An Optimal Scheduling Method for Ice Storage Air Conditioning Based on Load Forecasting
By optimizing the equipment control of the ice storage air conditioning system using a particle swarm optimization algorithm based on load prediction, the problems of high electricity costs and high equipment energy consumption in the existing technology are solved, thereby reducing electricity costs and energy consumption and improving the economic efficiency of the ice storage air conditioning system.
Patent Information
- Application Number
- CN202111618565.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-29
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2041-12-29
AI Technical Summary
The existing control strategies for ice storage air conditioning systems cannot effectively reduce electricity costs and equipment energy consumption; instead, they increase initial investment costs, leading to reduced economic efficiency.
The operation strategy of the ice storage air conditioning system is optimized by using a particle swarm optimization algorithm based on load forecasting. Through load forecasting model and equipment control model, the equipment control of the ice storage air conditioning system is optimized, including the particle swarm optimization of ethylene glycol outlet temperature, cooling water inlet temperature, chilled water circulation pump frequency and cooling tower fan frequency, so as to achieve the goal of minimizing total electricity cost and equipment energy consumption.
This resulted in a 6.9% reduction in electricity costs for building air conditioning systems and an 11.5% reduction in equipment energy consumption, while also reducing peak grid load.
Smart Images

Figure CN114282729B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the intersection field of building energy management and artificial intelligence, and specifically relates to an ice storage air conditioning optimization scheduling method based on load prediction. Background Art
[0002] Building operation energy consumption is an important aspect of my country's energy consumption, and air conditioning energy consumption accounts for a large proportion of building operation. As a type of energy storage, the ice storage system uses ethylene glycol solution as a coolant. It saves users' electricity costs and reduces the peak load of the power grid by making ice during low electricity price periods and melting ice during high electricity price periods.
[0003] Ice storage air conditioners often operate in combination with ice storage tank melting conditions and dual-condition chiller refrigeration conditions. At present, most control strategies are to prioritize ice melting or distribute cooling loads to chillers, which causes the ice in the ice storage tank to be used up in advance and chillers to be in low load rate conditions, increasing electricity costs and equipment energy consumption. This fails to achieve the goal of saving costs and reducing the peak load of the power grid. Instead, the economic efficiency of the ice storage air conditioner system is reduced due to the increase in initial investment. Summary of the invention
[0004] In view of this, the present invention provides an ice storage air conditioning optimization scheduling method based on load prediction, which predicts the hourly cooling load and adopts a particle swarm algorithm to optimize the operation strategy, and controls the ice storage tank, dual-mode chiller, cold water circulation pump, and cooling tower fan of the ice storage air conditioning system to simultaneously achieve the two operating goals of minimizing the total electricity cost of the air conditioning system within a day and minimizing the total energy consumption of each equipment hourly.
[0005] To achieve the above purpose, the present invention proposes an ice storage air conditioning optimization scheduling method based on load forecasting, and its technical route is as follows: Figure 1 As shown, the following steps are included:
[0006] Step 1: Combined with the characteristics of the forecast day, correlation analysis is used to determine the input characteristics of the load forecast model, including historical daily load, meteorological parameters, and building internal parameters. The historical daily load is determined by calculating the correlation using the angle cosine method, and the meteorological parameters and building internal parameters are determined by calculating the Spearman correlation coefficient;
[0007] Step 2: Select temperature difference classification, divide the data set into historical similar days and training days according to the characteristics of the prediction day and historical day, use the training day data as the training set to train the neural network, correct the parameters and weights of the last two layers of the neural network according to the historical similar day data, and use the trained prediction model for load prediction to obtain the load prediction value Q pre,t ;
[0008] Step 3: Collect the cooling capacity Q of the dual-condition cooling system L , electric power P, glycol outlet water temperature Tchw , Cooling water inlet temperature T cow , establish a dual-condition chiller cop model; collect the flow rate M of the cold water circulation pump chw , frequency n chw Data, establish the cold water circulation pump frequency-flow support vector machine regression (SVR) model; collect the cooling tower inlet water temperature T in , water outlet temperature T out , Outdoor dry bulb temperature T g , relative humidity RH, outdoor wind speed V, cooling tower fan frequency n tower , establish n tower About T in , T out , T g , RH, V cooling tower SVR model;
[0009] Step 4: Get the load forecast value Q in step 2) pre,t , hourly electricity price λ t , fixed chiller glycol outlet water temperature T chw , Cooling water inlet temperature T cow , the particle swarm optimization algorithm is used to calculate the hourly cooling capacity Q of the ice storage tank ice,t Cooling capacity Q of dual-mode refrigeration system chiller,t , under the constraints, the total electricity cost in one day is minimized;
[0010] Step 5: Obtain the hourly cooling capacity Q from step 4 ice,t and cooling capacity Q chiller,t , combined with the hourly load forecast value Q in step 2 pre,t , introduce the dual-condition chiller cop model, cold water circulation pump SVR model, and cooling tower SVR model established in step 3, and use particle swarm optimization to obtain the hourly optimal ethylene glycol outlet water temperature T chw , Cooling water inlet temperature T cow , cold water circulation pump frequency n chw , cooling tower fan frequency n tower , so that the electricity cost per hour is minimized and the control of the main equipment of the air-conditioning system is completed. The optimization objective function of step 4 and step 5 is:
[0011]
[0012] Beneficial Effects
[0013] 1. The present invention predicts building load based on neural network technology, adopts angle cosine method and Spearman correlation coefficient to determine characteristic variables, and improves prediction accuracy; introduces historical similar days, modifies the prediction model, and further improves prediction accuracy.
[0014] 2. The present invention optimizes the distribution of building cooling load through particle swarm algorithm, reduces electricity costs, and cuts peak load.
[0015] 3. The present invention optimizes the control strategy of ice storage air conditioning equipment, improves the energy efficiency of each equipment, and reduces the energy consumption of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A technical roadmap of an ice storage air conditioning optimization scheduling method based on load forecasting of the present invention;
[0017] Figure 2 A comparison diagram of the predicted load and the actual operating load in one embodiment of the present invention;
[0018] Figure 3 A diagram showing the cooling load distribution result in one embodiment of the present invention;
[0019] Figure 4 This is a diagram showing the effect of optimizing equipment energy consumption in one embodiment of the present invention. DETAILED DESCRIPTION
[0020] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.
[0021] The present invention provides an ice storage air conditioning optimization scheduling method based on load forecasting, and its flow chart is as follows: Figure 1 As shown, the following steps are included:
[0022] Step 1: Quantify the day of the week, weather, date, and temperature, and combine them into historical day characteristics. Use the angle cosine method to calculate the correlation between the predicted day and the historical day, and select the load of the historical day with the largest correlation in the previous month as the historical similar day load; use the Spearman correlation coefficient to determine the selected meteorological parameters and building internal parameter characteristics. The load forecasting model characteristics include: historical daily load, dry bulb temperature, maximum temperature, minimum temperature, minimum temperature, air pressure, body temperature, body temperature, air conditioning terminal power consumption, ventilation power consumption, lighting power consumption;
[0023] Step 2: Calculate the temperature change value and average temperature in a day, determine the temperature difference classification, and determine the historical similar day and training day based on the classification; first use the training day to train the neural network, and then use the historical similar day data to modify the last two layers of the neural network parameters and connection weights.
[0024] Specifically, the comparison between the predicted load and the actual operating load in this embodiment is as follows: Figure 2As shown. After calculation, the load forecasting accuracy index CVRMSE after introducing historical similar day correction is 5.46%, which meets the requirement that the forecast error in the project is no more than 30%. This shows that this model can predict building loads.
[0025] Step 3: Collect the cooling capacity Q of the dual-condition cooling system L , electric power P, glycol outlet water temperature T chw , Cooling water inlet temperature T cow , establish a dual-condition chiller cop model; collect the flow rate M of the cold water circulation pump chw , frequency n chw Data, establish the cold water circulation pump frequency-flow support vector machine regression (SVR) model; collect the cooling tower inlet water temperature T in , water outlet temperature T out , Outdoor dry bulb temperature T g , relative humidity RH, outdoor wind speed V, cooling tower fan frequency n tower , establish n tower About T in 、T out 、T g , RH, V cooling tower SVR model;
[0026] Specifically, the dual-condition chiller COP model is a linear model, expressed as formula (3). The chilled water pump circulation pump and cooling tower modeling adopts the support vector machine regression method, and the Gaussian kernel function is used to map the low-dimensional expression to the high-dimensional space. The relationship between the input and output variables is shown in formula (4) and formula (5):
[0027] cop=β 1 PLR+β 2 PLR 2 +β 3 ·PLR·T chw +β 4 ·T cow +β 5 ·PLR·T cow +β 6 (3)
[0028] n chw =f 1 (M chw ) (4)
[0029] n tower =f 2 (T in , T out , T g , RH, V) (5)
[0030] In formula (3), PLR is the partial load rate of the dual-mode chiller, β is a constant obtained by fitting; in formula (4) and formula (5), f 1 (·),f 2 (·) is the SVR regression model.
[0031] Step 4: Get the load forecast value Q in step 2 pre,t , hourly electricity price λ t , fixed chiller glycol outlet water temperature T chw , Cooling water inlet temperature T cow , the particle swarm optimization algorithm is used to calculate the hourly cooling capacity Q of the ice storage tank ice,t Cooling capacity Q of dual-mode refrigeration system chiller,t , while satisfying the constraints, minimize the total electricity cost in one day.
[0032] Specifically, the constraints are:
[0033]
[0034]
[0035] In formula (6), count{X, Y} represents the number of times X and Y events occur simultaneously in the optimized time length, which aims to avoid the frequent start and stop of the dual-condition cooling machine.
[0036] In this embodiment, the prediction results of step 2 are used to optimize the allocation, and the results are as follows: Figure 3 Compared with the ice melting priority strategy, it can reduce electricity costs by 6.9%.
[0037] Step 5: Obtain the hourly cooling capacity Q from step 4 ice,t and cooling capacity Q chiller,t , combined with the hourly load forecast value Q in step 2 pre,t , introduce the dual-condition chiller cop model, cold water circulation pump SVR model, and cooling tower SVR model established in step 3, and use particle swarm optimization to obtain the hourly optimal ethylene glycol outlet water temperature T chw , Cooling water inlet temperature T cow , cold water circulation pump frequency n chw , cooling tower fan frequency n tower , so as to minimize the electricity cost at each hour and complete the control of the main equipment of the air-conditioning system.
[0038] In this embodiment, the cooling machine, water pump and cooling tower are optimized and controlled within one day, and the energy consumption optimization effect is as follows: Figure 4 Compared with the original control strategy, the optimized proposed strategy can reduce energy consumption by 11.5%.
Claims
1. An optimal scheduling method for ice storage air conditioning based on load forecasting, It is characterized in that The following steps are involved: Step 1): Combined with the characteristics of the forecast day, correlation analysis is used to determine the input characteristics of the load forecast model, including historical daily loads, meteorological parameters, and building internal parameters; Step 2): Select temperature difference classification, divide the data set into historical similar days and training days according to the characteristics of the prediction day and the historical day, use the training day data as the training set to train the neural network, modify the neural network model according to the historical similar day data, and use the trained prediction model for load prediction to obtain the load prediction value Q pre,t ; Step 3): Collect the cooling capacity Q of the dual-condition cooling system L , electric power P, ethylene glycol outlet temperature T chw , Cooling water inlet temperature T cow , establish a dual-condition chiller cop model; collect the flow rate M of the cold water circulation pump chw , frequency n chw Data, establish the cold water circulation pump frequency-flow support vector machine regression SVR model; collect the cooling tower inlet water temperature T in , water outlet temperature T out , Outdoor dry bulb temperature T g , relative humidity RH, outdoor wind speed V, cooling tower fan frequency n tower , establish n tower About T in 、T out 、T g , RH, V cooling tower SVR model; Step 4): Get the load forecast value Q in step 2) pre,t , hourly electricity price λ t , fixed chiller glycol outlet water temperature T chw , Cooling water inlet temperature T cow , the particle swarm optimization algorithm is used to calculate the hourly cooling capacity Q of the ice storage tank ice,t Cooling capacity Q of dual-mode refrigeration system chiller,t , under the constraints, the total electricity cost in one day is minimized; Step 5): Obtain the hourly cooling capacity Q from step 4) ice,t and cooling capacity Q chiller,t , combined with step 2) hourly load forecast value Q pre,t , introduce the dual-condition chiller cop model, cold water circulation pump SVR model, and cooling tower SVR model established in step 3), and use particle swarm algorithm optimization to obtain the hourly optimal ethylene glycol outlet water temperature T chw , Cooling water inlet temperature T cow , cold water circulation pump frequency n chw , cooling tower fan frequency n tower , so as to minimize the electricity cost at each hour and complete the control of the main equipment of the air-conditioning system.
2. According to the load forecasting-based ice storage air conditioning optimization scheduling method of claim 1, It is characterized in that In the steps 1) and 2), the day of the week, weather, date, and temperature need to be quantified, and the angle cosine method is used to calculate the correlation between the predicted day and the historical day, and the load of the historical day with the largest correlation in the previous month is selected as one of the input features; the Spearman correlation coefficient is calculated to determine the selected meteorological parameters and building internal parameter characteristics; Calculate the temperature change value and average temperature in a day, determine the temperature difference classification, and determine the historical similar day and training day based on the classification; first use the training day to train the neural network, and then use the historical similar day data to modify the last two layers of the neural network parameters and connection weights.
3. According to the load forecasting-based ice storage air conditioning optimization scheduling method of claim 1, It is characterized in that In step 4) and step 5), a multi-objective hierarchical optimization method is adopted, that is, the dispatching layer optimizes the cooling load distribution to ensure the minimum electricity cost within a day, and the control layer optimizes the air conditioning system operating parameters to ensure the minimum hourly energy consumption of the system. The objective expression is formula (1), so that J 1 , J 2 At the same time, the minimum: In formula (1), cop t Indicates the hourly cooling coefficient of the dual-condition chiller, cop rate Indicates the rated cooling coefficient, Q ice Represents the total ice storage capacity of the ice storage tank, λ v Indicates the valley electricity price, W chiller,t Indicates hourly cooling energy consumption, W chw,t Indicates the energy consumption of the cold water circulation pump, W cow,t Indicates the energy consumption of cooling water circulation pump, W tower,t Represents the energy consumption of the cooling tower.
