Power utilization real-time scheduling method, system and equipment for machine room

Abstract

The invention relates to an electricity consumption real-time scheduling method, system and equipment for a machine room. A load electricity consumption prediction curve of a load in a peak and flat valley time period of one day is established according to historical data of load electricity consumption in the machine room; the time period is judged, and the real-time energy storage capacity of the energy storage system in the machine room and the real-time power consumption of the load are obtained in real time; and analyzing and comparing the real-time data with the load power consumption prediction curve, estimating whether the energy storage system can meet the power consumption demand of the load and the charging time required by the energy storage system, reducing the power consumption of the load when the energy storage system cannot meet the estimated power consumption demand of the load in the sharp time period and the peak time period, and reducing the charging time required by the energy storage system in the valley time period and the flat time period. Charging the energy storage system, and if the actual charging time is less than the estimated charging time, reducing the power consumption of the load; meanwhile, the invention discloses a system and equipment adopting the method. The power utilization in the machine room can be adjusted in real time, and benefit maximization is ensured.

Description

Technical Field

[0001] The present invention relates to the field of power consumption in a computer room, and in particular to a real-time power consumption scheduling method, system and equipment for a computer room. Background Art

[0002] At present, with the improvement of the information level of data centers and the implementation of green energy-saving projects in computer rooms, the monitoring and management of power consumption in computer rooms has become increasingly important. Energy storage systems are used in computer rooms as a fast two-way adjustment device for power and electric energy. Comprehensive energy storage optimization operation strategies and research on energy storage optimization configuration that considers the impact of energy storage optimization operation on operating costs are important foundations for investing in energy storage.

[0003] At present, the common site computer room energy storage management strategy in the industry is to only adjust the charging and discharging capacity of the energy storage system, that is, to statically set the charging and discharging time periods of the energy storage system according to the peak and valley periods through the EMS system (energy management system). In each time period, the energy storage system often operates at a single charging and discharging rate, which accelerates the aging of the energy storage system battery and increases the cost of operating the energy storage system.

[0004] The existing operation optimization strategy for energy storage systems based on the peak-valley electricity price difference is:

[0005] 1) Charging the energy storage during the off-peak electricity price period is the first charge;

[0006] 2) Discharge the energy storage during the morning peak period, which is the first discharge;

[0007] 3) Recharge the energy storage during the normal period at noon, which is the second charge;

[0008] 4) The energy storage is discharged again during the afternoon peak period, which is the second discharge.

[0009] Its disadvantage is that it cannot effectively issue corresponding charging and discharging strategies in real time and efficiently according to actual power consumption, nor can it link other equipment in the area to form an overall adjustable load to participate in optimized operation, and it cannot maximize the benefits in terms of battery safety and revenue. Summary of the invention

[0010] In view of the existing deficiencies, the present invention provides a method, system and device for real-time scheduling of electricity consumption in a computer room.

[0011] The technical solution adopted by the present invention to solve the technical problem is: a real-time power dispatching method for a computer room, comprising:

[0012] Based on the historical data of load power consumption in the computer room, a load power consumption prediction curve for the peak and valley time periods of the day is established;

[0013] Determine the time period and obtain real-time data in the computer room in real time, the real-time data including the real-time energy storage capacity of the energy storage system in the computer room and the real-time power consumption of the load;

[0014] Analyze and compare real-time data and load power consumption forecast curves, and estimate whether the energy storage system can meet the load power demand and the charging time required for the energy storage system based on the difference between the real-time energy storage capacity of the energy storage system and the real-time power consumption of the load and the estimated value of the load power consumption forecast curve. When the energy storage system cannot meet the load power demand during peak time periods, the load power consumption is reduced. During valley time periods and normal time periods, the energy storage system is charged. If the actual charging time is less than the estimated charging time, the load power consumption is reduced.

[0015] Preferably, the load power consumption prediction curve is established by preprocessing the historical data of load power consumption and then training and learning the feature data set through a recurrent neural network algorithm to obtain the load power consumption prediction curve.

[0016] Preferably, the preprocessing includes: performing differential encoding based on historical data to obtain a compressed differential data sequence, performing outlier detection on the differential data sequence to obtain marked valid data values, and then performing interpolation processing on the valid data values ​​to obtain standard data.

[0017] Preferably, the interpolation process is a linear interpolation process, and the formula is as follows: t =(X t-1 +X t+1 ) / 2, where X t is the data after outlier filling, X t-1 and X t+1 It is the valid data of the time period adjacent to the outlier time period.

[0018] Preferably, the estimated charging time is calculated by the following formula: t = 1.2*(Q 额 -Q 剩 )*R / T r *I, where: t is the estimated charging time, Q 额 is the rated capacity of the energy storage system, Q 剩 is the real-time energy storage capacity of the energy storage system, T r is the charging efficiency of the energy storage system, R is the attenuation rate of the energy storage system, and I is the charging current of the energy storage system.

[0019] Preferably, the real-time energy storage capacity of the energy storage system is acquired within a preset time interval.

[0020] A real-time power dispatching system for a computer room, comprising:

[0021] The real-time data subsystem is used to obtain and store the real-time data of the load and energy storage system in the computer room, and analyze and compare the real-time data and the load power consumption forecast curve;

[0022] Real-time control subsystem, used to control the execution of instructions;

[0023] The load forecasting subsystem is used to establish a load power consumption forecasting curve for the peak and valley periods of the day based on the historical data of load power consumption in the computer room;

[0024] The safety operation and maintenance subsystem is used to monitor the safe operation of the load and energy storage system in the computer room, and to control the load and energy storage system through the real-time control subsystem based on the safety status information;

[0025] The intelligent dispatching subsystem generates corresponding control instructions for the load and energy storage system in the computer room through the analysis results of the real-time data subsystem and / or the information of the safety operation and maintenance subsystem, and sends them to the real-time control subsystem.

[0026] Preferably, the safety operation and maintenance subsystem includes an alarm processing module and an intelligent patrol module. The alarm processing module is used to process the safety status information generated by the intelligent patrol module and determine whether an alarm is required, and to issue an alarm message when an alarm is required; the intelligent patrol module is used to monitor the operating status of the load and energy storage system in the computer room and generate safety information and transmit it to the alarm processing module.

[0027] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the computer program stored in the memory, so as to implement the real-time electricity scheduling method for a computer room as described in any of the preceding items.

[0028] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the real-time power scheduling method for a computer room as described in any of the preceding items is implemented.

[0029] The beneficial effect of the present invention is that the present invention can issue corresponding charging and discharging strategies in real time according to the real-time power consumption of the actual load and energy storage system in the base station room, combined with the load power consumption prediction curve, to achieve a scheduling strategy for battery safety protection and maximum profit. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the principle of the scheduling method according to an embodiment of the present invention;

[0031] Figure 2 is a structural block diagram of a scheduling system according to an embodiment of the present invention;

[0032] Names and serial numbers of components in the figure: 1- Real-time data subsystem 2- Real-time control subsystem 3- Load forecasting subsystem 4- Safety operation and maintenance subsystem 40- Alarm processing module 41- Intelligent inspection module 5- Intelligent scheduling subsystem. DETAILED DESCRIPTION

[0033] In order to more clearly illustrate the purpose, technical solutions and advantages of the embodiments of the present invention, the present invention will be further described in conjunction with the accompanying drawings and embodiments below for a clear and complete description. Obviously, the described embodiments are partial embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention. In addition, the directional terms mentioned in the present invention, such as "up", "down", "front", "back", "left", "right", "inside", "outside", etc., are only referenced to the directions of the attached diagrams. The directional terms used are for better and clearer explanation and understanding of the present invention, rather than indicating or implying the orientation that the present invention must have, and therefore cannot be understood as limitations on the present invention.

[0034] Embodiments of the present invention Figure 1 As shown in , a real-time power scheduling method for a computer room includes:

[0035] Based on the historical data of load power consumption in the computer room, a load power consumption prediction curve for the peak, flat and valley time periods of the day is established; the historical data includes information such as the location of the site, whether it is a weekday or a holiday, weather conditions, etc., which can be obtained from the corresponding database; the load includes communication equipment load, air-conditioning equipment load, etc. Among these loads, some loads can be adjusted in power consumption, which are called adjustable loads, such as air conditioners, and some loads cannot be adjusted in power consumption, which are called fixed loads; the peak and valley time periods are the peak time periods, peak time periods, valley time periods and flat time periods divided by the power system, and the divisions are different in different places. For example, according to the following division, 23:00-6:00 is the valley time period, 6:00-8:00, 12:00-16:00, 21:00-23:00 are flat time periods, and 8:00-9:00 and 16:00-21:00 are peak time periods. , 9:00-12:00 is the peak time period; the load power consumption prediction curve is the prediction curve of all load power consumption in the computer room; at this time, the step of establishing the load power consumption prediction curve is to preprocess the historical data of load power consumption and then train and learn the feature data set through the recurrent neural network algorithm to obtain the load power consumption prediction curve; the preprocessing of historical data includes: differential encoding based on historical data to obtain a compressed differential data sequence, and performing outlier detection on the differential data sequence to obtain the marked valid data value, and then interpolating the valid data value to obtain the standard data; differential encoding is to reduce data redundancy by the difference between adjacent data points. For example, the load power consumption data sequence of a site computer room during normal working hours and good weather in the peak time period of 8:00-9:00 is Q(t1), Q(t2)Q(t3),..., Q(t n ), the data sequence after differential encoding is ⊿Q(t i )=Q(t i )-Q(t i-1 ), t represents time, i.e. a certain day, and Q represents electricity consumption; then the compressed differential data sequence is tested for outliers, and outliers are identified and marked. Statistical methods, such as the isolation forest algorithm based on machine learning, can be used to detect outliers; after the outlier detection is completed, valid data values ​​are obtained, and then the valid data values ​​are interpolated to fill the outliers, thereby obtaining continuous standard data values. The interpolation process selects linear interpolation, and its formula is as follows: X t =(X t-1 +X t+1 ) / 2, where X t is the data after outlier filling, X t-1 and X t+1is the valid data value of the time period adjacent to the abnormal value time period; when performing the recurrent neural network algorithm, factors affecting power consumption such as the location of the computer room, weather conditions, whether it is a weekday or a holiday, which time period is the peak or valley, the number and type of loads in the computer room, etc. are used as input variables of the recurrent neural network model, such as the LSTM neural network. The standard data is divided into test data and training data, the LSTM network architecture is defined, and the LSTM neural network model is created. The created LSTM neural network model is an LSTM network with two hidden layers, each layer has a number of hidden units, and the drop layer with a certain drop rate is set to prevent overfitting. The training data is trained for a set number of rounds using the ADAM algorithm, and the gradient threshold is set to prevent gradient explosion. The initial learning rate is specified, and the learning rate is reduced by multiplying the factor after a certain number of rounds of training is performed; after the training is completed, the test data is used to predict the data for the next few days and cross-validate, the predicted data is de-standardized and the error is calculated, and the prediction result is found as the load power consumption prediction curve;

[0036] Determine the time period and obtain the real-time data in the computer room in real time. The real-time data includes the real-time energy storage capacity of the energy storage system in the computer room and the real-time power consumption of the load. That is to say, first determine whether the time period is the peak time period, the valley time period, or the flat time period, and perform corresponding scheduling according to the time period and the real-time data obtained in the computer room, so as to achieve a scheduling strategy that maximizes battery safety protection and benefits; the real-time power consumption of the load can be calculated through the load running time and power; the real-time energy storage capacity of the energy storage system is acquired within the preset time interval, that is, the parameter data of the energy storage system is acquired within the preset time interval, such as once every 60 seconds, each acquisition time is 2-4 seconds and starts at the 0th second of each minute. The collected parameter data includes the voltage, current, charge and discharge time, and charge and discharge times of the rechargeable battery in the energy storage system, and then these parameters are combined with the battery charging efficiency, attenuation rate, and rated capacity to calculate the real-time energy storage capacity of the energy storage system.

[0037] Analyze and compare the real-time data and the load power consumption forecast curve, that is, analyze and compare the real-time energy storage capacity of the energy storage system and the real-time power consumption of the load with the load power consumption forecast curve, and estimate whether the energy storage system can meet the power demand of the load and the charging time required by the energy storage system based on the difference between the real-time energy storage capacity of the energy storage system and the real-time power consumption of the load and the estimated value of the load power consumption forecast curve. If the real-time energy storage capacity of the energy storage system is greater than the difference between the real-time power consumption of the load and the estimated value of the load power consumption forecast curve, it means that the remaining capacity of the energy storage system can support the use of the load in the computer room. At this time, no matter which time period is in the peak time period, peak time period, valley time period or flat time period, the load in the computer room is running normally and the energy storage system does not need to be charged; if it is less than, it means that the remaining capacity of the energy storage system cannot meet the power demand of the load. At this time, in the peak time period and peak time period, the load power consumption is reduced. Reducing the load power consumption means reducing the power consumption of the adjustable load, such as air conditioning, Figure 1 The figure shows the scheduling of air conditioning power consumption in the computer room. In this case, the energy storage system is stopped from charging. During the valley time and the normal time, the energy storage system is charged. If the actual charging time is less than the estimated charging time, the load power consumption is reduced. In other words, during the valley time and the normal time, if the remaining capacity of the energy storage system is insufficient, it is charged. If the actual charging time, that is, the duration of the valley time or the normal time, is less than the estimated charging time, the load power consumption is reduced and charging is performed again when the next valley time or the normal time is reached. The estimated charging time is calculated by the following formula: t = 1.2*(Q 额 -Q 剩 )*R / T r *I, where: t is the estimated charging time, Q 额 is the rated capacity of the energy storage system, Q 剩 is the real-time energy storage capacity of the energy storage system, T r is the charging efficiency of the energy storage system, R is the attenuation rate of the energy storage system, and I is the charging current of the energy storage system.

[0038] A real-time power dispatching system for a computer room, such as Figure 2 As shown in, including:

[0039] The real-time data subsystem 1 is used to obtain and store the real-time data of the load and the energy storage system in the computer room. The real-time data includes the real-time energy storage capacity of the energy storage system in the computer room and the real-time power consumption of the load. At the same time, the real-time data and the load power consumption prediction curve are analyzed and compared, that is, the real-time energy storage capacity of the energy storage system and the real-time power consumption of the load are analyzed and compared with the load power consumption prediction curve, and the results of the analysis and comparison are transmitted to the intelligent scheduling subsystem 5;

[0040] The real-time control subsystem 2 is used for executing the control instructions, receiving the control instructions from the intelligent scheduling subsystem 5 and then executing the control instructions on the corresponding load;

[0041] The load forecasting subsystem 3 is used to establish a load power consumption forecasting curve for the peak and valley time periods of the load in the computer room based on the historical data of load power consumption in the computer room. That is, the historical data of load power consumption is preprocessed and then the characteristic data set is trained and fitted by a recurrent neural network algorithm to obtain the load power consumption forecasting curve;

[0042] The safety operation and maintenance subsystem 4 is used to monitor the safe operation of the load and energy storage system in the computer room, and control the load and energy storage system through the real-time control subsystem 2 according to the safety status information. The safety operation and maintenance subsystem 4 is used to monitor whether the computer room is running normally. When an abnormal situation occurs, it transmits the abnormal information to the intelligent energy scheduling subsystem 5 and issues an alarm. When the intelligent scheduling subsystem 5 receives the abnormal information, it issues corresponding control instructions to the real-time control subsystem 2 to control the corresponding equipment; the safety operation and maintenance subsystem 4 includes an alarm processing module 40 and an intelligent inspection module 41. The alarm processing module 40 is used to process the safety status information generated by the intelligent inspection module 41 and determine whether an alarm is required. If an alarm is required, an alarm message is issued, that is, if there is an abnormality in the equipment in the computer room, the alarm processing module 40 will alarm and send an alarm message to the intelligent scheduling subsystem 5; the intelligent inspection module 41 is used to monitor the operating status of the load and energy storage system in the computer room and generate safety information and transmit it to the alarm processing module 40;

[0043] The intelligent dispatching subsystem 5 generates corresponding control instructions for the load and energy storage system in the computer room through the analysis results of the real-time data subsystem 1 and / or the information of the safety operation and maintenance subsystem 4 and sends them to the real-time control subsystem 2. That is to say, the intelligent dispatching subsystem 5 issues corresponding control instructions when it receives any one or more of the analysis results of the real-time data subsystem 1 and the alarm information of the safety operation and maintenance subsystem 4.

[0044] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the computer program stored in the memory to implement the real-time electricity scheduling method for a computer room as described in any of the preceding items.

[0045] A computer-readable storage medium stores a computer program. The computer-readable storage medium may include: any entity or device, medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory and software distribution medium, etc. that can carry the computer program code. The computer program code can be in any existing code form. When the computer program is executed by a processor, the real-time electricity scheduling method for a computer room as described in any of the preceding items is implemented.

[0046] Although the present invention has been described in detail above with general descriptions and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made on the basis of the present invention. Therefore, these modifications or improvements made on the basis of not departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A real-time power dispatching method for a computer room, characterized in that: include: Based on the historical data of load power consumption in the computer room, a load power consumption prediction curve for the peak and valley time periods of the day is established; Determine the time period and obtain real-time data in the computer room in real time, the real-time data including the real-time energy storage capacity of the energy storage system in the computer room and the real-time power consumption of the load; Analyze and compare real-time data and load power consumption forecast curves, and estimate whether the energy storage system can meet the load power demand and the charging time required for the energy storage system based on the difference between the real-time energy storage capacity of the energy storage system and the real-time power consumption of the load and the estimated value of the load power consumption forecast curve. When the energy storage system cannot meet the load power demand during peak time periods, the load power consumption is reduced. During valley time periods and normal time periods, the energy storage system is charged. If the actual charging time is less than the estimated charging time, the load power consumption is reduced.

2. The real-time power dispatching method for a computer room according to claim 1 is characterized in that The establishment of the load power consumption prediction curve is specifically as follows: after preprocessing the historical data of load power consumption, the characteristic data set is trained and learned by a recurrent neural network algorithm to obtain the load power consumption prediction curve.

3. The real-time power dispatching method for a computer room according to claim 2 is characterized in that The preprocessing includes: performing differential encoding based on historical data to obtain a compressed differential data sequence, performing outlier detection on the differential data sequence to obtain marked valid data values, and then performing interpolation processing on the valid data values ​​to obtain standard data.

4. The real-time power dispatching method for a computer room according to claim 3 is characterized in that The interpolation process is a linear interpolation process, and its formula is as follows: t =(X t-1 +X t+1 ) / 2, where X t is the data after outlier filling, X t-1 and X t+1 It is the valid data of the time period adjacent to the outlier time period.

5. The real-time power dispatching method for a computer room according to claim 1 is characterized in that The estimated charging time is calculated by the following formula: t = 1.2*(Q 额 -Q 剩 )*R / T r *I, where: t is the estimated charging time, Q 额 is the rated capacity of the energy storage system, Q 剩 is the real-time energy storage capacity of the energy storage system, T r is the charging efficiency of the energy storage system, R is the attenuation rate of the energy storage system, and I is the charging current of the energy storage system.

6. The real-time power dispatching method for a computer room according to claim 1 is characterized in that The real-time energy storage capacity of the energy storage system is acquired within a preset time interval.

7. A real-time power dispatching system for a computer room, characterized in that: include: The real-time data subsystem is used to obtain and store the real-time data of the load and energy storage system in the computer room, and analyze and compare the real-time data and the load power consumption forecast curve; Real-time control subsystem, used to control the execution of instructions; The load forecasting subsystem is used to establish a load power consumption forecasting curve for the peak and valley periods of the day based on the historical data of load power consumption in the computer room; The safety operation and maintenance subsystem is used to monitor the safe operation of the load and energy storage system in the computer room, and control the load and energy storage system through the real-time control subsystem based on the safety status information; The intelligent dispatching subsystem generates corresponding control instructions for the load and energy storage system in the computer room through the analysis results of the real-time data subsystem and / or the information of the safety operation and maintenance subsystem, and sends them to the real-time control subsystem.

8. The real-time power dispatching system for a computer room according to claim 7 is characterized in that: The safety operation and maintenance subsystem includes an alarm processing module and an intelligent inspection module. The alarm processing module is used to process the safety status information generated by the intelligent inspection module and determine whether an alarm is required, and issue an alarm message when an alarm is required; the intelligent inspection module is used to monitor the operating status of the load and energy storage system in the computer room and generate safety information and transmit it to the alarm processing module.

9. An electronic device, characterized in that : includes a memory and a processor, the memory stores a computer program, and the processor is used to execute the computer program stored in the memory to implement the real-time electricity scheduling method for a computer room as described in any one of claims 1-6.

10. A computer-readable storage medium, characterized in that : The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the real-time electricity scheduling method for a computer room described in any one of claims 1-6 is implemented.

Images