Energy system optimization scheduling method and system based on data center load transfer

Abstract

The invention discloses an energy system optimization scheduling method and system based on data center load transfer, and relates to the technical field of power system optimization scheduling, and the method comprises the steps: collecting data through a sensor, and carrying out the preprocessing; constructing a comprehensive energy system model based on the data center based on the preprocessed data; constructing a target function and a constraint condition of a calculation task transfer model; and an optimization problem of the target function is converted into a mixed integer linear programming problem, and the problem is solved by using a branch and bound method. According to the method, the comprehensive energy system model and the calculation task transfer model are constructed, so that efficient transfer of data center loads and optimal scheduling of energy are realized, the energy utilization efficiency is remarkably improved, the operation cost is reduced, and the environmental influence is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization and scheduling, and in particular to an energy system optimization and scheduling method and system based on data center load transfer. Background Art

[0002] Efficient uninterruptible power supply and sustainable energy system are essential to achieve the sustainability and environmental friendliness of data centers. The scientific and technological revolution and industrial revolution represented by digitalization are driving fundamental changes in human lifestyles, production methods and ways of thinking. A large amount of data needs to be sensed, stored, encrypted, transmitted, called, calculated and traded. The amount of global data is growing at an average rate of 40% per year. Data centers (DCs) are places where centralized electronic information equipment is operated. As an important carrier of digital infrastructure, the scale of DCs will continue to expand, and the "computing power" will continue to increase. Therefore, the energy consumption and operating costs of DC will continue to rise. For example, a typical DC can consume the electricity of 25,000 households, and it consumes 100-20 times the energy of a standard office of the same size. In China, the overall market size of cloud computing reached 94.61 billion yuan, an increase of 32.4%, and the overall energy consumption of DCs reached 120 billion to 130 billion kWh, accounting for about 2% of China's total electricity consumption. Globally, developing countries will become the world's largest energy consumers, rising from 3% in 2017 to 4.5% in 2025.

[0003] Accelerating the construction of green DC, actively promoting safe and controllable energy-saving technologies and renewable energy applications, and integrating the concept of green development into all aspects of DC construction are important issues that need to be urgently addressed in DC energy conservation and emission reduction in my country and the world. Google, Microsoft, Yahoo, and Apple are already using renewable energy to power their DC and reduce their reliance on traditional fuels. Google is the first company to purchase 100% renewable energy to meet its annual electricity consumption for global operations (including DC) to make its DC sustainable. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to optimize the energy consumption and cost-effectiveness of the entire energy system while ensuring the operational efficiency of the data center, and at the same time solve the problems of low efficiency, high cost and great environmental impact of existing methods in energy allocation and task transfer.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: an energy system optimization scheduling method based on data center load transfer, which comprises the following steps:

[0007] Collect data through sensors for pre-processing;

[0008] Build a data center-based integrated energy system model based on the preprocessed data;

[0009] Construct the objective function and constraints of the computational task transfer model;

[0010] The optimization problem of the objective function is converted into a mixed integer linear programming problem and solved using the branch and bound method.

[0011] As a preferred solution of the energy system optimization scheduling method based on data center load transfer described in the present invention, wherein: the data includes photovoltaic unit data, gas turbine data, absorption chiller data, ground source heat pump data, organic Rankine cycle data, gas boiler data and heat exchanger data;

[0012] The preprocessing includes converting non-digital environmental data and operation data into digital codes, standardizing all data, achieving consistency of data of different dimensions, processing missing values, using linear interpolation or spline interpolation to fill missing values ​​for time series data, using mean or median to fill missing values ​​for non-time series data, removing outliers, and detecting and removing outliers through the 3σ principle.

[0013] As a preferred solution of the energy system optimization scheduling method based on data center load transfer described in the present invention, wherein: the comprehensive energy system model based on the data center includes a photovoltaic unit model, a gas turbine model, an absorption chiller model, a ground source heat pump model, an organic Rankine cycle model, a gas boiler model and a heat exchanger model;

[0014] The photovoltaic unit model is expressed as:

[0015]

[0016]

[0017] Where, f is the power derating factor of the photovoltaic unit, N pv is the installed capacity of the photovoltaic unit, G p is the local solar radiation intensity, G stc is the solar radiation density under standard test environment, α is the temperature coefficient, T pv,p is the surface temperature of the photovoltaic panel, T pv,stc is the temperature of the photovoltaic panel under standard test conditions, T a,p is the ambient temperature, T pv,soc is the surface temperature of photovoltaic power generation under standard conditions, η e,pv is the photovoltaic efficiency under standard test environment, τ is the solar transmittance of the photovoltaic cover, β is the solar absorption rate, G soc is the standard solar radiation intensity, Ta,soc is the standard ambient temperature;

[0018] The gas turbine model is expressed as:

[0019] E ge =F ge η ge,e

[0020] Q r =F ge η ge,q η hr

[0021]

[0022] Among them, E ge is the power generation of the gas turbine, F ge is the natural gas consumption, Q r is the waste heat power generation, η ge,e , η ge,q , η hr are the power generation efficiency, waste heat efficiency and heat recovery efficiency of the gas turbine, V ge,gas is the natural gas volume, LHV gas It is the low calorific value of natural gas;

[0023] The absorption refrigerator model is expressed as:

[0024] Q ra =Q ac / COP ac

[0025] Among them, Q ra is the heat absorbed by the absorption refrigerator, Q ac is the cooling output of the absorption chiller, COP ac is the coefficient of performance;

[0026] The ground source heat pump model is expressed as:

[0027]

[0028]

[0029] Among them, E gshp,c and E gshp,h are the power consumption of the ground source heat pump in cooling and heating modes, Q gshp,c and Q gshp,h are cooling and heating output, COP gshp,c and COP gshp,h are the coefficient of performance in cooling and heating modes, respectively;

[0030] The organic Rankine cycle model is expressed as:

[0031] E ORC =Q ORC η ORC

[0032] Among them, E ORC is the output power of the organic Rankine cycle system, Q ORC , η ORC are the input heat and power efficiency respectively;

[0033] The gas boiler model is expressed as:

[0034] V b =Q b / η b LHV gas

[0035] Among them, LHV gas is the lower calorific value of natural gas, V b is the natural gas consumption of gas boiler, Q b , η b They are the heat output of the gas boiler and the thermal efficiency of the furnace;

[0036] The heat exchanger model is expressed as:

[0037] Q hx =Q rh η hx

[0038] Among them, Q hx is the heat output of the heat exchanger, Q rh , η hx are the heat input and thermal efficiency of the heat exchanger respectively.

[0039] As a preferred solution of the energy system optimization scheduling method based on data center load transfer described in the present invention, the objective function of the computing task transfer model is expressed as follows:

[0040] minATC=C cap +C om +C op

[0041] Among them, C cap , C om , C op They are the annual average cost of initial investment cost, maintenance cost of components, and operating cost;

[0042] The objective function of the computing task transfer model includes the annual average cost, the maintenance cost of the IES, and the operating cost of the IES, and the expression is:

[0043]

[0044]

[0045] C op =C gas +C grid +C tax +C pe

[0046] Among them, N m , C m are the installed capacity and unit investment cost of the mth equipment, i and n are the discount rate and equipment life, l is the number of equipment, ε is the proportional coefficient of investment cost, C gas , C gr id are the natural gas cost and electricity purchase cost respectively, C tax is the carbon tax, C pe The penalty cost for curtailing photovoltaic power.

[0047] As a preferred solution of the energy system optimization scheduling method based on data center load transfer described in the present invention, the natural gas cost, electricity purchase cost, carbon tax and penalty cost of photovoltaic power abandonment are expressed as follows:

[0048]

[0049]

[0050]

[0051]

[0052] Among them, c gas is the natural gas price, F gas,j is the hourly consumption of natural gas, c pe,j is the hourly electricity price of the power grid, E grid,j is the electricity purchased from the grid every hour, is the unit carbon tax, is the hourly carbon dioxide emissions, c PV is the unit penalty cost of PV abandonment, They are photovoltaic power generation and DC power consumption per hour respectively.

[0053] As a preferred solution of the energy system optimization scheduling method based on data center load transfer described in the present invention, wherein: the constraint conditions include power balance constraint, cooling load balance constraint, thermal energy balance constraint, and gas turbine equipment capacity constraint;

[0054] The power balance constraint is expressed as:

[0055] E ge +E grid +E ORC +E PV =E gshp1 +E gshp2 +E req

[0056] Among them, E ge , E grid , E ORC and E PV are the power generation of gas turbine, power grid, organic Rankine cycle and photovoltaic generator, E gshp1 、E gshp2 are the power consumption of ground source heat pumps 1 and 2, E req is the electrical load of the end user;

[0057] The cooling load balance constraint is expressed as:

[0058] Q c =Q c,gshp1 +Q gshp2 +Q ac

[0059] Q c =Q gshp2 +Q ac

[0060] Among them, Q c,gshp1 , Q gshp2 are the cooling outputs of ground source heat pumps 1 and 2, Q ac is the cooling output and cooling load of the absorption chiller;

[0061] The thermal energy balance constraint is expressed as:

[0062] Q h =Q hc +Q h,gshp1

[0063] Q b +Q r =Q orc +Q ra +Q rh +Q loss

[0064] Among them, Q hc , Q h,gshp1 , Q b are the heat outputs of the heat exchanger, ground source heat pump No. 1 and boiler, Q r is the waste heat from the gas turbine, Q orc , Q ra and Qrh are the heat inputs of the organic Rankine cycle, absorption refrigerator and heat exchanger, Q loss is the heat loss;

[0065] The gas turbine equipment capacity constraint is expressed as:

[0066] α on N ge S ge ≤E ge ≤N ge S ge

[0067] Among them, α on is the starting factor of the gas turbine, E ge is the power generation of the gas turbine, N ge , S ge are two decision variables.

[0068] As a preferred solution of the energy system optimization scheduling method based on data center load transfer described in the present invention, wherein: the conversion of the optimization problem of the objective function into a mixed integer linear programming problem is to obtain a linear programming problem by relaxing the original MILP problem, and to obtain a non-integral numerical solution of the linear programming by using the simplex method. In the IES optimization model, equation α on N ge S ge ≤E ge ≤N ge S ge N ge , S ge The nonlinear term caused by the two decision variables is converted into a linear term by introducing the infinite number M, and the expression is:

[0069]

[0070] Among them, E ge is the power generation of the gas turbine, N ge , S ge are two decision variables, M is a large positive number used to represent sufficiently large values ​​in integer programming to ensure the logical constraints in the optimization model, α on is the starting factor of the gas turbine, which indicates the minimum power generation of the gas turbine in the starting state;

[0071] There are nonlinear terms in the gas turbine equipment capacity constraints in the IES optimization model. After introducing the infinite number M to convert the nonlinear terms into linear terms, N ge , S geThe two non-integer decision variables are divided into two nearest integers, and the outputs obtained by the power balance constraint, cooling load balance constraint, thermal energy balance constraint, gas turbine equipment capacity constraint, natural gas cost model, power purchase cost model, carbon tax model and photovoltaic power abandonment penalty cost model are brought into the IES operating cost model to obtain the minimum IES operating cost C op , and finally the annual average cost C of the initial investment cost cap , component maintenance cost C om and IES operating costs C op By outputting the upper and lower limits of the objective function, the optimal solution with the minimum cost minATC is obtained, and the operating cost of the customer satisfaction (CSL) is minimized by matching the calculation task with the energy flow.

[0072] Another object of the present invention is to provide an energy system optimization and scheduling system based on data center load transfer, which can effectively optimize energy usage and cost by integrating a comprehensive energy system model and a computing task transfer model, thereby solving the shortcomings of existing systems in real-time data processing, energy utilization efficiency and cost control.

[0073] In order to solve the above technical problems, the present invention provides the following technical solutions: an energy system optimization scheduling system based on data center load transfer, comprising: a data preprocessing module, a comprehensive energy system model construction module, a computing task transfer model construction module and an optimization solution module;

[0074] The data preprocessing module collects sensor data and preprocesses the data, converts non-digital environmental data and operation data into digital codes, standardizes all data, completes the consistency of data of different dimensions, processes missing values, fills missing values ​​with linear interpolation or spline interpolation for time series data, fills missing values ​​with mean or median for non-time series data, removes outliers, and detects and removes outliers through the 3σ principle;

[0075] The integrated energy system model building module builds a mathematical model of the integrated energy system of the data center based on the preprocessed data;

[0076] The computing task transfer model building module is to build the objective function and constraint conditions of computing task transfer;

[0077] The optimization solution module converts the objective function and its constraints into a mixed integer linear programming problem and solves it by using the branch and bound method.

[0078] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the energy system optimization scheduling method based on data center load transfer as described above are implemented.

[0079] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the energy system optimization scheduling method based on data center load transfer as described above.

[0080] Beneficial effects of the present invention: By constructing a comprehensive energy system model and a computing task transfer model, the present invention realizes efficient transfer of data center loads and optimal scheduling of energy, significantly improves energy utilization efficiency, reduces operating costs and reduces environmental impact. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0082] Figure 1 An overall flow chart of an energy system optimization scheduling method based on data center load transfer provided by the first embodiment of the present invention;

[0083] Figure 2 A structural diagram of an energy system optimization and scheduling system based on data center load transfer provided by a second embodiment of the present invention;

[0084] Figure 3 A diagram showing the energy balance effect before and after load transfer in the energy system optimization scheduling method based on data center load transfer provided in the third embodiment of the present invention. DETAILED DESCRIPTION

[0085] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0086] Example 1

[0087] Reference Figure 1 According to an embodiment of the present invention, a method for optimizing and scheduling an energy system based on data center load transfer is provided, comprising:

[0088] Firstly, a comprehensive energy system model based on a data center is constructed. The model consists of photovoltaic power generation modules (PV), gas turbines (GE), absorption chillers (AC), ground source heat pumps (GSHP), organic Rankine cycles (ORC), gas boilers (GB), and heat exchangers (HX).

[0089] The specific model is as follows:

[0090] 1) PV model:

[0091]

[0092] Where f is the power derating factor of the photovoltaic unit, N pv is the installed capacity of the photovoltaic unit (kW), G p is the local solar radiation intensity (kW / m2), G stc is the solar radiation density under standard test environment, 1kW / m2, α is the temperature coefficient (% / ℃), T pv,p is the surface temperature of the photovoltaic panel (℃), T pv,stc is the temperature of the photovoltaic panel under standard test conditions, 25°C, T pv,p Calculated by the following formula.

[0093]

[0094] Among them, T a,p is the ambient temperature (℃), T pv,soc is the surface temperature of photovoltaic power generation under standard working conditions (45-48°C), η e,pv is the photovoltaic efficiency under the standard test environment, τ is the solar transmittance of the photovoltaic cover, β is the solar absorption rate, τβ=0.9, and the standard solar radiation intensity G soc =0.8kW / m2, standard ambient temperature T a,soc =20℃.

[0095] 2) GE model:

[0096] E ge =F ge η ge,e

[0097] Q r =F ge η ge,q η hr

[0098]

[0099] Among them, E ge is GE's power generation (kW), F ge is the natural gas consumption, Qr is the waste heat power generation (kW), η ge,e , η ge,q , η hr are power generation efficiency, waste heat efficiency and heat recovery efficiency, V ge,gas is the amount of natural gas (m3), LHV gas It is the low calorific value of natural gas.

[0100] 3) AC model

[0101] Q ra =Q ac / COP ac

[0102] Among them, Q ra is the heat absorbed by the absorption refrigerator, Q ac is the cooling output of the absorption chiller, COP ac is the performance coefficient.

[0103] 4) GSHP model

[0104]

[0105]

[0106] Among them, E gshp,c and E gshp,h are the power consumption of the ground source heat pump in cooling and heating modes, Q gshp,c and Q gshp,h are cooling and heating output, COP gshp,c and COP gshp,h are the coefficient of performance in cooling and heating modes, respectively.

[0107] 5) ORC model

[0108] E ORC =Q ORC η ORC

[0109] Among them, E ORC is the output power of the organic Rankine cycle system, Q ORC , η ORC are the input heat and power efficiency respectively;

[0110] 6) GB model

[0111] V b =Q b / η b LHV gas

[0112] Among them, LHV gas is the lower calorific value of natural gas, Vb is the natural gas consumption of gas boiler, Q b , η b They are the heat output of the gas boiler and the thermal efficiency of the furnace.

[0113] 7) HX Model

[0114] Q hx =Q rh η hx

[0115] Among them, Q hc is the heat output of the heat exchanger (kW), Q rh is the heat input (kW), η hc is the thermal efficiency of the heat exchanger.

[0116] Then, the objective function and constraints of the computing task transfer model are constructed for the integrated energy system model based on the data center.

[0117] Construct the objective function of the computational task transfer model:

[0118] minATC=C cap +C om +C op

[0119] Among them, C cap is the annual average cost of the initial investment cost, C om is the maintenance cost of the component, C op for operating costs including fuel, electricity purchased from the grid and carbon taxes.

[0120] The average annual cost is defined as:

[0121]

[0122] Among them, N m , C m They represent the installed capacity and unit investment cost ($ / kW) of the mth equipment, i is the discount rate, n is the equipment life, and l is the number of equipment.

[0123] The maintenance cost of IES is defined as:

[0124]

[0125] Among them, ε is the proportional coefficient of investment cost.

[0126] The operating costs of IES are defined as:

[0127] C op =C gas +C grid +Ctax +C pe

[0128] Among them, C gas , C grid are the natural gas cost and the electricity purchase cost respectively. According to their unit prices and consumption, C tax is the carbon tax, C pe is the penalty cost of photovoltaic power abandonment. They are respectively expressed as:

[0129]

[0130]

[0131]

[0132]

[0133] Among them, c gas is the natural gas price, F gas,j is the hourly consumption of natural gas, c pe,j is the hourly electricity price of the power grid, E grid,j is the electricity purchased from the grid every hour, is the unit carbon tax, is the hourly carbon dioxide emissions, c PV is the unit penalty cost of PV abandonment, They are photovoltaic power generation and DC power consumption per hour respectively.

[0134] Next, construct the constraints for computing the task transfer model:

[0135] 1) Power balance constraints

[0136] E ge +E grid +E ORC +E PV =E gshp1 +E gshp2 +E req

[0137] Among them, E ge , E grid , E ORC and E PV are the power generation of GE, grid, ORC and PV respectively, E gshp1 、E gshp2 are the power consumption of #1 and #2 GSHP, E req It is the electrical load of the end user.

[0138] In the traditional separated production (SP) system, the power load is fully met by the power grid, that is, the power balance constraint of the SP system is expressed as:

[0139]

[0140] in, is electricity purchased from the grid, Power consumption of GSHP in SP system.

[0141] 2) Cooling load balance constraints

[0142] During the non-heating period, GSHP#1 adopts cooling mode, with two GSHPs and absorption chillers providing cooling. During the winter heating period, only GSHP#2 and absorption chillers are used for cooling, that is, the cooling load balance in IES is expressed as:

[0143] Q c =Q c,gshp1 +Q gshp2 +Q ac

[0144] Q c =Q gshp2 +Q ac

[0145] Among them, Q c,gshp1 , Q gshp2 are the cooling outputs of GSHP#1 and #2, Q ac is the cooling output and cooling load of the absorption chiller. In the traditional SP system, the required cooling load is fully met by the GSHP, that is, the load balance of the SP system is expressed as:

[0146]

[0147] in, is the cooling output of GSHP in the SP system.

[0148] 3) Thermal energy balance constraints

[0149] The system heat balance constraints include the heat balance of the heating system and the balance of waste heat utilization of the system. GE generates a lot of heat during the power generation process, which can drive the ORC system, AC and heat exchanger to provide electricity, cooling and heat respectively, while the unrecovered heat is released to the environment, that is, the heat balance in IES is expressed as:

[0150] Q h =Q hc +Q h,gshp1

[0151] Q b +Q r =Q orc +Q ra +Q rh +Qloss

[0152] Among them, Q hc , Q h,gshp1 and Q b are the heat outputs of the heat exchanger, GSHP#1 and boiler, respectively, Q r It's GE's waste heat, Q orc , Q ra and Q rh are the heat inputs of the ORC system, AC and heat exchanger, Q loss is heat loss.

[0153] In winter, the heat load is borne by the boiler, that is, the heat balance of the SP system is expressed as:

[0154]

[0155] in, is the output heat of the boiler (kW), η hx is the efficiency of the heat exchanger.

[0156] 4) GE equipment capacity constraints

[0157] α on N ge S ge ≤E ge ≤N ge S ge

[0158] Among them, α on It is the promoter of GE (α on =25%)Set αα on Designed to constrain GE's operation at higher load rates.

[0159] 5) Other equipment operating constraints on power and heat

[0160]

[0161]

[0162] Finally, the optimization problem is reformulated as a mixed integer linear programming, which is solved and evaluated using the branch and bound method in MATLAB.

[0163] It is further explained that:

[0164] 1) By relaxing the original MILP problem, a linear programming problem is obtained, and the non-integral numerical solution of the linear programming is obtained using the simplex method. As the IES optimization model, equation α on N ge S ge ≤E ge≤N ge S ge The nonlinear terms caused by the two decision variables Nge and Sge can be converted into linear terms by introducing the infinite number M:

[0165]

[0166] Among them, E ge is the power generation of the gas turbine, N ge , S ge are two decision variables, M is a large positive number used to represent sufficiently large values ​​in integer programming to ensure the logical constraints in the optimization model, α on It is the starting factor of the gas turbine, which indicates the minimum power generation of the gas turbine in the starting state.

[0167] Non-integer-valued decision variables are divided into the nearest integers. Conditions are added to the original problem to form two sub-problems (or branches) to solve them separately. Thus, an upper bound (upper limit) or lower bound on the objective function value can be obtained to obtain a limit (lower bound), from which the optimal solution can be obtained.

[0168] It should be further explained that there are nonlinear terms in the gas turbine equipment capacity constraints in the IES optimization model. After introducing the infinite number M to convert the nonlinear terms into linear terms, N ge , S ge The two non-integer decision variables are divided into two nearest integers, and the outputs obtained by the power balance constraint, cooling load balance constraint, thermal energy balance constraint, gas turbine equipment capacity constraint, natural gas cost model, power purchase cost model, carbon tax model and photovoltaic power abandonment penalty cost model are brought into the IES operating cost model to obtain the minimum IES operating cost C op , and finally the annual average cost C of the initial investment cost cap , component maintenance cost C om and IES operating costs C op By outputting the upper and lower limits of the objective function, the optimal solution with the minimum cost minATC is obtained, and the operating cost of the customer satisfaction (CSL) is minimized by matching the calculation task with the energy flow.

[0169] 2) Evaluation of the plan:

[0170] Compared with the SP system, three indicators of IES, namely annual cost savings rate (ACSR), carbon dioxide emission reduction rate (CDERR) and primary energy savings rate (PESR), are used to evaluate the comprehensive performance of the optimal solution.

[0171] The detailed calculation of the three indicators is as follows:

[0172]

[0173]

[0174]

[0175] Where o is the annual operating hours, ATC ref and ATC IES are the annual operating hours and annual total cost (ATC) of SP and IES, respectively, CDE ref 、CDE IES are the carbon dioxide emissions (CDE) and fuel consumption of SP and IES, respectively. ref and F IES are the fuel consumption of SP and IES respectively.

[0176] Example 2, reference Figure 2 , which is an embodiment of the present invention, provides a system for an energy system optimization scheduling method based on data center load transfer, including: a data preprocessing module, a comprehensive energy system model building module, a computing task transfer model building module and an optimization solution module.

[0177] The data preprocessing module collects sensor data and preprocesses the data, converts non-digital environmental data and operation data into digital codes, standardizes all data, completes the consistency of data of different dimensions, handles missing values, and uses linear interpolation or spline interpolation to fill missing values ​​for time series data. For non-time series data, the mean or median is used to fill missing values, removes outliers, and detects and removes outliers through the 3σ principle.

[0178] The integrated energy system model building module builds a mathematical model of the data center's integrated energy system based on preprocessed data.

[0179] The computational task transfer model building module is used to build the objective function and constraints of computational task transfer.

[0180] The optimization solution module converts the objective function and its constraints into a mixed integer linear programming problem and solves it using the branch and bound method.

[0181] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0182] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0183] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0184] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiment, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0185] Example 3

[0186] In this embodiment, in order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments. This embodiment experiments the existing traditional method and the method of this embodiment respectively.

[0187] The present invention considers a data center containing 3100 cabinets (18 servers per cabinet), and the IT area includes 56000 processors and auxiliary components such as power distributors and lights. The hourly cooling load and heating load of the office area are simulated in DesT software, and the hourly power load of the office area is obtained according to the operation strategy of lights and equipment in DesT simulation model. The real-time utilization of processors and network loads conforms to normal distribution and is generated in Monte Carlo simulation.

[0188] Table 1 Cost comparison before and after load transfer

[0189]

[0190] The whole year can be divided into heating period and non-heating period. Due to the low ambient temperature in winter, the proportion of DC cooling load in the heating season is much smaller than that in the non-heating season. The maximum annual DC power load is 9768.19 kW, the cooling load is 8948.93 kW, and the heating load is 3290.24 kW.

[0191] By comparing the performance after load transfer, it is found that load transfer has the greatest impact on CO2 emissions. Compared with the initial situation, the CDERR in the non-heating period and heating period reached 6.31% and 4.11%, respectively, and the ACSR was 2.76% and 3.47%, respectively. Due to the low solar radiation intensity in the heating period, the penalty cost for abandoning light before load transfer is relatively small, about 10,000 yuan.

[0192] Load shifting mainly reduces operating costs by responding to real-time electricity prices to reduce the cost of electricity purchase from 24,300 yuan to 19,900 yuan. As a result, the OCSR in the non-heating period is 3.47%. The PESR changes the least, which is 2.56% and 2.17% in the non-heating period and heating period, respectively.

[0193] Therefore, load transfer achieved through optimized task scheduling will not waste photovoltaic power generation, reduce IES's dependence on the grid, and achieve zero electricity purchase during average and peak electricity prices. When optimizing the power load, the cooling load is optimized, and the two are closely related. The proportion of AC cooling is increased, and more GE waste heat is utilized.

[0194] The present invention also proposes an IES consisting of solar photovoltaic (PV), gas engine (GE) and organic Rankine cycle (ORC), and constructs an optimization model to determine the component capacity to minimize the total annual cost. Compared with the individual optimizations, the proposed system configuration and the comprehensive optimization of energy and computational flow management strategies have achieved more benefits. The proposed optimization method is used to perform the optimal configuration of the IES, where the optimal capacity of the configuration is shown in Table 2, and the energy balance before and after the system load transfer is shown in Figure 3 The proposed design and scheduling optimization method with computational tasks provides an effective tool for the integration between IES and DC.

[0195] Table 2 Optimal capacity configuration

[0196]

[0197] It can be seen from Table 1 that due to the addition of gas turbines, organic Rankine cycle, photovoltaics, and absorption chillers in IES, when optimizing their capacity, the energy conversion forms between the integrated energy systems are more diverse, so compared with the individual optimization, the configuration capacity of IES's ground source heat pump #2, gas boiler, and heat exchanger is reduced.

[0198] pass Figure 3 It can be seen that the load shifting achieved by optimizing task scheduling achieves no waste of photovoltaic power generation, reduces IES's dependence on the grid, and achieves zero electricity purchase during average and peak electricity prices. When optimizing the power load, the cooling load is also optimized, and the two are closely related, increasing the cooling ratio generated by air conditioning and utilizing more GE waste heat.

[0199] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An energy system optimization scheduling method based on data center load transfer, characterized in that: include: Collect data through sensors for pre-processing; Build a data center-based integrated energy system model based on the preprocessed data; Construct the objective function and constraints of the computational task transfer model; The optimization problem of the objective function is converted into a mixed integer linear programming problem and solved using the branch and bound method.

2. The energy system optimization scheduling method based on data center load transfer according to claim 1, characterized in that: The data includes photovoltaic unit data, gas turbine data, absorption chiller data, ground source heat pump data, organic Rankine cycle data, gas boiler data and heat exchanger data; The preprocessing includes converting non-digital environmental data and operation data into digital codes, standardizing all data, achieving consistency of data of different dimensions, processing missing values, using linear interpolation or spline interpolation to fill missing values ​​for time series data, using mean or median to fill missing values ​​for non-time series data, removing outliers, and detecting and removing outliers through the 3σ principle.

3. The energy system optimization scheduling method based on data center load transfer according to claim 2 is characterized in that: The comprehensive energy system model based on the data center includes a photovoltaic unit model, a gas turbine model, an absorption chiller model, a ground source heat pump model, an organic Rankine cycle model, a gas boiler model and a heat exchanger model; The photovoltaic unit model is expressed as: Where, f is the power derating factor of the photovoltaic unit, N pv is the installed capacity of the photovoltaic unit, G p is the local solar radiation intensity, G stc is the solar radiation density under standard test environment, α is the temperature coefficient, T pv,p is the surface temperature of the photovoltaic panel, T pv,stc is the temperature of the photovoltaic panel under standard test conditions, T a,p is the ambient temperature, T pv,soc is the surface temperature of photovoltaic power generation under standard conditions, η e,pv is the photovoltaic efficiency under standard test environment, τ is the solar transmittance of the photovoltaic cover, β is the solar absorption rate, G soc is the standard solar radiation intensity, T a,soc is the standard ambient temperature; The gas turbine model is expressed as: Ege=Fgeηge,e Q r =F ge or ge,q or hr Among them, Ege is the power generation of the gas turbine, F ge is the natural gas consumption, Q r is the waste heat power generation, η ge,e , η ge,q , η hr are the power generation efficiency, waste heat efficiency and heat recovery efficiency of the gas turbine, V ge,gas is the natural gas volume, LHV gas It is the low calorific value of natural gas; The absorption refrigerator model is expressed as: Q ra =Q ac / COP ac Among them, Q ra is the heat absorbed by the absorption refrigerator, Q ac is the cooling output of the absorption chiller, COP ac is the performance coefficient; The ground source heat pump model is expressed as: Among them, E gshp,c and E gshp,h are the power consumption of the ground source heat pump in cooling and heating modes, Q gshp,c and Q gshp,h are cooling and heating output, COP gshp,c and COP gshp,h are the coefficient of performance in cooling and heating modes, respectively; The organic Rankine cycle model is expressed as: E ORC =Q ORC or ORC Among them, E ORC is the output power of the organic Rankine cycle system, Q ORC , η ORC are the input heat and power efficiency respectively; The gas boiler model is expressed as: V b =Q b / η b LHV gas Among them, LHV gas is the lower calorific value of natural gas, V b is the natural gas consumption of gas boiler, Q b , η b They are the heat output of the gas boiler and the thermal efficiency of the furnace; The heat exchanger model is expressed as: Q hx =Q rh the hx Among them, Q hx is the heat output of the heat exchanger, Q rh , η hx are the heat input and thermal efficiency of the heat exchanger respectively.

4. The energy system optimization scheduling method based on data center load transfer according to claim 3 is characterized in that: The objective function of the computing task transfer model is expressed as: minATC=C cap +C om +C op Among them, C cap , C om , C op They are the annual average cost of initial investment cost, maintenance cost of components, and operating cost; The objective function of the computing task transfer model includes the annual average cost, the maintenance cost of the IES, and the operating cost of the IES, and the expression is: C op =C gas +C grid +C tax +C pe Among them, N m , C m are the installed capacity and unit investment cost of the mth equipment, i and n are the discount rate and equipment life, l is the number of equipment, ε is the proportional coefficient of investment cost, C gas , C grid are the natural gas cost and the electricity purchase cost respectively, C tax is the carbon tax, C pe The penalty cost for curtailing photovoltaic power.

5. The energy system optimization scheduling method based on data center load transfer according to claim 4 is characterized in that: The IES operating cost includes natural gas cost, electricity purchase cost, carbon tax and penalty cost for photovoltaic power abandonment, and the expression is: Among them, c gas is the natural gas price, F gas,j is the hourly consumption of natural gas, c pe,j is the hourly electricity price of the power grid, E grid,j is the electricity purchased from the grid every hour, is the unit carbon tax, is the hourly carbon dioxide emissions, c PV is the unit penalty cost of PV abandonment, They are photovoltaic power generation and DC power consumption per hour respectively.

6. The energy system optimization scheduling method based on data center load transfer according to claim 5 is characterized in that: The constraints include power balance constraints, cooling load balance constraints, heat balance constraints, and gas turbine equipment capacity constraints; the constraints are input into the following constraints using the output data obtained from the gas turbine model, organic Rankine cycle model, and photovoltaic unit model: The power balance constraint is expressed as: AND ge +E grid +E ORC +E PV =And gshp1 +E gshp2 +E req Among them, E ge , E grid , E ORC and E PV are the power generation of gas turbine, power grid, organic Rankine cycle and photovoltaic generator, E gshp1 、E gshp2 are the power consumption of ground source heat pumps 1 and 2, E req is the electrical load of the end user; The cooling load balance constraint, the input parameter comes from the output parameter obtained from the absorption refrigerator model, and the expression is: Q c =Q c,gshp1 +Q gshp2 +Q ac Q c =Q gshp2 +Q ac Among them, Q c,gshp1 , Q gshp2 are the cooling outputs of ground source heat pumps 1 and 2, Q ac is the cooling output and cooling load of the absorption chiller; The heat balance constraint, the input parameters are from the output parameters obtained from the gas boiler model and the heat exchange model, and the expression is: Q h =Q hc +Q h,gshp1 Q b +Q r =Q orc +Q ra +Q rh +Q loss Among them, Q hc , Q h,gshp1 , Q b are the heat outputs of the heat exchanger, ground source heat pump No. 1 and boiler, Q r is the waste heat from the gas turbine, Q orc , Q ra and Q rh are the heat inputs of the organic Rankine cycle, absorption refrigerator and heat exchanger, Q loss is the heat loss; The gas turbine equipment capacity constraint, the input parameter comes from the output parameter obtained by the gas turbine model, and the expression is: a on N ge S ge ≤E ge ≤N ge S ge Among them, α on is the starting factor of the gas turbine, E ge is the power generation of the gas turbine, N ge , S ge are two decision variables.

7. The energy system optimization scheduling method based on data center load transfer according to claim 6 is characterized in that: The objective function optimization problem is converted into a mixed integer linear programming problem by relaxing the original MILP problem to obtain a linear programming problem, and using the simplex method to obtain a non-integral numerical solution of the linear programming. In the IES optimization model, equation α on N ge S ge ≤E ge ≤N ge S ge N ge , S ge The nonlinear term caused by the two decision variables is converted into a linear term by introducing the infinite number M, and the expression is: Among them, E ge is the power generation of the gas turbine, N ge , S ge are two decision variables, M is a large positive number used to represent sufficiently large values ​​in integer programming to ensure the logical constraints in the optimization model, α on is the starting factor of the gas turbine, which indicates the minimum power generation of the gas turbine in the starting state; There are nonlinear terms in the gas turbine equipment capacity constraints in the IES optimization model. After introducing the infinite number M to convert the nonlinear terms into linear terms, N ge , S ge The two non-integer decision variables are divided into two nearest integers, and the outputs obtained by the power balance constraint, cooling load balance constraint, thermal energy balance constraint, gas turbine equipment capacity constraint, natural gas cost model, power purchase cost model, carbon tax model and photovoltaic power abandonment penalty cost model are brought into the IES operating cost model to obtain the minimum IES operating cost C op , and finally the annual average cost C of the initial investment cost cap , component maintenance cost C om and IES operating costs C op By outputting the upper and lower limits of the objective function, the optimal solution with the minimum cost minATC is obtained, and the operating cost of the customer satisfaction (CSL) is minimized by matching the calculation task with the energy flow.

8. A system using the energy system optimization scheduling method based on data center load transfer as claimed in any one of claims 1 to 7, characterized in that: It includes data preprocessing module, integrated energy system model building module, computing task transfer model building module and optimization solution module; The data preprocessing module collects sensor data and preprocesses the data, converts non-digital environmental data and operation data into digital codes, standardizes all data, completes the consistency of data of different dimensions, processes missing values, fills missing values ​​with linear interpolation or spline interpolation for time series data, fills missing values ​​with mean or median for non-time series data, removes outliers, and detects and removes outliers through the 3σ principle; The integrated energy system model building module builds a mathematical model of the integrated energy system of the data center based on the preprocessed data; The computing task transfer model building module is to build the objective function and constraint conditions of computing task transfer; The optimization solution module converts the objective function and its constraints into a mixed integer linear programming problem and solves it by using the branch and bound method.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the energy system optimization scheduling method based on data center load transfer described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the energy system optimization scheduling method based on data center load transfer described in any one of claims 1 to 7 are implemented.

Images