 = r3, and the weights are added in equal proportion; Compare and O h The fitness value of each individual position in is replaced by O h In the individual position, we have: In the formula, is the pth individual position of the population after the hth generation comprehensive position update; Finally, calculate O h The fitness values ​​of all individual positions in the , and sort them according to the size of the fitness value to select the optimal solution Sort and filter according to fitness and select the most global optimal solution; The specific details of enhancing the solution quality are as follows: The d-dimensional search space is Gaussian mutated with non-inertial weights dimension by dimension to find the optimal solution Perform disturbance; the formula is as follows: In the formula, is an n×1 matrix, which is the optimal solution The position of the qth (q=1,2,…,d)th dimension in the d-dimensional search space, in,iher max is the maximum number of iterations, randn is a random number that satisfies the standard normal distribution with a mean of 0 and a variance of 1; Let the individual position after Gaussian mutation with non-inertial weights on the qth dimension be Compare and The size of the fitness value, replace it with the best one The individual positions of are: The final output That is the optimal control plan for the total running time in the iteration result.

Images