Non-intrusive load decomposition method and system

Abstract

The invention discloses a non-intrusive load decomposition method and system, and the method is characterized in that the method comprises the following steps: obtaining a flexible load reduction amount expression, and building a load decomposition model with the minimum scheduling deviation as a target and the falling of the flexible load reduction amount into a comfort degree range as a constraint; and receiving a regulation and control instruction sent by the power grid, collecting historical loads, and solving the load decomposition model by adopting a particle swarm algorithm by taking minimization of the total regulation quantity as a target so as to obtain a regulation and control target of each flexible load. A user flexible regulation and control pool is established, the adjustable load of each branch of a user is tracked in real time, and a regulation and control target value is generated according to a power grid regulation and control demand and a user load baseline.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and more specifically, to a non-invasive load decomposition method and system. Background Art

[0002] In the power system, with the large-scale access of renewable energy and the volatility of load demand, flexible regulation has become an important means to ensure the stable operation of the power system. Existing regulation methods usually rely on fixed regulation equipment, while ignoring the distribution and flexibility of each user's load. In addition, traditional load regulation algorithms are inefficient when dealing with large-scale complex systems and cannot respond quickly to grid demand. Therefore, there is an urgent need for a method that can combine the characteristics of distributed loads and achieve the optimal regulation strategy through optimization algorithms.

[0003] In view of the above problems, a non-invasive load decomposition method and system are urgently needed. Summary of the invention

[0004] In order to solve the deficiencies in the prior art, the present invention provides a non-invasive load decomposition method and system, which establishes a user flexible regulation pool, tracks the adjustable loads of each user branch in real time, and generates a regulation target value based on the grid regulation demand and the user load baseline.

[0005] The present invention adopts the following technical solution.

[0006] The first aspect of the present invention relates to a non-invasive load decomposition method, which includes the following steps: obtaining an expression for a flexible load reduction amount, and constructing a load decomposition model with the goal of minimizing the scheduling deviation and the constraint that the flexible load reduction amount falls within a comfort range; receiving control instructions issued by a power grid, collecting historical loads, and using a particle swarm algorithm to solve the load decomposition model with the goal of minimizing the total regulation amount, so as to obtain the control target of each flexible load.

[0007] Preferably, an expression for flexible load reduction is obtained, including: flexible loads include electric vehicle charging equipment, energy storage systems, lighting systems and air-conditioning equipment; the load reduction methods of electric vehicle charging equipment are charging power regulation, charging delay, and dynamic power allocation; the load reduction methods of energy storage equipment are charging and discharging regulation, frequency regulation, and demand response; the load reduction methods of air-conditioning equipment are changing the operating mode, adjusting the operating parameters, and changing the operating method.

[0008] Preferably, the load reduction method of the air conditioning equipment is to change the operation mode, adjust the operation parameters and change the operation method, including: changing the operation mode includes global temperature control and pre-cooling; adjusting the operation parameters includes increasing the chilled water temperature, increasing the supply air temperature, limiting the cooling valve, and fan frequency conversion control; changing the operation method includes limiting the cooling demand, shutting down the medium conveying equipment, and reducing the number of fans used. 6BD; 7;

[0009] Preferably, the load decomposition model is constructed with the minimum scheduling deviation as the goal and the flexible load reduction amount falling within the comfort range as the constraint, including: constructing a contribution formula of various control measures in the air-conditioning system to the total power regulation ΔP:

[0010] ΔP=ΔP t +ΔP a +ΔP w +ΔP v +ΔP f +ΔP c +ΔP d

[0011] Various control measures for global temperature control ΔP t , pre-cooling ΔP a , increase the chilled water temperature ΔP w , cold water valve limit ΔP v , fan frequency conversion control ΔP f , Cooling capacity demand limit ΔP c and reduce the number of fans ΔP d .

[0012] Preferably, the load decomposition model is constructed with the minimum dispatch deviation as the goal and the flexible load reduction falling within the comfort range as the constraint, including: using a polynomial regression analysis method to establish the relationship between the comfort level C and each flexible control strategy X1 to X8 as follows:

[0013]

[0014] In the formula, A 10 To A 1n , …, A 70 To A 7n is the polynomial regression coefficient; j is the regression order, n is the highest term, X1 to X8 are the flexible control strategies; i is the number of each flexible control strategy.

[0015] Preferably, the load decomposition model is constructed with the minimum dispatch deviation as the goal and the flexible load reduction falling within the comfort range as the constraint, including: solving the relationship between the comfort level C and each flexible control strategy X1 to X8 through linear regression analysis to obtain each regression coefficient A ij .

[0016] Preferably, the control instructions issued by the power grid are received, the historical loads are collected, and the load decomposition model is solved by using a particle swarm algorithm with the goal of minimizing the total control amount to obtain the control target of each flexible load, including: the objective function of minimizing the total control amount is:

[0017]

[0018] ΔP i is the adjustment value of the ith branch at time point t;

[0019] The constraint condition is that, according to whether the time point t belongs to the peak or valley period, the constraint P is obtained. i (t-1)≥P i (t).

[0020] A second aspect of the present invention relates to a non-intrusive load decomposition system, which includes a building module and a decomposition module; wherein the building module is used to obtain an expression for a flexible load reduction amount, and to construct a load decomposition model with the goal of minimizing the scheduling deviation and the constraint that the flexible load reduction amount falls within a comfort range; the decomposition module is used to receive control instructions issued by a power grid, collect historical loads, and with the goal of minimizing the total regulation amount, use a particle swarm algorithm to solve the load decomposition model to obtain the control target of each flexible load.

[0021] The third aspect of the present invention relates to a terminal, comprising a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the method described in the first aspect of the present invention.

[0022] A fourth aspect of the present invention relates to a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of the method in the first aspect of the present invention.

[0023] The beneficial effect of the present invention is that, compared with the prior art, the non-intrusive load decomposition method and system in the present invention, by establishing a user flexible control pool, tracks the adjustable loads of each branch of the user in real time, and generates a control target value according to the power grid control demand and the user load baseline. The control process constructs the target equation and the constraint equation, and optimizes and solves them in combination with the branch control priority, and finally uses the particle swarm algorithm to find the optimal control strategy. This method ensures the achievement of the control target, while minimizing the total control amount, and realizing accurate and flexible load control. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Figure 2 is a diagram of the non-intrusive load decomposition method. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the present invention clearer and more correct, the technical solution of the present invention is described in detail through multiple specific implementation methods below. It is worth noting that the embodiments adopted in the present invention are only used to explain the present invention and are not used to limit the content of the present invention.

[0026] Figure 1 The figure is a non-intrusive load decomposition method. The method of the present invention establishes a user flexible control pool, tracks the adjustable loads of each branch of the user in real time, and generates a control target value according to the power grid control demand and the user load baseline. The control process constructs the target equation and the constraint equation, and optimizes and solves them in combination with the branch control priority, and finally uses the particle swarm algorithm to find the optimal control strategy. This method ensures the achievement of the control target, while minimizing the total control amount, and realizing accurate and flexible load control.

[0027] The present invention discloses a non-intrusive load decomposition method, which includes the following steps: obtaining a flexible load reduction expression, and constructing a load decomposition model with the goal of minimizing the scheduling deviation and the flexible load reduction falling within the comfort range as a constraint; receiving a control instruction issued by the power grid, collecting historical loads, and solving the load decomposition model with the goal of minimizing the total adjustment amount by using a particle swarm algorithm to obtain the control target of each flexible load. Obtaining a flexible load reduction expression includes: the flexible load includes electric vehicle charging equipment, energy storage system, lighting system and air conditioning equipment; the load reduction method of the electric vehicle charging equipment is charging power regulation, charging delay, and dynamic power allocation; the load reduction method of the energy storage equipment is charging and discharging regulation, frequency regulation, and demand response; the load reduction method of the air conditioning equipment is changing the operation mode, adjusting the operation parameters, and changing the operation method. The load reduction method of the air conditioning equipment is changing the operation mode, adjusting the operation parameters, and changing the operation method, including: changing the operation mode includes global temperature control and pre-cooling; adjusting the operation parameters includes increasing the chilled water temperature, increasing the air supply temperature, limiting the cooling valve, and fan frequency conversion control; changing the operation method includes limiting the cooling demand, closing the medium conveying equipment, and reducing the number of fans used.

[0028] In practical applications, it is first necessary to establish a user flexible regulation pool, including tracking the branch load of each user and confirming the adjustable load. At the same time, the regulation priority and response time are preset for each user device to ensure rapid response and effectiveness during regulation. In practical applications, the user flexible regulation pool is the core of realizing flexible regulation of the power system. To this end, the specific conditions of various types of users and their branch loads must be tracked and managed in detail. Specifically, the regulation pool should include electric vehicle charging equipment, energy storage systems and lighting systems, as well as air conditioning equipment.

[0029] Electric vehicle charging piles can be adjusted by delaying charging or reducing charging power. They have large adjustable capabilities and flexible response duration, generally several hours. The response delay time depends on the charging strategy of the electric vehicle, and the reliability is relatively general, especially during the low load period at night. The regulation of electric vehicle charging equipment mainly includes charging power regulation, charging delay, and dynamic power allocation.

[0030] Charging power regulation adapts to the real-time needs of the power grid by adjusting the output power of the charging equipment. During peak hours of the power grid, you can choose to reduce the charging power to reduce the pressure on the power grid. Charging delay During periods of high electricity prices or heavy grid loads, you can choose to delay the charging process and resume charging after the electricity price drops or the grid load is reduced to achieve economic benefits and load balance. Dynamic power allocation For the situation of multiple charging piles, the intelligent control system dynamically allocates charging power, optimizes the overall charging efficiency, and avoids instantaneous impact on the power grid. In the control process of electric vehicle charging equipment, the following formula can be used to calculate the adjusted charging power and its impact on the load:

[0031] The charging power adjustment is ΔP e =P charging -P target , ΔP e is the adjusted charging power, in kW; P charging is the original charging power, in kW; P target is the target charging power, in kW.

[0032] Delayed charging affects ΔP delay = C·k·T delay , ΔP delay The load reduction caused by delayed charging, kW; C is the charging capacity of each electric vehicle, in kWh. k is the charging efficiency (no unit, usually between 0.8 and 0.9); T delay The delay time, in hours.

[0033] When regulating electric vehicle charging equipment, the following factors should be considered to determine their priority. During peak hours, the charging power should be controlled to reduce the load, while during off-peak hours, the charging power can be appropriately increased. If users have urgent travel needs, their charging needs should be met first. Charging operations should be prioritized during periods with lower electricity prices to make efficient use of electricity resources.

[0034] Energy storage devices such as batteries play a key role in balancing energy supply and demand. They can be regulated through charging and discharging behaviors, with high adjustability and flexible response time planning. Due to their high energy density, energy storage devices can provide stable and long-term power supply support. They have short response delays and high reliability, making them an important regulatory resource in emergencies or emergencies.

[0035] The regulation of energy storage systems mainly includes charge and discharge regulation, frequency regulation, and demand response. When the grid load is low or the electricity price is cheap, the energy storage system absorbs excess electrical energy and stores it as chemical energy or mechanical energy. When the grid load is high, the stored electrical energy is quickly released to meet the power demand and reduce the grid load. The energy storage system can quickly respond to fluctuations in grid frequency and charge and discharge as needed to maintain the stability of the grid. For example, when the frequency drops, it can be discharged to support the grid; when the frequency rises, it can be charged to absorb excess energy. According to the regulation signal of the grid, the energy storage system can flexibly adjust its charging and discharging strategy to meet the changes in the demand of the electricity market and participate in the competition in the electricity market.

[0036] In the regulation process of the energy storage system, the following formula can be used to calculate the charging and discharging power and its impact on the load. The energy storage charging and discharging power is ΔP s =P in -P out , ΔP s P is the power adjusted by the energy storage system, in kW; in is the energy storage charging power, in kW; P out is the energy storage discharge power, unit kW. The energy state is: E stored =E initial +ΔE in -ΔE out , E stored is the current stored energy, kWh; E initial Initial storage energy, kWh; ΔE in is the charging energy, i.e. the energy input during the charging process, in kWh; ΔE out It is the discharge energy, i.e. the energy output during the discharge process, in kWh.

[0037] When regulating the energy storage system, several factors, including the state of the grid, the integration of renewable energy, and economic efficiency, should be considered to determine its priority. When the grid experiences frequency fluctuations or peak loads, the energy storage system should give priority to releasing stored energy; charging should be given priority during periods of low electricity prices. When there is an oversupply of renewable energy, such as solar or wind power, the energy storage system should be given priority to charging; when renewable energy is insufficient, discharging should be given priority to ensure the balance of the system. According to the price dynamics of the electricity market, select the best charging and discharging time to maximize benefits. For example, discharging when electricity prices are high can obtain higher profits, while charging when electricity prices are low can reduce operating costs.

[0038] Although the lighting load is small, it is widely distributed and easy to integrate. When the regulation demand is not high, the load can be adjusted in time by adjusting the brightness or the switch state. Its adjustable capacity is medium, and the response duration is usually short, but the delay time is almost negligible, and the reliability rate is also very high.

[0039] In addition, air conditioning equipment is an important flexible load. Air conditioning systems usually have high adjustability and can respond to grid control needs by adjusting temperature settings. Its response duration is generally short, but it can be started and stopped frequently. The response delay time of air conditioning is usually small and has a high reliability rate, which is due to its relatively fast control feedback mechanism. The system architecture of air conditioning loads participating in grid control can be divided into equipment layer, terminal layer, communication layer, master station layer and dispatching control layer from bottom to top. The equipment layer includes air conditioning systems and monitoring devices, which are responsible for uploading information to the terminal and receiving control instructions and policy configuration. The functions of the terminal layer include capability evaluation, policy configuration and data management. The communication layer mainly realizes signal transmission through communication channels such as power optical fiber.

[0040] For central air-conditioning systems, according to the analysis of their operating characteristics and energy consumption characteristics, there are three main types of flexible control methods: changing the operating mode, adjusting operating parameters, and changing the operating method. Adjusting the operating mode includes global temperature control and pre-cooling. Global temperature control involves adjusting the overall temperature setting of the air-conditioning control area, or modifying the set value of the indoor air-conditioning return air temperature. Pre-cooling is to increase the system's cooling output before the strategy is executed to lower the room temperature and reserve cooling capacity in advance for the implementation of the strategy. Adjustment of operating parameters includes increasing the chilled water temperature, increasing the supply air temperature, limiting the cooling valve, and fan frequency conversion control. Increasing the chilled water temperature means raising the outlet water temperature of the air-conditioning host; increasing the supply air temperature is intended to increase the outlet air temperature of the indoor supply fan; limiting the cooling valve is to adjust the chilled water valve of the indoor supply air device to reduce the cooling entering the room; fan frequency conversion control is achieved by changing the motor output frequency of the supply air device. Changing the operating method includes limiting the cooling demand, shutting down some medium conveying equipment, and reducing the number of fans used. Limiting the cooling demand to reduce the cooling load, and reducing the number of fans is achieved by shutting down some indoor supply fans.

[0041] Although these measures are proposed independently, they are usually interrelated in actual operation. For example, global temperature control reduces cooling demand by increasing the indoor temperature setting, which may affect the implementation effect of the pre-cooling strategy because pre-cooling needs to be carried out at a lower base temperature. Similarly, fan variable frequency control and chilled water valve limitation need to work together when adjusting the wind speed and chilled water flow to ensure that the air conditioning system achieves energy saving while meeting comfort. Cooling demand limitation and reducing the number of fans directly affect the total load output of the system, and their regulation effect will also be affected by other control strategies. Therefore, when implementing these control measures, it is necessary to comprehensively consider the interactions between them to optimize the energy efficiency and performance of the entire system.

[0042] When the room temperature is set to 26°C as the benchmark, the energy consumption can be reduced by about 7.5% for every 1°C increase in the air conditioner set temperature. However, the implementation of this strategy requires a certain buffer time and has high requirements for the degree of building automation. The calculation of the load reduction of global temperature control can refer to the formula ΔP t =7.5%×P air ΔT, ΔP t The unit is kW. P air is the power of the air-conditioning system, which is the sum of the powers of all equipment, in kW. ΔT is the temperature adjustment value, in ℃. After the pre-cooling strategy is implemented, the indoor temperature will drop to a certain level. At this time, the air-conditioning system can be temporarily shut down until the room temperature reaches the set maximum limit. The sum of the powers of all equipment shut down after the pre-cooling strategy is completed is the theoretical load reduction value. When calculating the actual load reduction, it is necessary to consider the user behavior reflection caused by the rise in indoor temperature, and the reduction in cooling output caused by the self-regulation of the central air-conditioning system when all cold source equipment is not completely shut down. Therefore, it is necessary to make appropriate corrections based on the original equipment power, which is calculated as ΔP a =δ1·P m +P a , ΔP a The pre-cooling power reduction is in kW, 61 is the correction value, ranging from 60% to 100%, P m is the host operating power, in kW, P a = is the operating power of other auxiliary equipment, in kW. For constant power centrifugal chillers, every 1°C increase in chilled water outlet temperature can achieve 0.91% to 1.97% energy savings. For centrifugal units equipped with frequency conversion devices, when the load is less than 80%, every 1°C increase in chilled water outlet temperature can achieve an energy saving effect of 3.64% to 5.46%. The load reduction calculation is ΔP w =ΔT'·δ2·P m , ΔP wThe power cut to increase the chilled water temperature, in kW; ΔT' is the upward adjustment of the water temperature, in °C; δ2 is the correction value, the constant power centrifugal unit ranges from 0.91% to 1.97%, and the variable frequency centrifugal unit ranges from 3.64% to 5.46%. The restriction of the cold water valve will adjust the flow of cold water into the fan heat exchange coil. However, during the adjustment process, the rise in indoor temperature may cause discomfort to the user. In this case, the user is likely to increase the speed of the fan, which will result in a reduction in the load reduction effect. Calculation of load reduction ΔP v =(V a / V r )·δ3·P m , ΔP v The power reduction for cold water valve regulation, unit: kW, V a To adjust the total water flow, that is, the upper limit of the cold water flow rate limited by the cold water valve, V r is the total running water flow, that is, the water flow output value at the time of implementing the cold water valve restriction, δ3 is the correction value, and the value range is 20% to 80%. In the fan controlled by frequency conversion, there is a certain relationship between the frequency and power of the fan. At the same time, it is necessary to consider the impact of the reduction of fan capacity on the host output. Therefore, the load reduction ΔP f =∑(1-(f l / f) 3 )·P b +δ4·P m , ΔP f Power reduction for fan frequency conversion control, kW; f l is the limited frequency, i.e. the limited operating frequency of the fan after frequency conversion, Hz; f is the reference frequency, Hz; P b is the fan base load, kW; δ4 is the correction value of 5%, when f l =f, the value is 0, and the fan frequency conversion control is not performed. When the host load is reduced, the host power presents a nonlinear relationship. To simplify the calculation, it can be assumed that the host power changes linearly in certain operating areas and adjusted by introducing a correction factor. Load reduction ΔP c =δ5·∑(N1 / N r )·P m , ΔP c The power reduced to limit the cooling demand, kW; δ5 is the correction parameter; N1 is the host limit load value, which limits the cooling capacity output of the refrigeration unit, that is, the upper limit value of the cooling capacity output of the refrigeration unit; N rIt is the output value of the host operation, that is, the output value of the cooling capacity at the time when the cooling capacity demand is limited. When the number of fans is reduced, in addition to directly reducing the fan operation load, the terminal cooling output will also be reduced, and the host load will also be reduced accordingly. The temperature difference between the indoor area where the fan is turned off and the cooling area will cause gas convection, thereby increasing the cooling load of the cooling area. Therefore, it is necessary to make certain corrections to the host load reduction, load reduction ΔP d =n·P d +δ6·P m , ΔP d Power reduction to reduce the number of fans, kW; P d is the power of a single fan, kW; n is the number of reduced fans; δ6 is the correction value, ranging from 20% to 80%.

[0043] The demand response terminal screens and optimizes the combination of flexible control strategies according to the peak-shaving instructions and user comfort, and finally forms the best flexible control operation strategy that meets the comfort requirements and peak-shaving needs. How the demand response terminal achieves the optimal combination of flexible control operation strategies can be described as a mathematical programming problem of minimizing the peak-shaving deviation while meeting the comfort requirements. The specific model is described as: Min(|P-ΔP|), P is the grid regulation target value received by the user; AP is the total load reduced after flexible regulation.

[0044] ΔP=ΔP a +ΔP t1 +ΔP t2 +ΔP d +ΔP f +ΔP v +ΔP w +ΔP c

[0045] The relationship between comfort and each flexible control strategy is established by using polynomial regression analysis. Each term represents the contribution of various control measures in the air-conditioning system to the total power regulation, including global temperature control / increasing supply air temperature, pre-cooling, increasing chilled water temperature, chilled water valve restriction, fan variable frequency control, cooling demand restriction and reducing the number of fans.

[0046] Assuming that the maximum number of each of the eight independent variables in the multivariate nonlinear regression model is n, then:

[0047]

[0048] C is comfort; A is ij are the polynomial coefficients; Reduce the load variable value for each flexible control strategy (i ranges from 1 to 8, The variables are ΔP a , ΔPt1 , ΔP t2 , ΔP d , ΔP f , ΔP v , ΔP w , ΔP c ).

[0049] make but Solve the above equation by linear regression analysis to obtain the regression coefficients A ij . In order to ensure the reliability of the building central air-conditioning load participating in the power grid regulation, the optimization strategy decision is made from two time scales, day-ahead and real-time. In the day-ahead peak regulation, the user side makes the optimization decision of the flexible regulation strategy according to the peak regulation instruction, with the day-ahead dispatch interval of the power grid usually being 15 minutes, and then forms the peak regulation plan for the next day. Therefore, in the day-ahead power grid peak regulation, the optimization decision model of the regulation strategy is as follows:

[0050]

[0051] st C(t)≤C UP (T)

[0052] In the formula, P(t) is the real-time load predicted based on the collected historical load, which can be the sum of the real-time loads of all flexible loads in the whole network, and ΔP(t) is the load reduction mentioned above, which is the sum of the reduction of all flexible loads in the whole network. The objective function is constructed with the minimum adjustment amount as the goal, and the disassembled reduction amount, that is, the sum of the comfort level, is used as a constraint. Among them, C UP (T) is the upper limit of comfort, which is a curve with different sizes at different times. It can be set to a few kW at a time, as long as C(t) is below it. According to the actual constraint range, the comfort range can be known.

[0053] In the real-time operation of the next day, due to changes in external environment, climate, and weather factors, the output deviation may be large. Therefore, the demand response terminal collects real-time external environmental factors and air conditioning operation status to make real-time corrections to the control strategy.

[0054]

[0055] st C(t')≤C UP (T)1≤t≤T

[0056] Where ΔP′(t′) is the peak load regulation amount in the real-time correction period t′, kW; Tf′ is the number of real-time correction periods, which is the integer of the ratio of the day-ahead scheduling interval to the real-time correction interval.

[0057] Linear Programming (LP): When the problem can be transformed into a linear relationship, LP is a valid choice. Although minimizing the absolute value sum is not a linear problem, it can be transformed into a linear form by introducing additional variables and constraints. For example, an auxiliary variable can be introduced to replace the absolute value function. Integer Linear Programming (ILP): If the problem involves integer decision variables (such as on / off states or other discrete choices), integer linear programming can be used.

[0058] Quadratic Programming (QP): If the optimization objective or constraints contain quadratic terms, then quadratic programming can be applied. Dynamic Programming (DP): Dynamic programming is particularly effective when the problem has staged decisions and time series dependencies. Heuristic and meta-heuristic algorithms: including simulated annealing, genetic algorithms, and particle swarm optimization (PSO), etc., are particularly suitable for solving large-scale and complex optimization problems, or when traditional methods become infeasible due to problem size or nonlinear constraints. Particle swarm optimization can be combined with other algorithms (such as dynamic programming or linear programming) to improve the quality of the solution or the speed of solution. For example, a particle swarm algorithm can be used to obtain an approximate solution as a starting solution, and then a more accurate algorithm (such as linear programming) can be used for refinement and optimization. In addition, the particle swarm algorithm can continuously adjust parameters during the iteration process to adapt to the dynamic characteristics of the problem, which is particularly useful in dynamic systems such as power system regulation.

[0059] The above algorithms are all used to solve the load shedding distribution process. The objective function and constraints are calculated to obtain the control target.

[0060] When the power grid issues a control instruction, it should receive the instruction according to the load demand at different time points and distribute it to each user branch. To ensure that users can reasonably arrange their equipment loads to respond to the power grid control requirements, the superior agency generally sends a control instruction notification to the system one day in advance. This advance notification time allows users to adjust their operation plans, thereby effectively reducing the sudden impact caused by load adjustment. After receiving the control instruction, the system will calculate the specific control amount based on the load adjustment demand of the power grid. The setting of the control amount is based on real-time load data, historical load trend analysis, and target load adjustment strategy. For example, if the power grid instruction requires a load reduction of 5000kW, the system will allocate this control amount based on the adjustable capacity of various types of equipment to ensure that the load of different users can be adjusted evenly. The control instruction clearly specifies the time period for control, which is usually 1-2 hours. During this period, the system will continuously monitor the user's load situation and make real-time adjustments according to the instruction requirements. For example, when the power grid instruction requires a reduction of 5000kW from 16:00 to 17:00, the system will dynamically adjust the load situation of each user during this time period to ensure that the total load is within the expected range of the power grid. For each user, the system will reasonably allocate the control amount according to the adjustability and priority of their load. For equipment with fast response speed and strong adjustment ability, such as energy storage equipment and air conditioning systems, more control allocation will be given; while for equipment with slower response, such as some industrial equipment, the control amount will be appropriately reduced. In addition, the system will track the load changes in real time. If it is found that a certain device fails to adjust according to the predetermined steps, it will issue a secondary instruction in time to ensure the load adjustment of each device.

[0061] The system calculates the load average based on the user's load data for the past five days as the baseline load. This is an important reference for evaluating load changes in the subsequent regulation process. Combining the load baseline with the grid regulation requirements, the regulation target value at each time point is calculated.

[0062] The goal is to minimize the total adjustment amountMinimize: ΔP i is the regulation value of the ith branch at time point t. It satisfies the grid control target value and the control constraints of each device at the same time. In extreme cases, if the constraints conflict with the minimization objective, the grid control target value T(t) is preferentially satisfied to ensure grid stability.

[0063] The control amount must meet the control target value, that is Where T(t) is the target value of grid regulation. The load regulation of each branch must meet its own constraints, that is, except for energy storage and charging piles, the load at the time point before peak shaving should be greater than the load at the time point after peak shaving, and the opposite is true for valley filling. i (t-1)≥P i (t).

[0064] After the control amount of all branches meets the constraints, the particle swarm algorithm is used to optimize the control strategy. The individual particles represent the control amount of each user branch. The optimal strategy is found through iterative solution.

[0065] In the actual calculation process, the average load of the resource pool is calculated from all available devices. This step needs to consider the load characteristics of different types of equipment and accurately record the load data of each device in a specific time period. The calculation process can be expressed as: N is the total number of devices, P i is the load of the ith device. This includes but is not limited to air conditioners, charging piles, energy storage, lighting and other facilities and equipment. Based on the calculated average load, the next step is to select devices that are less than the maximum average load in the target load curve. This process will generate a pool of eligible devices to ensure that the selected devices do not exceed the preset load limit to maintain system stability. Any device selection needs to meet P i <The maximum average load of the target curve. Add the above selected equipment to the control equipment pool. The control equipment pool is the core part for subsequent load control and optimization. It contains a collection of load equipment that can be flexibly adjusted. Calculate the load of the selected equipment in the control equipment pool, find the total load, and compare it with the target curve to calculate the difference between the two AP = P 目标 -P 总负荷 , P 目标 is the target load, P 总负荷 is the total load of the control equipment pool. In each iteration, it is necessary to determine whether the average difference is less than 5% of the load. If it is less than 5%, it is considered that the control has achieved the expected goal and the cycle ends. The threshold |ΔP|<0.05×P can be set 目标 If the condition is met, the optimization stops; otherwise, proceed to the next step. If the average difference is greater than 5% of the load, the second step needs to be repeated, and the equipment that meets the requirements is reselected to update the control equipment pool. This process is iterative until the target load is finally met.

[0066] Initialize particle position and velocity P best (i) G best

[0067] Update particle positions and velocities:

[0068] v i (t+1)=w*v i (t)+C1*r1*(P best (i)-x i (t))+C2*r2*(G best (i)-x i (t))

[0069] X i(t+1)=x i (t)+V i (t+1)

[0070] Among them, v i (t) is the particle velocity, x i (t) is the particle position, P best (i) and G best (i) represents the individual and global optimal positions, w is the inertia weight, C1 and C2 are learning factors, and r1 and r2 are random numbers;

[0071] The control quantity of each branch device is represented by individual particles and solved by the optimization algorithm so that the control quantity satisfies the overall objective equation and constraints.

[0072] Finally, the optimized control strategy is output to guide the actual load regulation operation, so as to meet the target load requirements more efficiently and achieve the best resource allocation. The system tracks the load status of each branch in real time. If the load value increases, the priority of the branch is reduced to ensure the effectiveness of the control strategy. If the total regulation reaches the power grid control target, the system will no longer adjust; if not, the above steps are repeated until the control is completed at all time points.

[0073] The above load distribution process can be calculated using the same method such as particle swarm algorithm, or the first particle swarm can be used to distribute the load to the air-conditioning unit, and then the second particle swarm can be used to distribute the load reduction of the air-conditioning unit. The calculation method is various planning algorithms or heuristic algorithms. The calculation means is to solve the adjustment amount of the distributed flexible load with the minimum total adjustment amount as the goal and the comfort level falling within the constraint as the condition. The power of each flexible load substituted into the above formula can be a historical typical load or a power predicted based on the historical typical load. Through historical data training, the coefficients of the polynomial can also be obtained during model training as model parameters.

[0074] In practical applications, the load decomposition process is limited by the actual dispatching means. For example, it is impossible to adjust all loads at the same time. Therefore, the present invention also has the following improvements. When the power grid issues a control instruction, the instruction should be received according to the load demand at different time points and distributed to each user branch. In order to ensure that users can reasonably arrange their equipment loads to respond to the power grid control requirements, the superior agency generally sends a control instruction notification to the system one day in advance. The advance notification time enables users to adjust their operation plans, thereby effectively reducing the sudden impact caused by load adjustment. After receiving the control instruction, the system will calculate the specific control amount according to the load adjustment demand of the power grid. The setting of the control amount is based on real-time load data, historical load trend analysis and target load adjustment strategy. For example, if the power grid instruction requires a load reduction of 5000kW, the system will allocate this control amount according to the adjustable capacity of various types of equipment to ensure that the loads of different users can be adjusted evenly. The control instruction clearly stipulates the time period for control, which is usually 1-2 hours. During this period, the system will continue to monitor the user's load situation and make real-time adjustments according to the instruction requirements. For example, when the grid directive requires a 5000kW reduction from 16:00 to 17:00, the system will dynamically adjust the load of each user during this time period to ensure that the total load is within the expected range of the grid. For each user, the system reasonably allocates the control amount according to the adjustability and priority of its load (the specific allocation method will be mentioned below). For equipment with fast response speed and strong adjustment ability (such as energy storage equipment, air-conditioning system), a higher control amount will be allocated; for equipment with slower response (such as some industrial equipment), the control amount will be appropriately reduced. In addition, the system will track load changes in real time. If it is found that a certain device fails to adjust according to the predetermined steps, it will issue a secondary instruction in time to ensure that the load of each device is adjusted in place.

[0075] After the control amount of all branches meets the constraints, the particle swarm algorithm is used to optimize the control strategy. The individual particles represent the control amount of each user branch. The optimal strategy is found through iterative solution.

[0076] Compared with the prior art, the flexible regulation and real-time control method for power systems disclosed in the present invention provides an innovative solution, which significantly enhances the pertinence of power load response technology among various types of power users. This method is applicable to multiple fields including central air conditioning, electric vehicle charging piles, energy storage systems and lighting equipment. By establishing an economic benefit optimization objective function based on a particle swarm optimization algorithm, the load characteristics of different users are analyzed, and dynamic adjustments are implemented in actual operations to adapt to different demand responses. Its flexibility and adaptability enable it to cope with the changing power environment and maximize the optimization control effect of power load demand response.

[0077] This method not only achieves load optimization response, but also helps to save energy and reduce emissions, and maintain the stability of the power system. Specifically, by reasonably configuring the user's flexible control pool, real-time monitoring and adjustment of each user's load response, the system's ability to respond to emergencies and peak loads has been greatly improved.

[0078] Suppose that in order to improve the regulation capability of the power system during peak load periods, a city power company decides to introduce a flexible regulation strategy for power users, aiming to optimize the matching of power supply and demand and improve the overall efficiency of the system by dynamically adjusting the user's load response.

[0079] The control priority and response time are preset for each user device, which can be dynamically adjusted according to its load characteristics and the real-time needs of the power grid. This systematic load management not only enhances the ability to respond quickly, but also ensures the effectiveness and reliability of each device during the control process. This detailed control strategy can more accurately match the needs of the power grid and achieve higher control efficiency.

[0080] In one embodiment,

[0081] The user flexible control pool includes the following devices:

[0082] 1. Central air conditioning system (load adjustable range 1000-4000kW)

[0083] 2. Electric vehicle charging pile (load adjustable range 2000-4000kW)

[0084] 3. Energy storage system (charging and discharging power -10000-+10000kW)

[0085] 4. Lighting system (load adjustable range 0-1000kW)

[0086] Record the adjustable load data of various users for subsequent regulation. During the peak period of the power grid, the power supply company sends a regulation instruction to reduce the total system load by 5000kW from 16:00 to 17:00 on the same day, that is, to cut the peak by 5000kW for one hour. The given target value is 5000kW, and the focus is on regulation from the energy storage system. The specific allocation is as follows: Energy storage system: adjust 4000kW (discharge); central air-conditioning system: adjust 700kW; electric vehicle charging pile: adjust 300kW; lighting system: adjust 0kW (no adjustment). In the given regulation target value (5000kW), the adjustment amount allocated by the energy storage system (4000kW), the central air-conditioning system (700kW) and the electric vehicle charging pile (300kW) add up to just meet the target value. Specifically, according to the load characteristics of each device and its flexibility in responding to the power grid, the energy storage system is regarded as the main regulating force, with strong discharge capacity, and is preferentially adjusted to the maximum discharge level; at the same time, the load adjustment amount of the central air-conditioning system and the electric vehicle charging pile is based on its adjustable range and the actual user's demand response. The particle swarm algorithm will be used to optimize the control strategy to achieve effective load reduction. In the specific implementation process, the initial position and speed of the particles need to be reasonably set during initialization, and dynamically updated according to the load adjustment historical data. In order to ensure the convergence of the algorithm, it is necessary to set the maximum number of iterations and the convergence standard, such as stopping the iteration when the average difference is less than 5%. Assume that the initial loads of each device at a certain moment are: central air conditioning: 3800kW; electric vehicle charging pile: 3500kW; energy storage system: -2000kW (discharge state); lighting system: 600kW. According to the required adjustment target, the required adjustment amount is calculated. The energy storage system needs to be adjusted to -6000kW, with an adjustment amount of 4000kW (continue to discharge); the central air conditioning needs to be adjusted to 3100kW, with an adjustment amount of 700kW; the electric vehicle needs to be adjusted to 3200kW, with an adjustment amount of 300kW; the lighting system is maintained at 600kW, and no adjustment is required. If the load of a certain device does not change according to the predetermined steps, the system will monitor the status of the device in real time and make dynamic adjustments. The control strategy will combine the real-time load data with historical feedback, and adjust the priority and adjustment range of other devices by calculating the adjustable margin to achieve the overall load target. This process achieves rapid response with the help of feedback loops.

[0087] Assuming the response electricity price is 3 yuan / kWh, the control cost of the equipment is as follows:

[0088] Central air conditioning: 1 yuan / kWh

[0089] Electric vehicle charging station: 1 yuan / kWh

[0090] Energy storage system: 0.5 yuan / kWh

[0091] Lighting system: 0.2 yuan / kWh

[0092] The optimized load regulation is:

[0093] Energy storage system: Adjust 4000kW, cost 2000 yuan;

[0094] Central air conditioning: adjust 700kW, cost is 700 yuan;

[0095] Electric vehicle: Adjust to 300kW, cost is 300 yuan;

[0096] Lighting system: No adjustment, cost is 0 yuan;

[0097] The total demand response regulation is:

[0098] Total regulation capacity = 4000 + 700 + 300 + 0 = 5000kW

[0099] The overall control cost is:

[0100] Total cost = 2000 + 700 + 300 + 0 = 3000 yuan

[0101] Actual benefit calculation:

[0102] Actual income = total regulation amount × electricity price = 5000·3 = 15000 yuan

[0103] The final economic benefits are:

[0104] Economic benefit = actual income - total cost = 15,000-3,000 = 12,000 yuan

[0105] Through this embodiment, the flexible regulation and real-time control method of the present invention successfully reduced the load demand of the power grid during peak hours by 5000kW and lasted for one hour, significantly improving the response economic benefits. Using the energy storage system as the main means of regulation, supplemented by the flexible adjustment of other equipment, not only optimizes the load response, but also achieves good economic results. This scheme provides a feasible solution for the intelligent dispatching of power systems and has positive practical application prospects.

[0106] The flexible regulation and real-time control method for the power system disclosed in the present invention constructs an objective function for optimizing economic benefits, promotes the intelligence and efficiency of load demand response in the power industry, and has good market application prospects.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention is described in detail with reference to the above embodiments, it should be understood by those skilled in the art that the technical solutions of the present invention still include contents that can modify or replace the specific embodiments of the present invention. Any modification or replacement that does not depart from the spirit and scope of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A non-invasive load decomposition method, characterized in that: The method comprises the following steps: The expression of flexible load reduction is obtained, and the load decomposition model is constructed with the minimum scheduling deviation as the goal and the flexible load reduction falling within the comfort range as the constraint; Receive the control instructions issued by the power grid, collect historical loads, minimize the total regulation amount, and use the particle swarm algorithm to solve the load decomposition model to obtain the control target of each flexible load.

2. A non-invasive load decomposition method according to claim 1, characterized in that: The step of obtaining the flexible load reduction amount expression includes: The flexible loads include electric vehicle charging equipment, energy storage systems, lighting systems and air conditioning equipment; The load reduction mode of the electric vehicle charging equipment is charging power regulation, charging delay, and dynamic power allocation; The load reduction mode of the energy storage device is charge and discharge regulation, frequency regulation, and demand response; The load reduction method of the air conditioning equipment is to change the operation mode, adjust the operation parameters and change the operation method.

3. A non-invasive load decomposition method according to claim 1, characterized in that: The load reduction method of the air conditioning equipment is to change the operation mode, adjust the operation parameters and change the operation method, including: The changing operation mode includes global temperature control and pre-cooling; The adjustment of operating parameters includes increasing the chilled water temperature, increasing the supply air temperature, limiting the cooling valve, and fan frequency conversion control; The operation change method includes limiting the refrigeration demand, shutting down the medium conveying equipment, and reducing the number of fans used.

4. A non-invasive load decomposition method according to claim 1, characterized in that: The load decomposition model is constructed with the minimum dispatch deviation as the goal and the flexible load reduction amount falling within the comfort range as the constraint, including: Construct the contribution formula of various control measures in the air-conditioning system to the total power regulation ΔP: ΔP=ΔP t +ΔPa+ΔP w +ΔP v +ΔP f +ΔP c +ΔP d The various control measures are global temperature control ΔP t , pre-cooling ΔP a , increase the chilled water temperature ΔP w , cold water valve limit ΔP v , fan frequency conversion control ΔP f , Cooling capacity demand limit ΔP c and reduce the number of fans ΔP d .

5. A non-invasive load decomposition method according to claim 1, characterized in that: The load decomposition model is constructed with the minimum dispatch deviation as the goal and the flexible load reduction amount falling within the comfort range as the constraint, including: The polynomial regression analysis method is used to establish the relationship between the comfort level C and each flexible control strategy X1 to X8: In the formula, A 10 To A 1n , …, A 70 To A 7n is the polynomial regression coefficient; j is the regression order, n is the highest term, X1 to X8 are flexible control strategies; i is the number of each flexible control strategy.

6. A non-invasive load decomposition method according to claim 1, characterized in that: The load decomposition model is constructed with the minimum dispatch deviation as the goal and the flexible load reduction amount falling within the comfort range as the constraint, including: The relationship between comfort level C and each flexible control strategy X1 to X8 is solved by linear regression analysis to obtain the regression coefficient A ii .

7. A non-invasive load decomposition method according to claim 1, characterized in that: The receiving control instructions issued by the power grid, collecting historical loads, minimizing the total adjustment amount, and using the particle swarm algorithm to solve the load decomposition model to obtain the control target of each flexible load include: The objective function of minimizing the total adjustment amount is: ΔP i is the adjustment value of the ith branch at time point t; The constraint condition is that, according to whether the time point t belongs to the peak or valley period, the constraint P is obtained. i (t-1)≥P i (t).

8. A non-intrusive load decomposition system, characterized in that: The system includes a building module and a decomposition module; wherein, The construction module is used to obtain an expression for the flexible load reduction amount, and to construct a load decomposition model with the minimum scheduling deviation as the goal and the flexible load reduction amount falling within the comfort range as the constraint; The decomposition module is used to receive the control instructions issued by the power grid, collect historical loads, and use the particle swarm algorithm to solve the load decomposition model with the goal of minimizing the total adjustment amount to obtain the control target of each flexible load.

9. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images