Power supply device, power supply system and data center

Abstract

The present invention relates to the field of power supply technology, specifically to a power supply device, a power supply system and a data center. The present invention obtains the first data and the second data of the previous time window of each company in the park, and inputs them into a power demand prediction model and a power loss prediction model respectively. The power demand prediction model preprocesses the first data through a preprocessing layer, and the feature extraction layer extracts historical power consumption, socio-economic and weather characteristics. The feature analysis layer uses an ARIMA model and a multivariate linear regression model to generate a first power demand function. Combined with the initial power allocation value and constraints, the comprehensive power demand allocation satisfaction is optimized through a particle swarm algorithm, and a global optimal power allocation plan is obtained. The power loss prediction model preprocesses the second data, extracts current, resistance, temperature and humidity characteristics, and generates a first power loss function through a linear regression model. The invention improves the overall efficiency of the power supply system by optimizing power distribution and loss prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply, and in particular to a power supply device, a power supply system and a data center. Background Art

[0002] With the rapid development of smart grid technology, the traditional power supply system is gradually shifting towards intelligence and automation. Smart grids can not only monitor and manage the production, transmission and distribution of electricity in real time, but also accurately predict and optimize power demand through technologies such as big data analysis and machine learning. This technological change provides strong support for the efficient operation, stability and reliability of the power supply system.

[0003] Power supply management in modern parks increasingly relies on data-driven decision-making. By deploying smart meters, sensors, and IoT devices, the power supply system can collect a large amount of operating data in real time, including current, voltage, power factor, environmental parameters, etc. The real-time collection and analysis of this data enables the power supply system to predict power loss based on actual conditions.

[0004] However, there are still some problems in the power supply management of the park. Due to the high volatility of electricity demand of various companies in the park, it is difficult for the traditional power supply system to accurately predict the electricity demand of each company, which can easily lead to unreasonable power distribution among companies in the park, thus affecting the operating stability and efficiency of the power supply system.

[0005] To this end, a power supply device, a power supply system and a data center are proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a power supply device, a power supply system and a data center. The present invention relates to the field of power supply technology, specifically to a power supply device, a power supply system and a data center. By obtaining the first data and the second data of the previous time window of each company in the park, they are input into the power demand prediction model and the power loss prediction model respectively. The power demand prediction model preprocesses the first data through the preprocessing layer, the feature extraction layer extracts the historical power consumption, socio-economic and weather characteristics, and the feature analysis layer uses the ARIMA model and the multivariate linear regression model to generate the first power demand function. Combined with the initial power allocation value and the constraints, the comprehensive power demand allocation satisfaction is optimized by the particle swarm algorithm, and the global optimal power allocation plan is obtained. The power loss prediction model preprocesses the second data, extracts the current, resistance, temperature and humidity characteristics, and generates the first power loss function through the linear regression model. The invention improves the overall efficiency of the power supply system by optimizing power distribution and loss prediction.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A power supply device, comprising:

[0009] A data acquisition unit, used to acquire the first data and the second data of the previous time window of each company in the park;

[0010] The power demand prediction unit is used to input the first data into the power demand prediction model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; the preprocessing layer preprocesses the first data to obtain first detection data; the feature extraction layer extracts features from the first detection data to obtain historical power consumption features, socio-economic features and weather features; the feature analysis layer analyzes the three features through an ARIMA model and a multivariate linear regression model to obtain a first power demand function;

[0011] A power demand matching optimization unit, configured to obtain a comprehensive power demand allocation satisfaction degree according to the first power demand function and the initial power allocation value, and optimize the comprehensive power demand allocation satisfaction degree according to constraint conditions and a particle swarm algorithm to obtain a global optimal power allocation solution;

[0012] The power loss prediction unit is used to input the second data into the power loss prediction model, and pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model in sequence; the preprocessing layer preprocesses the second data to obtain second detection data; the feature extraction layer extracts features from the second detection data to obtain current features, resistance features, temperature features and humidity features; the feature analysis layer analyzes these three features through a linear regression model to obtain a first power loss function.

[0013] Preferably, the first data includes historical electricity consumption data, socio-economic data and weather data; the second data includes accumulated power loss data, transmission line data, equipment data and environmental data; the historical electricity consumption data includes daily, weekly and monthly electricity consumption data; the socio-economic data includes holiday, working day and seasonal data; the weather data includes air temperature, humidity and wind speed; the transmission line data includes the length and material properties of the transmission line; the length is the length of the wire from the distribution point to the distribution cabinet of each company; the equipment data includes current data between the distribution point and the distribution cabinet of each company; the environmental data includes temperature and humidity.

[0014] Preferably, the first power demand function is:

[0015] ;

[0016] in, Indicates The electricity demand of the company in the next time window; represents the first weight matrix; An expression representing the time dependence of historical electricity consumption characteristics and electricity demand; Indicates The historical electricity consumption characteristics of the companies in the previous time window; Expressions representing the time-dependency between socio-economic data characteristics and electricity demand; represents the second weight matrix; Indicates The socioeconomic characteristics of the companies in the previous time window; An expression representing the time dependency between weather data characteristics and electricity demand; represents the third weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

[0017] Preferably, the comprehensive power demand matching satisfaction is:

[0018] ;

[0019] in, Indicates the comprehensive electricity demand allocation satisfaction; Indicates The power allocation value of each company in the next time window; Indicates the number of companies; Indicates The electricity demand of the company in the next time window; Indicates The demand matching satisfaction of the companies;

[0020] The said The demand matching satisfaction of the companies is:

[0021] ;

[0022] The constraint conditions include that the sum of the power allocation values ​​allocated by each company is not greater than the preset total power threshold of the park and the power demand matching satisfaction of each company is not less than the preset satisfaction threshold, which is specifically manifested as:

[0023] ;

[0024] in, Indicates the preset total power threshold of the park; represents the preset satisfaction threshold;

[0025] The global allocation scheme is:

[0026] ;

[0027] in, represents the global optimal power allocation scheme; Indicates The electricity distribution value of the company.

[0028] Preferably, the first power loss function is:

[0029] ;

[0030] in, Indicates The power loss of the company in the next time window; represents the first influence matrix; Indicates The current characteristics of the companies in the previous time window; represents the second influence matrix; Indicates The temperature characteristics of the companies in the previous time window; represents the third influence matrix; Indicates Humidity characteristics of the companies in the previous time window; represents the fourth influence matrix; Indicates The resistance characteristics of the companies in the previous time window; represents the second error term.

[0031] A power supply system, comprising:

[0032] A data acquisition module, used to acquire the first data and the second data of the previous time window of each company in the park;

[0033] The power demand forecasting module is used to input the first data into the power demand forecasting model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; the preprocessing layer preprocesses the first data to obtain first detection data; the feature extraction layer extracts features from the first detection data to obtain historical power consumption features, socio-economic features and weather features; the feature analysis layer analyzes the three features through an ARIMA model and a multivariate linear regression model to obtain a first power demand function;

[0034] A power demand matching optimization module, used to obtain a comprehensive power demand allocation satisfaction according to the first power demand function and the initial power allocation value, and optimize the comprehensive power demand allocation satisfaction according to constraints and a particle swarm algorithm to obtain a global optimal power allocation solution;

[0035] The power loss prediction module is used to input the second data into the power loss prediction model, and pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model in sequence; the preprocessing layer preprocesses the second data to obtain second detection data; the feature extraction layer extracts features from the second detection data to obtain current features, resistance features, temperature features and humidity features; the feature analysis layer analyzes these three features through a linear regression model to obtain a first power loss function.

[0036] Preferably, the first power demand function is:

[0037] ;

[0038] in, Indicates The electricity demand of the company in the next time window; represents the first weight matrix; An expression representing the time dependence of historical electricity consumption characteristics and electricity demand; Indicates The historical electricity consumption characteristics of the companies in the previous time window; Expressions representing the time-dependency between socio-economic data characteristics and electricity demand; represents the second weight matrix; Indicates The socioeconomic characteristics of the companies in the previous time window; An expression representing the time dependency between weather data characteristics and electricity demand; represents the third weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

[0039] Weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

[0040] Preferably, the comprehensive power demand matching satisfaction is:

[0041] ;

[0042] in, Indicates the comprehensive electricity demand allocation satisfaction; Indicates The power allocation value of each company in the next time window; Indicates The electricity demand of the company in the next time window; Indicates the number of companies; Indicates The demand matching satisfaction of the companies;

[0043] The said The demand matching satisfaction of the companies is:

[0044] ;

[0045] The constraint conditions include that the sum of the power allocation values ​​allocated by each company is not greater than the preset total power threshold of the park and the power demand matching satisfaction of each company is not less than the preset satisfaction threshold, which is specifically manifested as:

[0046] ;

[0047] in, Indicates the preset total power threshold of the park; represents the preset satisfaction threshold;

[0048] The global allocation scheme is:

[0049] ;

[0050] in, represents the global optimal power allocation scheme; Indicates The electricity distribution value of the company.

[0051] Preferably, the first power loss function is:

[0052] ;

[0053] in, Indicates The power loss of the company in the next time window; represents the first influence matrix; Indicates The current characteristics of the companies in the previous time window; represents the second influence matrix; Indicates The temperature characteristics of the companies in the previous time window; represents the third influence matrix; Indicates Humidity characteristics of the companies in the previous time window; represents the fourth influence matrix; Indicates The resistance characteristics of the companies in the previous time window; represents the second error term.

[0054] A power supply center comprises a load and a power supply system, wherein the power supply system is used to supply power to the load.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The present invention obtains historical electricity consumption data, socio-economic data and weather data of each company in the park and inputs them into the power demand forecasting model. The feature extraction layer extracts historical electricity consumption features, socio-economic features and weather features, and the feature analysis layer analyzes these features through the ARIMA model and the multivariate linear regression model to establish the time dependency relationship between the features and the power demand. Based on the time dependency relationship, a first power demand function is generated, which can accurately predict the future power demand of each company in the park, thereby effectively avoiding oversupply or shortage of electricity and improving the operational stability and efficiency of the power supply system.

[0057] 2. The present invention analyzes the first power demand function, calculates the comprehensive power demand allocation satisfaction, and optimizes it through the particle swarm algorithm. On the premise of meeting the basic power demand of each company, the limited power resources are reasonably allocated to ensure the maximization of the comprehensive power demand allocation satisfaction, thereby improving the operation stability and efficiency of the power supply system.

[0058] 3. The present invention obtains the accumulated power loss data, transmission line data, equipment data and environmental data of each company in the park, and inputs them into the power loss prediction model, extracts the current, resistance, temperature and humidity characteristics, and analyzes them through a linear regression model in combination with the characteristic analysis layer to generate a power loss prediction function. This function can accurately predict the power loss during the transmission process, help to identify potential high-loss areas in a timely manner, and provide an optimization basis so as to adjust strategies in the future, reduce energy waste, and improve the operational stability and efficiency of the power supply system. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A schematic diagram of a power supply device provided in an embodiment of the present invention;

[0060] Figure 2 A schematic diagram of a power supply system provided in an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of a power demand prediction model provided by an embodiment of the present invention;

[0062] Figure 4 A schematic diagram of a power loss prediction model provided by an embodiment of the present invention;

[0063] Figure 5 A schematic diagram of a data center provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] Embodiment 1

[0066] Park A is a small Internet park, which currently has some Internet companies settled in. In order to improve the operating stability and efficiency of the power supply system, Park A has applied a power supply device, such as Figure 1 As shown;

[0067] A power supply device, comprising:

[0068] A data acquisition unit, used to acquire the first data and the second data of the previous time window of each company in the park;

[0069] In this embodiment, the previous time window is represented as one year;

[0070] The power demand forecasting unit is used to input the first data into the power demand forecasting model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; further, the first data includes historical power consumption data, socio-economic data and weather data; the socio-economic data includes holidays, working days and seasonal data; the weather data includes temperature, humidity and wind speed; the schematic diagram of the power demand forecasting model is as shown in Figure 3 As shown;

[0071] The preprocessing layer preprocesses the first data to obtain first detection data; the preprocessing includes data cleaning and normalization;

[0072] The feature extraction layer extracts features from the first detection data to obtain historical electricity consumption features, socio-economic features, and weather features;

[0073] The historical electricity consumption characteristics include the long-term trend, seasonal fluctuations and periodic characteristics of electricity consumption data, which reflect the changing pattern of electricity consumption data over time. The specific extraction process is as follows: first, the moving average is calculated to extract the long-term trend in the historical electricity consumption data to identify the overall trend of electricity consumption and smooth out short-term fluctuations; the seasonal-trend decomposition (STL) technology is used to decompose the historical electricity consumption data into trend, seasonal and residual parts, and seasonal characteristics reflecting annual or quarterly fluctuations are extracted from them; according to the autocorrelation function (ACF), the periodic characteristics in the historical electricity consumption data are identified, such as daily, weekly and monthly electricity consumption patterns.

[0074] The socio-economic characteristics include the impact characteristics of holidays and working days reflected by time tags, as well as the periodic characteristics of economic activities after aggregation, which capture the impact of socio-economic factors on electricity demand. The extraction process includes: adding time tags such as holidays and working days to the data, and extracting features reflecting the impact of socio-economic activities by analyzing the impact of these tags on electricity consumption; aggregating socio-economic data by week to extract the periodic impact characteristics of socio-economic activities on electricity demand, such as the difference in electricity consumption between working days and weekends.

[0075] The weather characteristics include average temperature and humidity characteristics, which reflect the direct impact of weather changes on electricity demand. The extraction process includes: extracting the daily average temperature and humidity, and extracting the characteristics reflecting the impact of weather conditions on electricity demand by analyzing the impact of these meteorological data on electricity demand.

[0076] Furthermore, the feature analysis layer uses the ARIMA model to analyze the historical electricity consumption characteristics to obtain the time dependence relationship between the historical electricity consumption characteristics and the electricity demand; and uses the multivariate linear regression model to analyze the socio-economic characteristics and weather characteristics respectively to obtain the time dependence relationship between the socio-economic characteristics and weather characteristics and the electricity demand;

[0077] Furthermore, a first power demand function is obtained according to the dependency relationship; the first power demand function is:

[0078] ;

[0079] in, Indicates The electricity demand of the company in the next time window; represents the first weight matrix; An expression representing the time dependence of historical electricity consumption characteristics and electricity demand; Indicates The historical electricity consumption characteristics of the companies in the previous time window; Expressions representing the time-dependency between socio-economic data characteristics and electricity demand; represents the second weight matrix; Indicates The socioeconomic characteristics of the companies in the previous time window; An expression representing the time dependency between weather data characteristics and electricity demand; represents the third weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

[0080] In this embodiment, the power demand of some companies in Park A is shown in Table 1;

[0081] Table 1 Forecasted electricity demand of some companies

[0082] company Forecast electricity demand (GWh) 1 110 2 85 3 70 4 125

[0083] A power demand matching optimization unit, configured to obtain a comprehensive power demand allocation satisfaction degree according to the first power demand function and the initial power allocation value, and optimize the comprehensive power demand allocation satisfaction degree according to constraint conditions and a particle swarm algorithm to obtain a global optimal power allocation solution;

[0084] This embodiment obtains historical electricity consumption data, socio-economic data, and weather data of each company in the park and inputs them into the power demand forecasting model. The feature extraction layer extracts historical electricity consumption features, socio-economic features, and weather features. The feature analysis layer analyzes these features through the ARIMA model and the multivariate linear regression model to establish a time-dependent relationship between the features and power demand. Based on the time-dependent relationship, a first power demand function is generated, which can accurately predict the future power demand of each company in the park, thereby effectively avoiding oversupply or shortage of electricity and improving the operational stability and efficiency of the power supply system.

[0085] In this embodiment, the optimization variable is the allocated electricity of each company in Park A for the next year;

[0086] Furthermore, the comprehensive power demand matching satisfaction is:

[0087] ;

[0088] in, Indicates the comprehensive electricity demand allocation satisfaction; Indicates The power allocation value of each company in the next time window; Indicates The electricity demand of the company in the next time window; Indicates the number of companies; Indicates The demand matching satisfaction of the companies;

[0089] The said The demand matching satisfaction of the companies is:

[0090] ;

[0091] The constraint conditions include that the sum of the power allocation values ​​allocated by each company is not greater than the preset total power threshold of the park and the power demand matching satisfaction of each company is not less than the preset satisfaction threshold, which is specifically manifested as:

[0092] ;

[0093] in, Indicates the preset total power threshold of the park; represents the preset satisfaction threshold;

[0094] In this embodiment, the power value allocated to each company in Park A is taken as the optimization target, and the optimization objective function is the inverse of the comprehensive power demand allocation satisfaction, where the objective function is:

[0095] ;

[0096] in, represents the optimization objective function;

[0097] Combined with the constraints, 100 iterations are set and 100 populations are initialized with dimensions of , where n is the number of companies, and the position of each particle represents the power distribution value of each company; calculate the fitness value of each particle, and continuously update the speed and position of the particle; repeat the evaluation and update until the stop condition is met; the stop condition is that the maximum number of iterations is reached; and output the global allocation plan;

[0098] The global allocation scheme is:

[0099] ;

[0100] in, represents the global optimal power allocation scheme; Indicates The electricity distribution value of each company;

[0101] The power loss prediction unit is used to input the second data into the power loss prediction model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; the preprocessing layer preprocesses the second data to obtain second detection data; the feature extraction layer extracts features from the second detection data to obtain current features, resistance features, temperature features and humidity features; the schematic diagram of the power loss prediction model is as shown in Figure 4 As shown;

[0102] This embodiment analyzes the first power demand function, calculates the comprehensive power demand allocation satisfaction, and optimizes it through the particle swarm algorithm. On the premise of meeting the basic power demand of each company, the limited power resources are reasonably allocated to ensure the maximization of the comprehensive power demand allocation satisfaction, thereby improving the operation stability and efficiency of the power supply system.

[0103] The second data includes power accumulated loss data, transmission line data, equipment data and environmental data; the transmission line data includes the length and material properties of the transmission line; the length is the length of the line from the distribution point to the distribution cabinet of each company; the equipment data includes the current data between the distribution point and the distribution cabinet of each company; the environmental data includes temperature and humidity;

[0104] The temperature characteristics include the daily maximum temperature, daily minimum temperature, daily average temperature, weekly average temperature, and monthly average temperature; the current characteristics; the humidity characteristics include the daily maximum humidity, daily minimum humidity, daily average humidity, weekly average humidity, and monthly average humidity; the current characteristics include the daily peak current, monthly average current, monthly current fluctuation amplitude, and monthly current standard deviation; the resistance characteristics include length characteristics, material characteristics, and temperature influence characteristics; the temperature influence characteristics are expressed as the influence of temperature on resistance;

[0105] The feature analysis layer uses the above extracted features as independent variables, analyzes their influence on power loss through a linear regression model, and obtains a first power loss function.

[0106] Furthermore, the first power loss function is:

[0107] ;

[0108] in, Indicates The power loss of the company in the next time window; represents the first influence matrix; Indicates The current characteristics of the companies in the previous time window; represents the second influence matrix; Indicates The temperature characteristics of the companies in the previous time window; represents the third influence matrix; Indicates Humidity characteristics of the companies in the previous time window; represents the fourth influence matrix; Indicates The resistance characteristics of the companies in the previous time window; represents the second error term.

[0109] Embodiment 2

[0110] B Innovation Internet Park is a medium-sized Internet enterprise. Currently, a large number of Internet companies have settled in. In order to improve the operating stability and efficiency of the power supply system, B Innovation Internet Park has applied a power supply system such as Figure 2 As shown;

[0111] A power supply system, comprising:

[0112] A data acquisition module, used to acquire the first data and the second data of the previous time window of each company in the park;

[0113] In this embodiment, the previous time window is represented as one year;

[0114] The power demand forecasting module is used to input the first data into the power demand forecasting model, and pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model in sequence; further, the first data includes historical power consumption data, socio-economic data and weather data; the socio-economic data includes holidays, working days and seasonal data; the weather data includes temperature, humidity and wind speed; the schematic diagram of the power demand forecasting model is as shown in Figure 3 As shown;

[0115] The preprocessing layer preprocesses the first data to obtain first detection data; the preprocessing includes data cleaning and normalization;

[0116] The feature extraction layer extracts features from the first detection data to obtain historical electricity consumption features, socio-economic features, and weather features;

[0117] The historical electricity consumption characteristics include the long-term trend, seasonal fluctuations and periodic characteristics of electricity consumption data, which reflect the changing pattern of electricity consumption data over time. The specific extraction process is as follows: first, the moving average is calculated to extract the long-term trend in the historical electricity consumption data to identify the overall trend of electricity consumption and smooth out short-term fluctuations; the seasonal-trend decomposition (STL) technology is used to decompose the historical electricity consumption data into trend, seasonal and residual parts, and seasonal characteristics reflecting annual or quarterly fluctuations are extracted from them; according to the autocorrelation function (ACF), the periodic characteristics in the historical electricity consumption data are identified, such as daily, weekly and monthly electricity consumption patterns.

[0118] The socio-economic characteristics include the impact characteristics of holidays and working days reflected by time tags, as well as the periodic characteristics of economic activities after aggregation, which capture the impact of socio-economic factors on electricity demand. The extraction process includes: adding time tags such as holidays and working days to the data, and extracting features reflecting the impact of socio-economic activities by analyzing the impact of these tags on electricity consumption; aggregating socio-economic data by week to extract the periodic impact characteristics of socio-economic activities on electricity demand, such as the difference in electricity consumption between working days and weekends.

[0119] The weather characteristics include average temperature and humidity characteristics, which reflect the direct impact of weather changes on electricity demand. The extraction process includes: obtaining the daily average temperature and humidity, analyzing the impact of these meteorological data on electricity demand, and extracting the characteristics reflecting the impact of weather conditions on electricity demand.

[0120] Furthermore, the feature analysis layer uses the ARIMA model to analyze the historical electricity consumption characteristics to obtain the time dependence relationship between the historical electricity consumption characteristics and the electricity demand; and uses the multivariate linear regression model to analyze the socio-economic characteristics and weather characteristics respectively to obtain the time dependence relationship between the socio-economic characteristics and weather characteristics and the electricity demand;

[0121] Furthermore, a first power demand function is obtained according to the dependency relationship; the first power demand function is:

[0122] ;

[0123] in, Indicates The electricity demand of the company in the next time window; represents the first weight matrix; An expression representing the time dependence of historical electricity consumption characteristics and electricity demand; Indicates The historical electricity consumption characteristics of the companies in the previous time window; Expressions representing the time-dependency between socio-economic data characteristics and electricity demand; represents the second weight matrix; Indicates The socioeconomic characteristics of the companies in the previous time window; An expression representing the time dependency between weather data characteristics and electricity demand; represents the third weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

[0124] In this example, the power demand of some companies in Park B is shown in Table 2;

[0125] Table 2 Forecast of electricity demand of some companies

[0126] company Forecast electricity demand (GWh) 1 120 2 80 3 90 4 130

[0127] A power demand matching optimization module, used to obtain a comprehensive power demand allocation satisfaction according to the first power demand function and the initial power allocation value, and optimize the comprehensive power demand allocation satisfaction according to constraints and a particle swarm algorithm to obtain a global optimal power allocation solution;

[0128] In this embodiment, the optimization variable is the allocated power of each company in the B Internet Innovation Park in the next year; among them, the B Internet Innovation Park has 50 Internet companies; the power demand table of some Internet companies in the B Park is shown in Table 2;

[0129] Furthermore, the comprehensive power demand matching satisfaction is:

[0130] ;

[0131] in, Indicates the comprehensive electricity demand allocation satisfaction; Indicates The power allocation value of each company in the next time window; Indicates The electricity demand of the company in the next time window; Indicates the number of companies; Indicates The demand matching satisfaction of the companies;

[0132] The said The demand matching satisfaction of the companies is:

[0133] ;

[0134] The constraint conditions include that the sum of the power allocation values ​​allocated by each company is not greater than the preset total power threshold of the park and the power demand matching satisfaction of each company is not less than the preset satisfaction threshold, which is specifically manifested as:

[0135] ;

[0136] in, Indicates the preset total power threshold of the park; represents the preset satisfaction threshold;

[0137] The global allocation scheme is:

[0138] ;

[0139] in, represents the global optimal power allocation scheme; Indicates The electricity distribution value of each company;

[0140] The power loss prediction module is used to input the second data into the power loss prediction model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; the preprocessing layer preprocesses the second data to obtain second detection data; the feature extraction layer extracts features from the second detection data to obtain current features, resistance features, temperature features and humidity features; the schematic diagram of the power loss prediction model is as shown in Figure 4 As shown;

[0141] The second data includes power accumulated loss data, transmission line data, equipment data and environmental data; the transmission line data includes the length and material properties of the transmission line; the length is the length of the line from the distribution point to the distribution cabinet of each company; the equipment data includes the current data between the distribution point and the distribution cabinet of each company; the environmental data includes temperature and humidity;

[0142] The temperature characteristics include the daily maximum temperature, daily minimum temperature, daily average temperature, weekly average temperature, and monthly average temperature; the current characteristics; the humidity characteristics include the daily maximum humidity, daily minimum humidity, daily average humidity, weekly average humidity, and monthly average humidity; the current characteristics include the daily peak current, monthly average current, monthly current fluctuation amplitude, and monthly current standard deviation; the resistance characteristics include length characteristics, material characteristics, and temperature influence characteristics; the temperature influence characteristics are expressed as the influence of temperature on resistance;

[0143] The feature analysis layer uses the above extracted features as independent variables, analyzes their influence on power loss through a linear regression model, and obtains a first power loss function.

[0144] Furthermore, the first power loss function is:

[0145] ;

[0146] in, Indicates The power loss of the company in the next time window; represents the first influence matrix; Indicates The current characteristics of the companies in the previous time window; represents the second influence matrix; Indicates The temperature characteristics of the companies in the previous time window; represents the third influence matrix; Indicates Humidity characteristics of the companies in the previous time window; represents the fourth influence matrix; Indicates The resistance characteristics of the companies in the previous time window; represents the second error term.

[0147] This embodiment obtains the accumulated power loss data, transmission line data, equipment data and environmental data of each company in the park, and inputs them into the power loss prediction model, extracts the current, resistance, temperature and humidity characteristics, and analyzes them through a linear regression model in combination with the feature analysis layer to generate a power loss prediction function. This function can accurately predict the power loss during the transmission process, help to identify potential high-loss areas in a timely manner, and provide an optimization basis so that strategies can be adjusted in the future to reduce energy waste and improve the operational stability and efficiency of the power supply system.

[0148] Furthermore, the present embodiment includes a power supply center, such as Figure 5 As shown; the power supply center includes a load, and the power supply system, wherein the power supply system is used to supply power to the load.

[0149] This embodiment obtains the first data and the second data of the previous time window of each company in the park, and inputs them into the power demand prediction model and the power loss prediction model respectively. The power demand prediction model preprocesses the first data through the preprocessing layer, the feature extraction layer extracts historical electricity consumption, socio-economic and weather characteristics, and the feature analysis layer uses the ARIMA model and the multivariate linear regression model to generate the first power demand function. Combined with the initial power allocation value and constraints, the particle swarm algorithm is used to optimize the comprehensive power demand allocation satisfaction and obtain the global optimal power allocation plan. The power loss prediction model preprocesses the second data, extracts current, resistance, temperature and humidity characteristics, and generates the first power loss function through the linear regression model. The invention improves the overall efficiency of the power supply system by optimizing power distribution and loss prediction.

[0150] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A power supply device, characterized in that: The device comprises: A data acquisition unit, used to acquire the first data and the second data of the previous time window of each company in the park; The power demand prediction unit is used to input the first data into the power demand prediction model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; the preprocessing layer preprocesses the first data to obtain first detection data; the feature extraction layer extracts features from the first detection data to obtain historical power consumption features, socio-economic features and weather features; the feature analysis layer analyzes the three features through an ARIMA model and a multivariate linear regression model to obtain a first power demand function; The power demand matching optimization unit is used to obtain the comprehensive power demand allocation satisfaction according to the first power demand function and the initial power allocation value, and optimize the comprehensive power demand allocation satisfaction according to the constraint conditions and the particle swarm algorithm to obtain the global optimal power allocation plan; the comprehensive power demand allocation satisfaction is: ； in, Indicates the comprehensive electricity demand allocation satisfaction; Indicates The power allocation value of each company in the next time window; Indicates The electricity demand of the company in the next time window; Indicates the number of companies; Indicates Demand distribution satisfaction of companies; The said The satisfaction of demand distribution of companies is: ； The constraint conditions include that the sum of the power allocation values ​​allocated by each company is not greater than the preset total power threshold of the park and the power demand allocation satisfaction of each company is not less than the preset satisfaction threshold, which is specifically manifested as: ； in, Indicates the preset total power threshold of the park; represents the preset satisfaction threshold; The global optimal power allocation scheme is: ； in, represents the global optimal power allocation scheme; Indicates The electricity distribution value of each company; The power loss prediction unit is used to input the second data into the power loss prediction model, and pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model in sequence; the preprocessing layer preprocesses the second data to obtain second detection data; the feature extraction layer extracts features from the second detection data to obtain current features, resistance features, temperature features and humidity features; the feature analysis layer analyzes these three features through a linear regression model to obtain a first power loss function.

2. A power supply device according to claim 1, characterized in that: The first data includes historical electricity consumption data, socio-economic data and weather data; the second data includes accumulated power loss data, transmission line data, equipment data and environmental data; the historical electricity consumption data includes daily, weekly and monthly electricity consumption data; the socio-economic data includes holiday, working day and seasonal data; the weather data includes temperature, humidity and wind speed; the transmission line data includes the length and material properties of the transmission line; the length is the length of the wire from the distribution point to the distribution cabinet of each company; the equipment data includes current data between the distribution point and the distribution cabinet of each company; the environmental data includes temperature and humidity.

3. A power supply device according to claim 1, characterized in that: The first power demand function is: ； in, Indicates The electricity demand of the company in the next time window; represents the first weight matrix; An expression representing the time dependence of historical electricity consumption characteristics and electricity demand; Indicates The historical electricity consumption characteristics of the companies in the previous time window; Expressions representing the time-dependency between socio-economic data characteristics and electricity demand; represents the second weight matrix; Indicates The socioeconomic characteristics of the companies in the previous time window; An expression representing the time dependency between weather data characteristics and electricity demand; represents the third weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

4. A power supply device according to claim 1, characterized in that: The first power loss function is: ； in, Indicates The power loss of the company in the next time window; represents the first influence matrix; Indicates The current characteristics of the companies in the previous time window; represents the second influence matrix; Indicates The temperature characteristics of the companies in the previous time window; represents the third influence matrix; Indicates Humidity characteristics of the companies in the previous time window; represents the fourth influence matrix; Indicates The resistance characteristics of the companies in the previous time window; represents the second error term.

5. A power supply system, characterized in that: The system comprises: A data acquisition module, used to acquire the first data and the second data of the previous time window of each company in the park; The power demand forecasting module is used to input the first data into the power demand forecasting model, and sequentially pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model; the preprocessing layer preprocesses the first data to obtain first detection data; the feature extraction layer extracts features from the first detection data to obtain historical power consumption features, socio-economic features and weather features; the feature analysis layer analyzes the three features through an ARIMA model and a multivariate linear regression model to obtain a first power demand function; The power demand matching optimization module is used to obtain the comprehensive power demand allocation satisfaction according to the first power demand function and the initial power allocation value, and optimize the comprehensive power demand allocation satisfaction according to the constraint conditions and the particle swarm algorithm to obtain the global optimal power allocation plan; the comprehensive power demand allocation satisfaction is: ； in, Indicates the comprehensive electricity demand allocation satisfaction; Indicates The power allocation value of each company in the next time window; Indicates The electricity demand of the company in the next time window; Indicates the number of companies; Indicates Demand distribution satisfaction of companies; The said The satisfaction of demand distribution of companies is: ； The constraint conditions include that the sum of the power allocation values ​​allocated by each company is not greater than the preset total power threshold of the park and the power demand allocation satisfaction of each company is not less than the preset satisfaction threshold, which is specifically manifested as: ； in, Indicates the preset total power threshold of the park; represents the preset satisfaction threshold; The global optimal power allocation scheme is: ； in, represents the global optimal power allocation scheme; Indicates The electricity distribution value of each company; The power loss prediction module is used to input the second data into the power loss prediction model, and pass through the preprocessing layer, feature extraction layer and feature analysis layer of the model in sequence; the preprocessing layer preprocesses the second data to obtain second detection data; the feature extraction layer extracts features from the second detection data to obtain current features, resistance features, temperature features and humidity features; the feature analysis layer analyzes these three features through a linear regression model to obtain a first power loss function.

6. A power supply system according to claim 5, characterized in that: The first power demand function is: ； in, Indicates The electricity demand of the company in the next time window; represents the first weight matrix; An expression representing the time dependence of historical electricity consumption characteristics and electricity demand; Indicates The historical electricity consumption characteristics of the companies in the previous time window; Expressions representing the time-dependency between socio-economic data characteristics and electricity demand; represents the second weight matrix; Indicates The socioeconomic characteristics of the companies in the previous time window; An expression representing the time dependency between weather data characteristics and electricity demand; represents the third weight matrix; Indicates The weather data characteristics of the companies in the previous time window; represents the first error term.

7. A power supply system according to claim 5, characterized in that: The first power loss function is: ； in, Indicates The power loss of the company in the next time window; represents the first influence matrix; Indicates The current characteristics of the companies in the previous time window; represents the second influence matrix; Indicates The temperature characteristics of the companies in the previous time window; represents the third influence matrix; Indicates Humidity characteristics of the companies in the previous time window; represents the fourth influence matrix; Indicates The resistance characteristics of the companies in the previous time window; represents the second error term.

8. A power supply center, characterized in that: It comprises a load, and a power supply system as described in any one of claims 5 to 7, wherein the power supply system is used to supply power to the load.

Images