Power supply energy consumption optimization analysis method for smart stadium

Abstract

The invention discloses a power supply energy consumption optimization analysis method for a smart stadium, and relates to the technical field of energy optimization analysis, and the method comprises the following steps: determining electric equipment, and collecting historical power consumption data and power consumption of the electric equipment; determining an equipment power consumption influence factor which is an independent variable and the equipment power consumption as a dependent variable to construct a regression model; constructing a time sequence model to predict future influence factors and predict the power consumption demand of the equipment; quantifying the error of the regression model and the prediction uncertainty of the time sequence model; analyzing the sensitivity degree of the regression model to the change of the influence factors, and transmitting the uncertainty of the influence factors to the uncertainty of equipment power consumption prediction; and combining the inherent error of the regression model and the uncertainty of the input influence factors, and configuring a standby power supply. The method can effectively improve the situation that resources are wasted due to too large capacity or the emergency power utilization requirement cannot be met due to too small capacity due to lack of scientific basis in standby power supply capacity configuration in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of energy optimization analysis, and in particular to a method for optimizing the analysis of power supply energy consumption in a smart sports stadium. Background Art

[0002] In the field of power supply for smart sports stadiums, a variety of technical means are currently used to maintain power supply and management. Traditional power monitoring relies on smart meters to collect power consumption data, which can obtain power consumption information for the entire venue and some key equipment. However, its collection frequency is relatively low, usually in hours or even days. In terms of power demand forecasting, the simple moving average method or exponential smoothing method in time series analysis is often used to estimate future power consumption trends based on power consumption data over a period of time. In terms of power supply security, backup diesel generators or uninterruptible power supplies (UPS) are generally equipped to provide emergency power when problems occur in the main power supply line.

[0003] However, existing technologies have exposed many shortcomings in practical applications. When faced with partial line failures, traditional backup power systems lack accurate fault location and affected equipment identification mechanisms, and are unable to restore power to affected equipment in the first place, resulting in long-term power outages in some areas of the venue, seriously disrupting the normal development of the event or affecting the audience's viewing experience. At the same time, the configuration of backup power capacity lacks scientific basis, and is mostly set based on experience or the approximate size of the venue. It is very easy for the capacity to be too large, resulting in idle equipment, waste of resources, and rising operation and maintenance costs, or the capacity is too small to meet emergency power needs. Summary of the invention

[0004] The purpose of the present invention is to provide a method for optimizing the power supply energy consumption of a smart sports stadium to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for optimizing power consumption analysis of a smart sports stadium, the method comprising the following steps: Step 1: Identify the electrical equipment and collect historical power consumption data and power consumption of each electrical equipment at fixed time intervals; Step 2: Determine the influencing factors of power consumption of each power-consuming device, and build a regression model with the influencing factors as independent variables and the power consumption of each power-consuming device as the dependent variable; Step 3: Predict the influencing factors at future times based on known influencing factor data or by building a time series model, and further predict the power demand of each power-consuming device; Step 4: quantify the historical prediction error of the regression model of each power-consuming equipment and measure the prediction uncertainty of the time series model of each influencing factor; Step 5: Analyze the sensitivity of the regression model of each electrical equipment to the changes of each influencing factor, and transfer the uncertainty of the influencing factor to the uncertainty of the power consumption prediction of the electrical equipment; Step 6: Combine the inherent error of the regression model and the uncertainty of the input influencing factors to form an overall prediction uncertainty index, and configure the backup power supply accordingly.

[0006] In step 1, the devices connected to the circuit are recorded, clustered according to their functions, and the power-consuming devices are determined, which are expressed as: [a1, a2, ..., a n ]; where n is a positive integer, indicating the number of types of electrical equipment; a1~a n Respectively represent the 1st to nth types of electrical equipment; The data collection time is expressed as: [T0, T1, T2, …, T m ]; where m is a positive integer, indicating the number of data collection moments, T0~T m They represent the 1st to m+1th data collection moments respectively; the time intervals between adjacent moments are the same.

[0007] Obtain historical electricity consumption data and power consumption of each electrical device based on smart meters and IoT sensors; The historical electricity consumption data is expressed as: [E(T 1-0 ),E(T 2-1 ),…,E(T m-(m-1) )]; where T 1-0 ~T m-(m-1) Respectively represent the time interval between T0 and T1 and the time interval between Tm-1 and Tm; E(T 1-0 )~E(T m-(m-1) ) respectively represent the power consumption data in the time interval from T0 to T1 ~ the power consumption data in the time interval from Tm-1 to Tm; The power consumption of each electrical equipment is expressed as: ; Among them, A[a1,T 1-0 ]~A[a n ,T 1-0 ] respectively represent the electrical equipment a1~a1 in the time interval from T0 to T1 n Power consumption; A[a1,T m-(m-1) ]~A[a n ,T m-(m-1) ] represent T m-1 to T m Electrical equipment a1~a within the time interval n Power consumption; In each time interval, the sum of the power consumption of each electrical device is equal to the power consumption data.

[0008] In step 2, sensors are deployed to collect influencing factor data, including environmental parameters, venue status parameters, equipment status parameters, and personnel flow data; the influencing factors are expressed as: [I1, I2, …, I u ]; where u is a positive integer, indicating the number of influencing factors, I1~I u Respectively represent the 1st to uth influencing factors; Align the influencing factor data with the device power consumption data at the same time interval; Preprocessing: normalize the same influencing factor data at different time intervals and the power consumption of electrical equipment; For a certain electrical equipment, the influencing factor data corresponding to different time intervals after preprocessing and the power consumption of the electrical equipment are used as input to construct an LSTM neural network regression model to obtain a regression model for predicting the power consumption of the electrical equipment; Build a regression model for power consumption prediction for all power-consuming devices.

[0009] In step 3, the known influencing factor data are obtained based on the meteorological agency or venue activity arrangement; for the unknown influencing factor data, a Prophet time series model is constructed for prediction: the influencing factor data of the historical time interval is used as input, and the Prophet time series model is constructed to output the future time interval T (m+1)-m The predicted value of the influencing factors; among them, T (m+1)-m Indicates T m to T m+1 the time interval between Based on the acquired data of known influencing factors and the predicted data of unknown influencing factors, combined with the regression model for predicting the power consumption of each electrical device, the predicted power consumption of each electrical device is obtained.

[0010] In step 4, the historical prediction error of the regression model of each power-consuming device is quantified: in the training phase, when building the LSTM prediction model for each power-consuming device, the residual between the predicted value and the true value is calculated using historical data; For electrical equipment i , the historical residual standard deviation is ΔA i ; Where i is a positive integer, representing the type sequence of electrical equipment, i∈{1,2,…,n}; When using the Prophet model to predict unknown influencing factors, obtain the uncertainty index of each influencing factor; For each influencing factor I k , the Prophet model outputs a confidence interval or standard deviation when predicting, and records the future time interval T (m+1)-m The prediction standard deviation is σ Ik,T(m+1)-m ; Among them, k is a positive integer, representing the type sequence of influencing factors, k∈{1,2,…,u}; σ Ik,T(m+1)-m Indicates the influencing factor I of the Prophet model prediction k At time interval T (m+1)-m The standard deviation of .

[0011] In step 5, the sensitivity of the LSTM model of each power-consuming device to the changes of each influencing factor is analyzed, so as to transfer the uncertainty of the influencing factor to the uncertainty of the power consumption prediction of the power-consuming device; For each device a i And each influencing factor I k , by making small perturbations on the model input and observing the changes in predicted power consumption, we can calculate the approximate local partial derivatives: ∂A i / ∂I k ≈(A i (I k +δ)-Ai(I k -δ)) / 2δ; Among them, A i Indicates that the influencing factor is I k Electrical equipment i The predicted power consumption; δ represents the numerical disturbance; A i (I k +δ) and A i (I k -δ) respectively represent the influencing factors I k +δ and I k -δ when electrical equipment a i The predicted power consumption; ∂A i / ∂I k Indicates electrical equipment i Influencing Factors I k sensitivity; This sensitivity coefficient is used to measure the contribution of the uncertainty of each influencing factor to the power consumption forecast, that is, the uncertainty propagation of a single influencing factor: V Ik,T(m+1)-m =|∂A i / ∂I k |⋅σ Ik,T(m+1)-m ; Among them, V Ik,T(m+1)-m Influencing factors I k At time interval T (m+1)-m The uncertainty of AI power consumption is propagated to the electrical equipment.

[0012] In step 6, the model inherent error and input uncertainty are combined to obtain the overall prediction uncertainty: For electrical equipment i At time interval T (m+1)-mBy combining the uncertainty propagation of each influencing factor, the input-related uncertainty is obtained: V input,I,T(m+1)-m =(V I1,T(m+1)-m 2 +V I2,T(m+1)-m 2 +…+V Iu,T(m+1)-m 2 ) 1 / 2 ; Among them, V input,I,T(m+1)-m Indicates that all influencing factors are within the time interval T (m+1)-m Propagate to the electrical equipment i Uncertainty in power consumption; Combining this input uncertainty with the model inherent residual uncertainty yields the total uncertainty: V total,I,T(m+1)-m =(ΔA i 2 +V input,I,T(m+1)-m 2 ) 1 / 2 ; Among them, V total,I,T(m+1)-m Indicates electrical equipment i At time interval T (m+1)-m The overall uncertainty within.

[0013] Use the predicted values ​​and uncertainties of each power-consuming device in the future time interval to guide the configuration of backup power supply; For electrical equipment i , the predicted power consumption is A i , the configured spare capacity is calculated according to the formula: E=A i +k·V total,I,T(m+1)-m ; Where k is the safety factor, k>1, determined according to risk tolerance, equipment importance and system redundancy requirements; E represents the power equipment ai in the time interval T (m+1)-m Built-in backup power capacity.

[0014] According to the corresponding backup power capacity calculated for each power-consuming device, mobile backup power of corresponding capacity is configured in advance; the backup power is deployed at key nodes or stored in a centralized location that is easy to dispatch; The operating status of each circuit is detected in real time. When a part of the circuit fails, the dispatch system will quickly identify the affected equipment and automatically instruct the corresponding mobile backup power supply to connect to the power supply to ensure the continuous operation of the power-consuming equipment.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention comprehensively considers multi-dimensional influencing factors such as environmental parameters, venue status, equipment status and personnel flow, establishes an LSTM regression model, captures nonlinear energy consumption laws, and combines the Prophet model's predictive ability for non-known influencing factors to achieve refined energy consumption forecasts for future time periods, providing a scientific basis for energy scheduling; the present invention transfers the uncertainty of influencing factors to the energy consumption forecast results through sensitivity coefficients, quantifies the error contribution of input data and the model itself, and forms an overall forecast uncertainty index; the present invention dynamically calculates the backup power capacity based on uncertainty indicators and safety factors, avoids redundancy or insufficiency caused by traditional "one-size-fits-all" configuration, and reduces costs while ensuring power supply reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of the steps of a method for optimizing power consumption analysis of a smart sports stadium according to the present invention. DETAILED DESCRIPTION

[0017] 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.

[0018] Example: Figure 1 As shown, the present invention provides a technical solution, a method for optimizing power consumption analysis of a smart sports stadium, the method comprising the following steps: Step 1: Identify the electrical equipment and collect historical power consumption data and power consumption of each electrical equipment at fixed time intervals; Step 2: Determine the influencing factors of power consumption of each power-consuming device, and build a regression model with the influencing factors as independent variables and the power consumption of each power-consuming device as the dependent variable; Step 3: Predict the influencing factors at future times based on known influencing factor data or by building a time series model, and further predict the power demand of each power-consuming device; Step 4: quantify the historical prediction error of the regression model of each power-consuming equipment and measure the prediction uncertainty of the time series model of each influencing factor; Step 5: Analyze the sensitivity of the regression model of each electrical equipment to the changes of each influencing factor, and transfer the uncertainty of the influencing factor to the uncertainty of the power consumption prediction of the electrical equipment; Step 6: Combine the inherent error of the regression model and the uncertainty of the input influencing factors to form an overall prediction uncertainty index, and configure the backup power supply accordingly.

[0019] In step 1, the devices connected to the circuit are recorded, clustered according to their functions, and the power-consuming devices are determined, which are expressed as: [a1, a2, ..., a n ]; where n is a positive integer, indicating the number of types of electrical equipment; a1~a n Respectively represent the 1st to nth types of electrical equipment; The data collection time is expressed as: [T0, T1, T2, …, T m ]; where m is a positive integer, indicating the number of data collection moments, T0~T m They represent the 1st to m+1th data collection moments respectively; the time intervals between adjacent moments are the same.

[0020] Obtain historical electricity consumption data and power consumption of each electrical device based on smart meters and IoT sensors; The historical electricity consumption data is expressed as: [E(T 1-0 ),E(T 2-1 ),…,E(T m-(m-1) )]; where T 1-0 ~T m-(m-1) Respectively represent the time interval between T0 and T1 and the time interval between Tm-1 and Tm; E(T 1-0 )~E(T m-(m-1) ) respectively represent the power consumption data in the time interval from T0 to T1 ~ the power consumption data in the time interval from Tm-1 to Tm; The power consumption of each electrical equipment is expressed as: ; Among them, A[a1,T 1-0 ]~A[a n ,T 1-0 ] respectively represent the electrical equipment a1~a1 in the time interval from T0 to T1 n Power consumption; A[a1,T m-(m-1) ]~A[a n ,T m-(m-1) ] represent T m-1 to T m Electrical equipment a1~a within the time interval n Power consumption; In each time interval, the sum of the power consumption of each electrical device is equal to the power consumption data.

[0021] In step 2, sensors are deployed to collect influencing factor data, including environmental parameters, venue status parameters, equipment status parameters, and personnel flow data; the influencing factors are expressed as: [I1, I2, …, I u ]; where u is a positive integer, indicating the number of influencing factors, I1~I u Respectively represent the 1st to uth influencing factors; Align the influencing factor data with the device power consumption data at the same time interval; Preprocessing: normalize the same influencing factor data at different time intervals and the power consumption of electrical equipment; For a certain electrical equipment, the influencing factor data corresponding to different time intervals after preprocessing and the power consumption of the electrical equipment are used as input to construct an LSTM neural network regression model to obtain a regression model for predicting the power consumption of the electrical equipment; Build a regression model for power consumption prediction for all power-consuming devices.

[0022] In step 3, the known influencing factor data are obtained based on the meteorological agency or venue activity arrangement; for the unknown influencing factor data, a Prophet time series model is constructed for prediction: the influencing factor data of the historical time interval is used as input, and the Prophet time series model is constructed to output the future time interval T (m+1)-m The predicted value of the influencing factors; among them, T (m+1)-m Indicates T m to T m+1 the time interval between Based on the acquired data of known influencing factors and the predicted data of unknown influencing factors, combined with the regression model for predicting the power consumption of each electrical device, the predicted power consumption of each electrical device is obtained.

[0023] In step 4, the historical prediction error of the regression model of each power-consuming device is quantified: in the training phase, when building the LSTM prediction model for each power-consuming device, the residual between the predicted value and the true value is calculated using historical data; For electrical equipment i , the historical residual standard deviation is ΔA i ; Where i is a positive integer, representing the type sequence of electrical equipment, i∈{1,2,…,n}; When using the Prophet model to predict unknown influencing factors, obtain the uncertainty index of each influencing factor; For each influencing factor I k , the Prophet model outputs a confidence interval or standard deviation when predicting, and records the future time interval T (m+1)-m The prediction standard deviation is σ Ik,T(m+1)-m ; Among them, k is a positive integer, representing the type sequence of influencing factors, k∈{1,2,…,u}; σ Ik,T(m+1)-m Indicates the influencing factor I of the Prophet model prediction k At time interval T (m+1)-m The standard deviation of .

[0024] In step 5, the sensitivity of the LSTM model of each power-consuming device to the changes of each influencing factor is analyzed, so as to transfer the uncertainty of the influencing factor to the uncertainty of the power consumption prediction of the power-consuming device; For each device a i And each influencing factor I k , by making small perturbations on the model input and observing the changes in predicted power consumption, we can calculate the approximate local partial derivatives: ∂A i / ∂I k ≈(A i (I k +δ)-Ai(I k -δ)) / 2δ; Among them, A i Indicates that the influencing factor is I k Electrical equipment i The predicted power consumption; δ represents the numerical disturbance; A i (I k +δ) and A i (I k -δ) respectively represent the influencing factors I k +δ and I k -δ when electrical equipment a i The predicted power consumption; ∂A i / ∂I k Indicates electrical equipment i Influencing Factors I k sensitivity; This sensitivity coefficient is used to measure the contribution of the uncertainty of each influencing factor to the power consumption forecast, that is, the uncertainty propagation of a single influencing factor: V Ik,T(m+1)-m =|∂A i / ∂I k |⋅σ Ik,T(m+1)-m ; Among them, V Ik,T(m+1)-m Influencing factors I k At time interval T (m+1)-m The uncertainty of AI power consumption is propagated to the electrical equipment.

[0025] In step 6, the model inherent error and input uncertainty are combined to obtain the overall prediction uncertainty: For electrical equipment i At time interval T (m+1)-m By combining the uncertainty propagation of each influencing factor, the input-related uncertainty is obtained: V input,I,T(m+1)-m =(V I1,T(m+1)-m 2 +V I2,T(m+1)-m 2 +…+V Iu,T(m+1)-m 2 )1 / 2 ; Among them, V input,I,T(m+1)-m Indicates that all influencing factors are within the time interval T (m+1)-m Propagate to the electrical equipment i Uncertainty in power consumption; Combining this input uncertainty with the model inherent residual uncertainty yields the total uncertainty: V total,I,T(m+1)-m =(ΔA i 2 +V input,I,T(m+1)-m 2 ) 1 / 2 ; Among them, V total,I,T(m+1)-m Indicates electrical equipment i At time interval T (m+1)-m The overall uncertainty within.

[0026] Use the predicted values ​​and uncertainties of each power-consuming device in the future time interval to guide the configuration of backup power supply; For electrical equipment i , the predicted power consumption is A i , the configured spare capacity is calculated according to the formula: E=A i +k·V total,I,T(m+1)-m ; Where k is the safety factor, k>1, determined according to risk tolerance, equipment importance and system redundancy requirements; E represents the power equipment ai in the time interval T (m+1)-m Built-in backup power capacity.

[0027] According to the corresponding backup power capacity calculated for each power-consuming device, mobile backup power of corresponding capacity is configured in advance; the backup power is deployed at key nodes or stored in a centralized location that is easy to dispatch; The operating status of each circuit is detected in real time. When a part of the circuit fails, the dispatch system will quickly identify the affected equipment and automatically instruct the corresponding mobile backup power supply to connect to the power supply to ensure the continuous operation of the power-consuming equipment.

[0028] In this embodiment, the power supply energy consumption of a smart football field is optimized: Data collection and equipment clustering: Classification of electrical equipment; Divide venue equipment into 5 categories according to function: a1 lighting system (venue lighting, emergency lighting); a2 air conditioning system (audience seats, locker room air conditioning); a3 ventilation system (ventilation equipment); a4 security system (cameras, access control); a5 catering equipment (kitchen appliances, vending machines); Historical data collection: Time interval: record once every hour (for example: 00:00 on May 1, 2025 to 23:00 on May 31); Total power consumption: total power consumption per hour (unit: kWh); Equipment power consumption: Through smart meters and IoT sensors, record the hourly power consumption of each type of equipment (for example, lighting system a1 consumes 1200kWh from 00:00 to 01:00 on May 1); Determination and preprocessing of influencing factors: List of influencing factors: Environmental parameters: temperature, humidity (collected by meteorological sensors); Venue status: event status (0 = no event, 1 = event), number of spectators (ticketing system data); Equipment status: equipment load rate (collected by IoT sensors); Data alignment and normalization: Alignment: Match the influencing factor data with the device power consumption data at the same time interval (every hour); Normalization: Scale data of different dimensions (such as temperature and number of people) to the interval [0,1] (for example: temperature 25℃ → 0.5, number of spectators 48,000 → 0.96); Model building and energy consumption prediction: LSTM regression model training (taking lighting system a1 as an example); Input: Influencing factor data for the past 3 hours (temperature, humidity, event status, number of spectators, equipment load rate); Output: Power consumption of lighting system a1 in the current hour; Training process: Use historical data (such as May 1 to May 31) to train the model and learn the nonlinear relationship between influencing factors and power consumption.

[0029] Prediction of future influencing factors: Known factors: event status (17:00-23:00 is 1), number of spectators (48,000); Prediction of unknown factors: Use the Prophet model to predict the temperature and humidity in the next 6 hours (for example, predict the temperature to be 28°C and the humidity to be 60% at 17:00); Energy consumption prediction results: Input the influencing factor data for the next three hours (sliding window), and use the LSTM model to predict the hourly power consumption of each device (for example: lighting system a1 is predicted to consume 1500kWh from 17:00 to 18:00); Uncertainty quantification: model-inherent errors; Calculate the standard deviation of the historical forecast residuals (ΔA): For example, the standard deviation of the historical forecast error of lighting system a1 is ΔA1=50kWh; Uncertainty of influencing factors: Get the confidence interval of the Prophet model prediction (for example: temperature prediction value 28℃, confidence interval 26-30℃, standard deviation σ=2℃); Sensitivity analysis: Calculate the sensitivity of the device to influencing factors: By perturbing the input data (such as temperature ±1°C), observe the change in power consumption: Formula: Sensitivity coefficient = (predicted value + δ - predicted value - δ) / (2δ); For example: for every 1°C increase in temperature, the power consumption of lighting system a1 increases by 30 kWh (sensitivity coefficient = 30); Uncertainty propagation: Multiply the uncertainty of the influencing factor by the sensitivity coefficient to obtain the impact on power consumption: temperature uncertainty propagation = 30kWh / ℃×2℃=60kWh; Backup power configuration: Total uncertainty calculation: Input uncertainty: the square root of the sum of the squares of all influencing factors’ uncertainties (e.g., the total impact of temperature, humidity, etc., Vinput=70kWh); Total uncertainty: (ΔA²+Vinput²) 1 / 2 =(50²+70²) 1 / 2 ≈86kWh; Spare capacity calculation: Formula: Reserve capacity = predicted value + k × total uncertainty (k = safety factor, take 2); For example: Lighting system a1 is predicted to consume 1500 kWh, and the backup capacity = 1500 + 2 × 86 = 1672 kWh; Dynamic deployment: Deploy the mobile power supply near the lighting system a1 distribution box; monitor the circuit load in real time, and automatically switch to the backup power supply when an abnormality is detected in the lighting system a1 line.

[0030] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for optimizing power consumption of a smart sports stadium, characterized in that: The method comprises the following steps: Step 1: Identify the electrical equipment and collect historical power consumption data and power consumption of each electrical equipment at fixed time intervals; Step 2: Determine the influencing factors of power consumption of each power-consuming device, and build a regression model with the influencing factors as independent variables and the power consumption of each power-consuming device as the dependent variable; Step 3: Predict future influencing factors based on known influencing factor data or build a time series model to further predict the power demand of each power-consuming device; Step 4: quantify the historical prediction error of the regression model of each power-consuming equipment and measure the prediction uncertainty of the time series model of each influencing factor; Step 5: Analyze the sensitivity of the regression model of each electrical equipment to the changes of each influencing factor, and transfer the uncertainty of the influencing factor to the uncertainty of the power consumption prediction of the electrical equipment; Step 6: Combine the inherent error of the regression model and the uncertainty of the input influencing factors to form an overall prediction uncertainty index, and configure the backup power supply accordingly.

2. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 1 is characterized in that: In step 1, the devices connected to the circuit are recorded, clustered according to their functions, and the power-consuming devices are determined, which are expressed as: [a1, a2, ..., a n ]; where n is a positive integer, indicating the number of types of electrical equipment; a1~a n Respectively represent the 1st to nth types of electrical equipment; The data collection time is expressed as: [T0, T1, T2, …, T m ]; where m is a positive integer, indicating the number of data collection moments, T0~T m They represent the 1st to m+1th data collection moments respectively; the time intervals between adjacent moments are the same.

3. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 2 is characterized in that: In step 1, historical electricity consumption data and power consumption of each power-consuming device are obtained based on smart meters and IoT sensors; The historical electricity consumption data is expressed as: [E(T 1-0 ),E(T 2-1 ),…,E(T m-(m-1) )]; where T 1-0 ~T m-(m-1) Respectively represent the time interval between T0 and T1 and the time interval between Tm-1 and Tm; E(T 1-0 )~E(T m-(m-1) ) respectively represent the power consumption data in the time interval from T0 to T1 ~ the power consumption data in the time interval from Tm-1 to Tm; The power consumption of each electrical equipment is expressed as: ; Among them, A[a1,T 1-0 ]~A[a n ,T 1-0 ] respectively represent the electrical equipment a1~a1 in the time interval from T0 to T1 n Power consumption; A[a1,T m-(m-1) ]~A[a n ,T m-(m-1) ] represent T m-1 to T m Electrical equipment a1~a within the time interval n Power consumption; In each time interval, the sum of the power consumption of each electrical device is equal to the power consumption data.

4. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 3 is characterized in that: In step 2, sensors are deployed to collect influencing factor data, including environmental parameters, venue status parameters, equipment status parameters, and personnel flow data; the influencing factors are expressed as: [I1, I2, …, I u ]; where u is a positive integer, indicating the number of influencing factors, I1~I u Respectively represent the 1st to uth influencing factors; Align the influencing factor data with the device power consumption data at the same time interval; Preprocessing: normalize the same influencing factor data at different time intervals and the power consumption of electrical equipment; For a certain electrical equipment, the influencing factor data corresponding to different time intervals after preprocessing and the power consumption of the electrical equipment are used as input to construct an LSTM neural network regression model to obtain a regression model for predicting the power consumption of the electrical equipment; Build a regression model for power consumption prediction for all power-consuming devices.

5. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 4 is characterized in that: In step 3, the known influencing factor data are obtained based on the meteorological agency or venue activity arrangement; for the unknown influencing factor data, a Prophet time series model is constructed for prediction: the influencing factor data of the historical time interval is used as input, and the Prophet time series model is constructed to output the future time interval T (m+1)-m The predicted value of the influencing factors; among them, T (m+1)-m Indicates T m to T m+1 the time interval between Based on the acquired data of known influencing factors and the predicted data of unknown influencing factors, combined with the regression model for predicting the power consumption of each electrical device, the predicted power consumption of each electrical device is obtained.

6. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 5 is characterized in that: In step 4, the historical prediction error of the regression model of each power-consuming device is quantified: in the training phase, when building the LSTM prediction model for each power-consuming device, the residual between the predicted value and the true value is calculated using historical data; For electrical equipment i , the historical residual standard deviation is ΔA i ; Where i is a positive integer, representing the type sequence of electrical equipment, i∈{1,2,…,n}; When using the Prophet model to predict unknown influencing factors, obtain the uncertainty index of each influencing factor; For each influencing factor I k , the Prophet model outputs a confidence interval or standard deviation when predicting, and records the future time interval T (m+1)-m The prediction standard deviation is σ Ik,T(m+1)-m ; Among them, k is a positive integer, representing the type sequence of influencing factors, k∈{1,2,…,u}; σ Ik,T(m+1)-m Indicates the influencing factor I of the Prophet model prediction k At time interval T (m+1)-m The standard deviation of .

7. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 6 is characterized in that: In step 5, the sensitivity of the LSTM model of each power-consuming device to the changes of each influencing factor is analyzed, so as to transfer the uncertainty of the influencing factor to the uncertainty of the power consumption prediction of the power-consuming device; For each device a i And each influencing factor I k , by making small perturbations on the model input and observing the changes in predicted power consumption, we can calculate the approximate local partial derivatives: ∂A i / ∂I k ≈(A i (I k +δ)-Ai(I k -δ)) / 2δ; Among them, A i Indicates that the influencing factor is I k Electrical equipment i The predicted power consumption; δ represents the numerical disturbance; A i (I k +δ) and A i (I k -δ) respectively represent the influencing factors I k +δ and I k -δ when electrical equipment a i The predicted power consumption; ∂A i / ∂I k Indicates electrical equipment i Influencing Factors I k sensitivity; This sensitivity coefficient is used to measure the contribution of the uncertainty of each influencing factor to the power consumption forecast, that is, the uncertainty propagation of a single influencing factor: V Ik,T(m+1)-m =|∂A i / ∂I k |⋅σ Ik,T(m+1)-m ; Among them, V Ik,T(m+1)-m Influencing factors I k At time interval T (m+1)-m The uncertainty of AI power consumption is propagated to the electrical equipment.

8. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 7 is characterized in that: In step 6, the model inherent error and input uncertainty are combined to obtain the overall prediction uncertainty: For electrical equipment i At time interval T (m+1)-m By combining the uncertainty propagation of each influencing factor, the input-related uncertainty is obtained: V input,I,T(m+1)-m =(V I1,T(m+1)-m 2 +V I2,T(m+1)-m 2 +…+V Iu,T(m+1)-m 2 ) 1 / 2 ; Among them, V input,I,T(m+1)-m Indicates that all influencing factors are within the time interval T (m+1)-m Propagate to the electrical equipment i Uncertainty in power consumption; Combining this input uncertainty with the model inherent residual uncertainty yields the total uncertainty: V total,I,T(m+1)-m =(ΔA i 2 +V input,I,T(m+1)-m 2 ) 1 / 2 ; Among them, V total,I,T(m+1)-m Indicates electrical equipment i At time interval T (m+1)-m The overall uncertainty within.

9. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 8 is characterized in that: In step 6, the predicted value of each power-consuming device in the future time interval and its uncertainty are used to guide the configuration of the backup power supply; For electrical equipment i , the predicted power consumption is A i , the configured spare capacity is calculated according to the formula: E=A i +k·V total,I,T(m+1)-m ; Where k is the safety factor, k>1, determined according to risk tolerance, equipment importance and system redundancy requirements; E represents the power equipment ai in the time interval T (m+1)-m Built-in backup power capacity.

10. The power supply energy consumption optimization analysis method of a smart sports stadium according to claim 9, characterized in that: In step 6, according to the backup power capacity calculated for each power-consuming device, a mobile backup power supply of corresponding capacity is configured in advance; the backup power supply is deployed at the key nodes of the circuit of the corresponding power-consuming device or is centrally stored in a location that is easy to dispatch; The operating status of each circuit is detected in real time. When a part of the circuit fails, the dispatch system will quickly identify the affected equipment and automatically instruct the corresponding mobile backup power supply to connect to the power supply to ensure the continuous operation of the power-consuming equipment.

Images