An intelligent building power demand side optimization scheduling strategy based on a particle swarm algorithm

Abstract

The application discloses a kind of intelligent building power demand side optimization scheduling strategies based on particle swarm algorithm, obtains the building controllable load power state by monitoring system, according to building load characteristics, it is classified into basic load, can be translated load, can be cut load, both can be translated and can be adjusted load;Based on demand side response price incentive and contract incentive two incentive mechanisms, on this basis, with the minimum economic index power cost, user power satisfaction and power grid stability as objective function to establish optimization scheduling model;Finally, the improved particle swarm algorithm is applied to solve the optimization scheduling model, and simulation analysis is carried out, and the expected simulation result;It can effectively improve the economy and comfort of demand side, to achieve the effect of power grid peak clipping and valley filling.The application considers the dispatching mode of electric vehicle and photovoltaic power generation in peak period, to further reduce power cost.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grid electric energy dispatching, and in particular to a smart building electricity demand-side optimization dispatching strategy based on a particle swarm algorithm. Background Art

[0002] With the rapid development of smart grid technology, and to improve the absorption of renewable energy and achieve peak load shifting on the power grid, a growing number of researchers are beginning to leverage wind, photovoltaic, and energy storage to dispatch unit loads, proposing dispatch strategies that are more economical, reliable, environmentally friendly, and user-satisfactory. This approach also explores the enormous dispatch potential on the demand side, optimizing power usage strategies from the user's perspective to achieve a win-win situation for both users and the grid.

[0003] Regarding demand-side response, my country has currently implemented measures such as "time-of-use electricity pricing," "direct load control," and "interruptible load" to reduce maximum load and minimize peak-to-valley fluctuations. Due to the randomness, volatility, and uncontrollability of distributed renewable energy (RDG) output, these current demand-side response measures are unable to effectively absorb the fluctuations in renewable energy, significantly impacting user satisfaction and resulting in very low user satisfaction. Summary of the Invention

[0004] In response to the above technical problems, this technical solution provides an intelligent building electricity demand-side optimization scheduling strategy based on particle swarm algorithm, which adds contract incentives for the load that can be reduced on the demand side, and adds a contribution calculation formula to the incentive fee, which can effectively reflect the peak-shaving index and can further encourage users to participate in the response; it can reduce the load implementation, and the user's active participation in the response will not affect the user's electricity experience, which can further improve the stability and economy of the power grid; it can effectively solve the above problems. The present invention is achieved through the following technical solutions:

[0005] A particle swarm algorithm-based strategy for optimizing the electricity demand side of smart buildings classifies the loads in the building and establishes mathematical models for each type of load. The strategy uses the economic indicator electricity cost, user satisfaction, and reliability indicator grid stability as objective functions, and then uses the particle swarm algorithm to iteratively optimize. The specific steps include:

[0006] Step 1: Obtain the power consumption status of the building's controllable loads through the monitoring system. Based on the building's load characteristics, classify them into base loads, shiftable loads, curtailable loads, and both shiftable and adjustable loads.

[0007] Step 2: Comprehensively consider the charging and discharging characteristics of electric vehicles and the output of the photovoltaic system, analyze the user's electric vehicle charging optimization strategy, and establish a scheduling optimization model. The scheduling optimization model includes: a shiftable load scheduling model, a curtailable load scheduling model, a shiftable and adjustable load scheduling model, a HVAC model, and an EV and PV model.

[0008] Step 3: Propose a DR incentive mechanism based on demand-side response, which is divided into electricity price incentive and contract incentive;

[0009] Step 4: Establish the objective function and its constraints, taking the minimization of electricity cost, user satisfaction and grid stability as the objective function, and use the linear weighted combination method to transform the multiple objectives into a single objective function;

[0010] The objective function is:

[0011]

[0012] The optimization adjustment is carried out with the goal of minimizing electricity costs. Considering that in the case of excessive solar energy, users will sell the PV system output that cannot be absorbed to the power company for profit. λ is the electricity price at time t.

[0013]

[0014] With the goal of user satisfaction and grid stability, C fort Indicates user satisfaction. Compare before and after optimization. If the load usage time changes less, C fort The smaller the value, the higher the user satisfaction; the grid stability is expressed by the difference between the daily peak and valley power consumption;

[0015] Step 5: The established scheduling optimization model is solved using the improved particle swarm algorithm.

[0016] Furthermore, the basic loads described in step 1 include lighting systems and televisions, which have a large user demand and do not participate in system scheduling;

[0017] The shiftable loads include printers and projectors. Users adjust the load power usage interval according to their actual situation. However, due to the load characteristics, the load power and power usage time cannot be adjusted. The overall load curve will be shifted.

[0018] The load reduction mentioned above is that under the conditions of the contract signed between the power supply company and the user, the power supply company will issue a load reduction instruction to the user in advance based on the actual load operation situation, and the user can also receive financial compensation from it;

[0019] The loads described as both movable and adjustable include electric vehicles and HVAC constant temperature systems. These loads offer significant scheduling flexibility for users, consume a significant portion of electricity, and are a key factor in user comfort. The characteristic of these loads is that their power consumption and duration can be flexibly adjusted to a certain extent. The grid provides electricity price information one day in advance, adopting a time-of-use (TOU) pricing incentive mechanism to encourage users to set electricity usage strategies.

[0020] Furthermore, the load scheduling model described in step 2 is:

[0021] L TL，i =P TL，i Δt (1);

[0022]

[0023] Formula 1 is the translation load i at t i The electric energy consumed in the period; where i is the set of load types that can be translated i∈{l w ,l wh ,...l d};P TL,i is the translatable load i at t i Power of the time period; S TL,i Indicates the power consumption status of load i; is the actual working range of load i; To envision the best scheduling interval;

[0024] The curtailable load dispatch model is:

[0025]

[0026] L in Formula 4 XJ.k Indicates the actual load reduction capacity, It represents the ideal load reduction capacity, r ki Indicates the fluctuation ratio coefficient of the reduction amount in different gears allowed by the contract incentive mechanism;

[0027] The load scheduling model that can be translated and adjusted is:

[0028] L FL，j =P FL，j S FL,j Δt (5);

[0029]

[0030] Formula 5 is the actual response of the dispatch of the shiftable and adjustable load j; where j is the set of shiftable and adjustable load types j∈{l w ,l wh ,...l d}number; P FL，j For a load that can be translated and adjusted, j is at t j Power of the time period; S FL,j Indicates the start and stop status of load j; is the actual working range of load j; is the optimal scheduling interval; σ in formula 8 is the proportional coefficient of the adjustable power of this type of load.

[0031] The HAVC model is:

[0032] L HAVC =|P t |Δt (9);

[0033] Where: |P t | is the HVAC power, the absolute value here is to distinguish between cooling and heating; Δt is the recording interval;

[0034] K t =a H K t-1 +(1-a H )E t +a p P t (10);

[0035]

[0036] It is necessary to consider the indoor temperature change, which is related to the frequency of the variable temperature control load; comfortable room temperature range: the comfortable room temperature for the general human body is between 18 degrees Celsius and 25 degrees Celsius; HAVC power limit conditions; where P T is the temperature control load power in the T period, which is the optimization variable; a H is the building insulation coefficient, which is generally less than 1; a P K is the HVAC heating / cooling capacity coefficient; t is the indoor temperature variation law; K t-1 is the room temperature at the last moment; E t The outdoor temperature.

[0037] The EV and PV models are:

[0038]

[0039] Where SOC(t) is the target charging capacity of the electric vehicle; SOC(t0) is the remaining capacity of the electric vehicle before it is connected to the charging pile; P EV is the charging and discharging power of the electric vehicle; η EV is the electric vehicle power conversion coefficient; t-t0 is the electric vehicle charging time constant; E is the total amount of electricity that can be stored in the electric vehicle;

[0040] P EV ≤P max (14);

[0041] SOC min ≤SOC(t)≤SOC max(15);

[0042] The above is to ensure the life of electric vehicle batteries and constrain their charging and discharging power and electric vehicle charge state;

[0043] The output power of the PV power generation system is:

[0044]

[0045] P in Equation 16 PV Indicates the output of the photovoltaic system, P ref The system standard output calculated based on the photovoltaic solar panel parameters; S t represents the solar radiation intensity at time t; α θ is the temperature coefficient, K so is the actual temperature of the solar panel at time t; S ref , K ref They are the standard solar radiation intensity and temperature of the photovoltaic system respectively; the standard output of the photovoltaic system is related to the voltage and current characteristics of the solar panel.

[0046] Photovoltaic power generation system, in order to minimize electricity costs to the greatest extent, photovoltaic power generation devices can be installed on buildings. Although photovoltaic power generation is unstable, it can still play the role of peak shaving and valley filling in most periods of time.

[0047] Furthermore, the electricity price incentives described in step 3 can be divided into fixed electricity prices, time-of-use electricity prices, and real-time electricity prices. The three electricity price mechanisms are in a progressive relationship. The simulation example mainly focuses on the time-of-use electricity price mechanism. The power grid company conducts load forecasts on the day before and proposes electricity price information for different time periods based on the forecast data. It then transmits electricity price information for residential users participating in the incentive mechanism to encourage users to make reasonable electricity usage strategies based on the time-of-use electricity price.

[0048] The contract incentive mentioned above refers to a contract signed between the power grid and the user. Based on the grid's transient load conditions, the dispatch center will issue load reduction instructions to the user 30-120 minutes in advance to reduce the peak load. The user can then benefit from this. The incentive benefits are calculated based on the peak-shaving effect index and the different incentive fee coefficients on the tiered index to determine the specific benefits.

[0049]

[0050] C mot =H mot C dr ν mot H gxd (19);

[0051] Formula 17 is the calculation formula for the peak regulation effect index, where P total (t), They represent the total power consumption of the user at time t and the average power consumption of the day, P sys (T) They represent the total power of the power system in time period T and the average power consumption of the day respectively; Formula 18 is the actual contribution of the user to the power grid, which is also an important indicator of incentive costs; Formula 19 is the incentive cost formula of the power grid, which is based on the DR electricity price standard set by the power grid.

[0052] Furthermore, the linear weighted combination method described in step 4 is used to transform multiple objectives into a single objective function. The importance of each objective is multiplied by the corresponding weight coefficient, and finally the sum is added to form an objective function for solution. The single objective function after unified dimension is:

[0053] F = β1F1 + β2F2 + β3F3 (20);

[0054] For buildings such as office buildings, the economic index F1 is taken as the most important indicator, and then the optimization strategy is used to achieve the effect of peak shaving and valley filling for the power grid. The importance level of power grid stability F3 is higher than that of user satisfaction F2. The operator relationship between F1, F2, and F3 is judged one by one and normalized.

[0055] Furthermore, the improved particle swarm algorithm described in step 5 is to add a link to update the particle swarm parameters at the PSO algorithm iteration to obtain nonlinearly decreasing particle swarm parameters;

[0056]

[0057] For conventional particle swarm optimization, the parameters are fixed as constants, and the inertia weight and learning factor are set as one-dimensional vectors. Then, Equations 21 and 22 are the update iterative formulas for the inertia weight and learning factor, respectively; where w max 、w min 、c max 、c min is the maximum and minimum value of the inertia weight and learning factor; d1 and d2 are dynamic error coefficients; N it 、 is the current number of iterations and the maximum number of iterations.

[0058] Furthermore, the improved particle swarm algorithm is used for iterative optimization in step 5, and the specific steps are as follows:

[0059] S51: Obtain building load data and outdoor temperature information, initialize the electric vehicle charge state, predict photovoltaic output, and perform normalization processing;

[0060] S52: Set the particle swarm parameters, input the learning factor, inertia weight and number of iterations;

[0061] S53: Initialize the population and randomly generate a population for the movable and adjustable loads and energy storage systems;

[0062] S54: Calculate the fitness function and input a mathematical model with minimum electricity cost, user satisfaction and grid stability index as the objective function;

[0063] S55: Determine the global optimal fitness value, first take out the individual extreme value and the group extreme value, obtain the global and individual optimal, and determine the individual optimal fitness value and the global optimal fitness value;

[0064] S56: Start iterating to find the optimal solution;

[0065] S57: judging whether the power grid needs peak load regulation based on the real-time power load curve;

[0066] S58: If not, dispatch the curtailable load based on the contract incentives and propose a final electricity utilization strategy;

[0067] S59: Result analysis.

[0068] Furthermore, the steps for starting the iterative search for the optimum in step S56 are as follows:

[0069] S561: Update rate and population;

[0070] S562: Calculate and update the particle target fitness value;

[0071] S563: Determine whether the particles have converged. If not, continue iteration; otherwise, output the optimal solution and provide a user-side scheduling strategy.

[0072] Beneficial effects

[0073] The particle swarm algorithm-based optimization scheduling strategy for smart building electricity demand side proposed in this paper has the following advantages compared with the existing technology:

[0074] (1) In order to enhance the security, economy and stability of the power grid, the present invention optimizes the demand side scheduling, classifies the loads in the building, and establishes mathematical models for each type of load; based on the two incentive mechanisms of demand-side response electricity price incentive and contract incentive, on this basis, an optimization scheduling model is established with the economic indicator of minimum electricity cost, user electricity satisfaction and grid stability as the objective function, and then the particle swarm algorithm is used to iteratively find the optimal solution. Based on the time-of-use electricity price environment, the influence of photovoltaic energy storage and electric vehicle charging load and demand-side response incentive is considered, and the contribution index is added to establish the power equipment scheduling model for office buildings. Comprehensively considering the electricity cost, user satisfaction and voltage stability, the simulation example shows that the optimization model provides users with the best electricity plan by changing the electricity consumption strategy of different electricity loads. The results show that the intelligent building optimization scheduling strategy can reduce electricity costs, improve user satisfaction, and achieve the effect of peak shaving and valley filling for the power grid.

[0075] (2) The present invention utilizes an improved particle swarm algorithm, which introduces an iterative formula of learning factors and inertia weights and realizes nonlinear parameter reduction through an exponential function, which is beneficial for the algorithm to quickly jump out of the local optimum. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 Schematic diagram of the operation process of the present invention.

[0077] Figure 2 Schematic diagram of the particle swarm algorithm steps in the present invention.

[0078] Figure 3 This is the photovoltaic load prediction output curve for Case 1.

[0079] Figure 4 This is the scheduling result diagram of each building's power load before and after optimization in Case 1.

[0080] Figure 5 This is a comparison chart of the total load curve before and after optimization in Case 1.

[0081] Figure 6 This is a comparison chart of the fitness curves before and after the improvement of the PSO algorithm in Case 1. DETAILED DESCRIPTION

[0082] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Under the premise of not departing from the design concept of the present invention, various modifications and improvements made by ordinary persons in this field to the technical solutions of the present invention should fall within the scope of protection of the present invention.

[0083] Example 1:

[0084] A particle swarm algorithm-based strategy for optimizing the demand-side dispatch of electricity in smart buildings. Typically, load scheduling involves shifting and avoiding peak loads. Based on load response characteristics, interruptible and shiftable loads can achieve peak shifting effects. Users can shift their peak electricity consumption from peak loads on the grid, thus shifting peak loads and filling valleys. Users can also reduce their electricity consumption during peak load periods by reducing their loads.

[0085] Classify the loads in the building and establish mathematical models for each type of load. Use the economic indicator electricity cost, user satisfaction, and reliability indicator grid stability as objective functions, and then use the particle swarm algorithm to iteratively find the optimal solution. The specific steps include:

[0086] Step 1: Obtain the power consumption status of the building's controllable loads through the monitoring system, and classify the loads in the building according to the building load characteristics; classify them into basic loads, shiftable loads, reducible loads, and both shiftable and adjustable loads.

[0087] The basic loads include lighting systems and televisions, which have a large user demand and do not participate in system scheduling.

[0088] The shiftable loads include printers and projectors. Users adjust the load power usage range according to their actual conditions, but due to their load characteristics, they cannot be interrupted, and the power and power usage time of the load cannot be adjusted, resulting in the overall load curve being shifted.

[0089] The load that can be reduced is that under the conditions of the contract signed between the power supply company and the user, the power supply company will issue a load reduction instruction to the user in advance based on the actual operation of the load, and the user can also receive financial compensation.

[0090] The loads described as both movable and adjustable include electric vehicles and HVAC constant temperature systems. These loads offer significant scheduling flexibility for users, consume a significant portion of electricity, and are a key factor in user comfort. The characteristic of these loads is that their power consumption and duration can be flexibly adjusted to a certain extent. The grid provides electricity price information one day in advance, adopting a time-of-use (TOU) pricing incentive mechanism to encourage users to set electricity usage strategies.

[0091] Step 2: Taking into account the charging and discharging characteristics of electric vehicles and the output of the photovoltaic system, analyze the user's electric vehicle charging optimization strategy and establish a scheduling optimization model; the scheduling optimization model includes: a shiftable load scheduling model, a curtailable load scheduling model, a shiftable and adjustable load scheduling model, a HVAC model, and an EV and PV model.

[0092] The shiftable load dispatching model is:

[0093] L TL，i =P TL，i Δt (1);

[0094]

[0095] Formula 1 is the translation load i at t i The electric energy consumed in the period; where i is the set of load types that can be translated i∈{l w ,l wh ,...l d};P TL,i is the translatable load i at t i Power of the time period; S TL,i Indicates the power consumption status of load i; is the actual working range of load i; To imagine the optimal scheduling interval; the parameter data are shown in Table 2.

[0096] The curtailable load dispatch model is:

[0097]

[0098] L in Formula 4 XJ.k Indicates the actual load reduction capacity, It represents the ideal load reduction capacity, r ki It represents the fluctuation ratio coefficient of the reduction amount in different gears allowed by the contract incentive mechanism.

[0099] The load scheduling model that can be translated and adjusted is:

[0100] L FL，j =P FL，j S FL,j Δt (5);

[0101]

[0102] σ1P FL，j <P FL，j <σ2P FL，j (8);

[0103] Formula 5 is the actual response of the dispatch of the shiftable and adjustable load j; where j is the set of shiftable and adjustable load types j∈{l w ,l wh ,...l d}number; P FL，j For a load that can be translated and adjusted, j is at t j Power of the time period; S FL,j Indicates the start and stop status of load j; is the actual working range of load j; is the optimal dispatch interval; σ in Equation 8 is the proportional coefficient for the adjustable power of this type of load; σ1 and σ2 are set to 0.5 and 0.8. For other load data, refer to Table 2.

[0104] The HAVC model is:

[0105] L HAVC =|P t |Δt (9);

[0106] Where: |P t | is the HVAC power, the absolute value here is to distinguish between cooling and heating; Δt is the recording interval;

[0107] K t =a H K t-1 +(1-a H )E t +a p P t (10);

[0108]

[0109] It is necessary to consider the indoor temperature change, which is related to the frequency of the variable temperature control load; comfortable room temperature range: the comfortable room temperature for the general human body is between 18 degrees Celsius and 25 degrees Celsius; HAVC power limit conditions; where P T is the temperature control load power in the T period, which is the optimization variable; a H is the building insulation coefficient, which is generally less than 1; a P K is the HVAC heating / cooling capacity coefficient; t is the indoor temperature variation law; K t-1 is the room temperature at the last moment; E t is the outdoor temperature;

[0110] The EV and PV models are:

[0111]

[0112] Where SOC(t) is the target charging capacity of the electric vehicle; SOC(t0) is the remaining capacity of the electric vehicle before it is connected to the charging pile; P EV is the charging and discharging power of the electric vehicle; η EV is the electric vehicle power conversion coefficient; t-t0 is the electric vehicle charging time constant; E is the total amount of electricity that can be stored in the electric vehicle;

[0113] The output power of the PV power generation system is:

[0114]

[0115] P in Equation 16 PV Indicates the output of the photovoltaic system, P ref The system standard output calculated based on the photovoltaic solar panel parameters; S t represents the solar radiation intensity at time t; α θ is the temperature coefficient, K so is the actual temperature of the solar panel at time t; S ref , K ref They are the standard solar radiation intensity and temperature of the photovoltaic system respectively; the standard output of the photovoltaic system is related to the voltage and current characteristics of the solar panel.

[0116] Photovoltaic power generation system, in order to minimize electricity costs to the greatest extent, photovoltaic power generation devices can be installed on buildings. Although photovoltaic power generation is unstable, it can still play the role of peak shaving and valley filling in most periods of time.

[0117] Step 3: Propose a DR incentive mechanism based on demand-side response, which is divided into electricity price incentive and contract incentive.

[0118] The electricity price incentives described can be divided into fixed electricity prices, time-of-use electricity prices, and real-time electricity prices. The three electricity price mechanisms are in a progressive relationship. The simulation example mainly focuses on the time-of-use electricity price mechanism. The power grid company conducts load forecasts on the day before and proposes electricity price information for different time periods based on the forecast data. It then transmits electricity price information for residential users participating in the incentive mechanism to encourage users to make reasonable electricity usage strategies based on the time-of-use electricity price.

[0119] The contract incentive mentioned above refers to a contract signed between the power grid and the user. Based on the grid's transient load conditions, the dispatch center will issue load reduction instructions to the user 30-120 minutes in advance to reduce the peak load. The user can then benefit from this. The incentive benefits are calculated based on the peak-shaving effect index and the different incentive fee coefficients on the tiered index to determine the specific benefits.

[0120]

[0121] C mot =H mot C dr ν mot H gxd (17);

[0122] Formula 15 is the calculation formula for the peak regulation effect index, where: They represent the total power consumption of the user at time t and the average power consumption of the day, P sys (T) They represent the total power of the power system in time period T and the average power consumption of the day respectively; Equation 16 is the actual contribution of the user to the power grid, which is also an important indicator of incentive costs; Equation 17 is the incentive cost formula of the power grid, which is based on the DR electricity price standard set by the power grid.

[0123] Step 4: Establish the objective function and its constraints, taking the minimum electricity cost, user satisfaction and grid stability as the objective function, and use the linear weighted combination method to transform the multiple objectives into a single objective function. The objective function is:

[0124]

[0125] The optimization adjustment is carried out with the goal of minimizing electricity costs. Considering that in the case of excessive solar energy, users will sell the PV system output that cannot be absorbed to the power company for profit. λ is the electricity price at time t.

[0126]

[0127] With the goal of user satisfaction and grid stability, C fort Indicates user satisfaction. Compare before and after optimization. If the load usage time changes less, C fort The smaller the value, the higher the user satisfaction; the grid stability is expressed by the difference between daily peak and valley power consumption.

[0128] The linear weighted combination method is used to transform multiple objectives into a single objective function. The importance of each objective is multiplied by the corresponding weight coefficient, and finally the sum is added to form an objective function for solution. The single objective function after unified dimension is:

[0129] F = β1F1 + β2F2 + β3F3 (21);

[0130] For buildings like office buildings, the economic index F1 is the most important indicator. Optimization strategies are then used to achieve peak load shifting and valley filling for the power grid. The importance of power grid stability F3 is higher than that of user satisfaction F2. The operator relationships between F1, F2, and F3 are determined one by one and normalized. Thus, β1 = 0.5, β2 = 0.2, and β3 = 0.3.

[0131] Step 5: The established scheduling optimization model is solved using the improved particle swarm algorithm.

[0132] The improved particle swarm algorithm is to add a link to update the particle swarm parameters at the iteration of the PSO algorithm to obtain nonlinear decreasing particle swarm parameters;

[0133]

[0134] For conventional particle swarm optimization, the parameters are fixed as constants, and the inertia weight and learning factor are set as one-dimensional vectors. Then, Equations 22 and 23 are the update iterative formulas for the inertia weight and learning factor, respectively. max =0.9,w min =0.4,c max =1.5,c min =0.5;d1=0.025,d2=0.75;

[0135] The specific steps of iterative optimization using the improved particle swarm algorithm are as follows:

[0136] S51: Obtain building load data and outdoor temperature information, initialize the electric vehicle charge state, predict photovoltaic output, and perform normalization processing;

[0137] S52: Set the particle swarm parameters, input the learning factor, inertia weight and number of iterations;

[0138] S53: Initialize the population and randomly generate a population for the movable and adjustable loads and energy storage system;

[0139] S54: Calculate the fitness function and input a mathematical model with minimum electricity cost, user satisfaction and grid stability index as the objective function;

[0140] S55: Determine the global optimal fitness value, first take out the individual extreme value and the group extreme value, obtain the global and individual optimal, and determine the individual optimal fitness value and the global optimal fitness value;

[0141] S56: Start iterating to find the optimal solution;

[0142] S561: Update rate and population;

[0143] S562: Calculate and update the particle target fitness value;

[0144] S563: Determine whether the particles have converged. If not, continue iteration; otherwise, output the optimal solution and provide a user-side scheduling strategy.

[0145] S57: judging whether the power grid needs peak load regulation based on the real-time power load curve;

[0146] S58: If not, dispatch the curtailable load based on the contract incentives and propose a final electricity utilization strategy;

[0147] S59: Result analysis.

[0148] The specific results analysis is as follows:

[0149] Case 1:

[0150] This example divides a 24-hour day into 48 time periods. The Jiangsu Province industrial electricity price policy, shown in Table 1, is used under the time-of-use electricity pricing policy. When prices are low, more electricity is consumed; when prices are high, less electricity is consumed.

[0151] Table 1 shows the time-of-use electricity price information, which is divided into three periods: peak period, off-peak period and valley period.

[0152] The power grid company will issue time-of-use electricity price information to the HEMS system in the near future. The objects involved in the dispatch are shiftable loads, shiftable and dispatchable loads, photovoltaic energy storage and electric vehicle SOC.

[0153] Table 1 Time-of-use electricity price information

[0154]

[0155] The peak electricity consumption period is generally in summer, so the data of one day in summer in a certain area of ​​Jiangsu Province was selected as the standard.

[0156] Table 2 shows the parameters of various power loads. These loads are common in office building scenarios. Based on human thinking habits and reference to multiple literature surveys, parameters such as the optimal scheduling range and operating time are proposed.

[0157] Table 2 Various power load parameters

[0158]

[0159] Figure 3 The invention collects photovoltaic power data absorbed by solar panels in a certain area in summer. The invention allows users to participate in demand-side response and sell the surplus electricity stored in the photovoltaic system to power companies in exchange for income.

[0160] As shown in Table 2, the building's electricity load is divided into basic load, curtailable load, shiftable load and both interruptible and shiftable load.

[0161] The base load is not subject to scheduling adjustment due to its rigid characteristics, so it is not involved in scheduling in this example.

[0162] The loads that can be reduced include computers. There are a number of computers in the building that consume electricity. The number of loads can be controlled to reduce power to a certain extent.

[0163] The same characteristics of shiftable and interruptible loads are that they can be used during low-price periods within the user-defined time range, but this may reduce user satisfaction to a certain extent. With the goal of minimizing electricity costs, user satisfaction, and grid stability, the particle swarm algorithm is used to solve the model, with the operating status of the electricity load as the decision variable and the optimal scheduling interval as the constraint condition. The results are as follows: Figure 4 shown.

[0164] observe Figure 4 First, after optimization, electric vehicles were considered and scheduled in the off-peak period from 04:00 to 08:00. Other loads were basically also scheduled to the range with the lowest electricity price within the user's allowed usage range.

[0165] Since the water consumption in the morning is relatively high, the water dispenser is scheduled to supply and boil water before 08:00.

[0166] After the implementation of the contract incentive, the load can be reduced. Compared with before optimization, there is a certain degree of reduction trend at 14:00, because some people will take a lunch break at this time and will not use the computer, and will finish get off work early at 17:00.

[0167] Table 3 Optimization results

[0168] stage <![CDATA[Electricity cost / (yuan·d -1 )]]> User satisfaction Peak-to-valley difference / Kw Before optimization 22.4273 1.5873 2.74 After optimization 5.0988 0.0925 1.65

[0169] Comparing the dispatch models before and after optimization, the model before optimization did not consider photovoltaic output and participation in demand-side response and electric vehicle models.

[0170] The electricity cost after optimization was reduced by 17.3285 yuan compared to before optimization. Of course, the invention uses a smaller load as a reference and has significant limitations. Regarding the user satisfaction index, the lower the value, the higher the satisfaction. Of course, the HAVC system's ability to regulate indoor temperature is also key to improving user satisfaction.

[0171] This example controls the indoor temperature within the range of 25 degrees Celsius and 27 degrees Celsius, which is the human comfort temperature. Figure 5 It can be seen that the peak-to-valley difference is reduced by 1.09Kw, which also enables it to achieve the effect of peak shaving and valley filling to a certain extent.

[0172] Figure 5 As shown in the figure, the fitness function of the total load curve before optimization does not take into account the weight coefficient and the grid stability objective function. Therefore, it can be clearly seen that the optimization reduces the peak-to-valley difference and enhances voltage stability, thereby playing a certain role in peak shaving and valley filling for the building load.

[0173] For the comparison before and after the improvement of the algorithm Figure 6 As shown in the figure, the optimization primarily improved the algorithm's inertia weight parameter. In the initial iterations, the two methods performed at similar speeds. The pre-optimization fitness curve clearly fell into a local optimum, breaking out of it after around 64 iterations, and reaching its optimal fitness value at 74 iterations. The post-optimization fitness value broke out of the local optimum much more quickly after 40 iterations, and the output fitness value was even better.

[0174] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes, replacements and improvements within the technical scope disclosed by the present invention are within the protection scope of the present invention.

Claims

1. A particle swarm optimization algorithm-based smart building electricity demand-side optimization scheduling strategy, characterized by: Classify the loads in the building and establish mathematical models for each type of load. Use the economic indicator electricity cost, user satisfaction, and reliability indicator grid stability as objective functions, and then use the particle swarm algorithm to iteratively find the optimal solution. The specific steps include: Step 1: Obtain the power consumption status of the building's controllable loads through the monitoring system. Based on the building's load characteristics, classify them into base loads, shiftable loads, curtailable loads, and both shiftable and adjustable loads. Step 2: Comprehensively consider the charging and discharging characteristics of electric vehicles and the output of the photovoltaic system, analyze the user's electric vehicle charging optimization strategy, and establish a scheduling optimization model. The scheduling optimization model includes: a shiftable load scheduling model, a curtailable load scheduling model, a shiftable and adjustable load scheduling model, a HVAC model, and an EV and PV model. The shiftable load dispatching model is: L TL，i =P TL，i Δt (1); Formula 1 is the translation load i at t i The electric energy consumed in the period; where i is the set of load types that can be translated i∈{l w ,l wh ,...l d };P TL,i is the translatable load i at t i Power of the time period; S TL,i Indicates the power consumption status of load i; is the actual working range of load i; To envision the best scheduling interval; The curtailable load dispatch model is: L in Formula 4 XJ.k Indicates the actual load reduction capacity, It represents the ideal load reduction capacity, r ki Indicates the fluctuation ratio coefficient of the reduction amount in different gears allowed by the contract incentive mechanism; The load scheduling model that can be translated and adjusted is: L FL，j =P FL，j S FL,j Δt (5); σ1P FL，j <P FL，j <σ2P FL，j (8); Formula 5 is the actual dispatch response of the shiftable and adjustable load j; where j is the set of shiftable and adjustable load types j∈{l w ,l wh ,...l d }number; P FL，j For a load that can be translated and adjusted, j is at t j Power of the time period; S FL,j Indicates the start and stop status of load j; is the actual working range of load j; is the optimal dispatching interval; σ in Equation 8 is the proportional coefficient of the load power adjustment; The HAVC model is: L HAVC =|P t |Δt (9); Where: |P t | is the HVAC power, the absolute value here is to distinguish between cooling and heating; Δt is the recording interval; K t =a H K t-1 +(1-a H )E t +a p P t (10); Where, P T is the temperature control load power in the T period, which is the optimization variable; a H is the building insulation coefficient, which is generally less than 1; a P K is the HVAC heating / cooling capacity coefficient; t is the indoor temperature variation law; K t-1 is the room temperature at the last moment; E t is the outdoor temperature; The EV and PV models are: Where SOC(t) is the target charging capacity of the electric vehicle; SOC(t0) is the remaining capacity of the electric vehicle before it is connected to the charging pile; P EV is the charging and discharging power of the electric vehicle; η EV is the electric vehicle power conversion coefficient; t-t0 is the electric vehicle charging time constant; E is the total amount of electricity that can be stored in the electric vehicle; P EV ≤P max (14); SOC min ≤SOC(t)≤SOC max (15); In order to ensure the battery life of electric vehicles, the charging and discharging power and the charge state of electric vehicles are constrained; The output power of the PV power generation system is: P in Equation 16 PV Indicates the output of the photovoltaic system, P ref The system standard output calculated based on the photovoltaic solar panel parameters; S t represents the solar radiation intensity at time t; α θ is the temperature coefficient, K so is the actual temperature of the solar panel at time t; S ref , K ref They are the standard solar radiation intensity and temperature of the photovoltaic system respectively; the standard output of the photovoltaic system is related to the voltage and current characteristics of the solar panel; Photovoltaic power generation systems are installed on buildings to minimize electricity costs. Although photovoltaic power generation devices are unstable, they can still play a role in peak load shifting in most periods. Step 3: Based on demand-side response, a DR incentive mechanism is proposed, which is divided into electricity price incentives and contract incentives. The electricity price incentives can be divided into fixed electricity prices, time-of-use electricity prices, and real-time electricity prices. The three electricity price mechanisms are in a progressive relationship. The simulation example mainly uses the time-of-use electricity price mechanism. The power grid company conducts load forecasts on the day before and proposes electricity price information for different time periods based on the forecast data. It transmits the electricity price information to the HEMS of residential users participating in the incentive mechanism, encouraging users to make reasonable electricity consumption strategies based on the time-of-use electricity price. The contract incentive mentioned above refers to a contract signed between the power grid and the user. Based on the grid's transient load conditions, the dispatch center will issue load reduction instructions to the user 30-120 minutes in advance to reduce the peak load. The user can then benefit from this. The incentive benefits are calculated based on the peak-shaving effect index and the different incentive fee coefficients on the tiered index to determine the specific benefits. C mot =H mot C dr ν mot H gxd (19); Formula 17 is the calculation formula for the peak regulation effect index, where: They represent the total power consumption of the user at time t and the average power consumption of the day, P sys (T) They represent the total power of the power system in the T period and the average power consumption of the day respectively; Formula 18 is the actual contribution of the user to the power grid, which is also an important indicator of incentive costs, among which H gxd is the actual contribution of the user to the grid; Equation 19 is the grid incentive fee formula, which is based on the DR electricity price standard set by the grid; Step 4: Establish the objective function and its constraints, taking the minimum electricity cost, user satisfaction and grid stability as the objective function, and use the linear weighted combination method to transform the multiple objectives into a single objective function; the objective function is: The optimization adjustment is carried out with the goal of minimizing electricity costs. Considering that in the case of excessive solar energy, users will sell the PV system output that cannot be absorbed to the power company for profit. λ is the electricity price at time t. With the goal of user satisfaction and grid stability, C fort Indicates user satisfaction. Compare before and after optimization. If the load usage time changes less, C fort The smaller the value, the higher the user satisfaction; the grid stability is expressed by the difference between the daily peak and valley power consumption; The linear weighted combination method is used to transform multiple objectives into a single objective function. The importance of each objective is multiplied by the corresponding weight coefficient, and finally the sum is added to form an objective function for solution. The single objective function after unified dimension is: F = β1F1 + β2F2 + β3F3 (23); For buildings like office buildings, the economic index F1 is taken as the most important index, and then the optimization strategy is used to achieve the effect of peak load reduction and valley filling for the power grid. The importance level of power grid stability F3 is higher than that of user satisfaction F2. The operator relationship between F1, F2, and F3 is judged one by one and normalized. Step 5: Solve the established scheduling optimization model using the improved particle swarm algorithm; The improved particle swarm algorithm is to add a link to update the particle swarm parameters at the iteration of the PSO algorithm to obtain nonlinearly decreasing particle swarm parameters; For conventional particle swarm optimization, the parameters are fixed as constants, and the inertia weight and learning factor are set as one-dimensional vectors. Then, Equations 24 and 25 are the update iterative formulas for the inertia weight and learning factor, respectively; where w max 、w min 、c max 、c min is the maximum and minimum value of the inertia weight and learning factor; d1 and d2 are dynamic error coefficients; N it 、 is the current number of iterations and the maximum number of iterations.

2. The particle swarm algorithm-based intelligent building electricity demand-side optimization scheduling strategy according to claim 1, characterized in that: The basic loads described in step 1 include lighting systems and televisions. These loads have high user demand and are not involved in system scheduling. The shiftable loads include printers and projectors. Users adjust the load power usage interval according to their actual situation. However, due to the load characteristics, the load power and power usage time cannot be adjusted. The overall load curve will be shifted. The load reduction mentioned above is that under the conditions of the contract signed between the power supply company and the user, the power supply company will issue a load reduction instruction to the user in advance based on the actual load operation situation, and the user can also receive financial compensation from it; The loads described as both movable and adjustable include electric vehicles and HVAC constant temperature systems. These loads offer significant scheduling flexibility for users, consume a significant portion of electricity, and are a key factor in user comfort. The characteristic of these loads is that their power consumption and duration can be flexibly adjusted to a certain extent. The grid provides electricity price information one day in advance, adopting a time-of-use (TOU) pricing incentive mechanism to encourage users to set electricity usage strategies.

3. The particle swarm algorithm-based intelligent building electricity demand-side optimization scheduling strategy according to claim 1, characterized in that: The improved particle swarm algorithm is used to iteratively search for the optimal solution. The specific steps are as follows: S51: Obtain building load data and outdoor temperature information, initialize the electric vehicle charge state, predict photovoltaic output, and perform normalization processing; S52: Set the particle swarm parameters, input the learning factor, inertia weight and number of iterations; S53: Initialize the population and randomly generate a population for the movable and adjustable loads and energy storage system; S54: Calculate the fitness function and input a mathematical model with minimum electricity cost, user satisfaction and grid stability index as the objective function; S55: Determine the global optimal fitness value, first take out the individual extreme value and the group extreme value, obtain the global and individual optimal, and determine the individual optimal fitness value and the global optimal fitness value; S56: Start iterating to find the optimal solution; S57: judging whether the power grid needs peak load regulation based on the real-time power load curve; S58: If not, dispatch the curtailable load based on the contract incentives and propose a final electricity utilization strategy; S59: Result analysis.

4. The particle swarm algorithm-based intelligent building electricity demand-side optimization scheduling strategy according to claim 3, characterized in that: The steps for starting the iterative search for the optimum in step S56 are: S561: Update rate and population; S562: Calculate and update the particle target fitness value; S563: Determine whether the particles have converged. If not, continue iteration; otherwise, output the optimal solution and provide a user-side scheduling strategy.

Images