A prediction method for building-new energy vehicle load considering personnel characteristics

Abstract

The present invention discloses a method for predicting the load of new energy vehicles in buildings taking into account personnel characteristics, specifically: based on statistical data, a mathematical model of spatiotemporal variables corresponding to different user characteristics is established, and a travel chain of personnel is established based on the mathematical model of spatiotemporal variables corresponding to different user characteristics; the number of new energy vehicles in a certain area is estimated, and an electric vehicle charging load model and a hydrogen charging load model of hydrogen energy vehicles are constructed through a Monte Carlo simulation model, and the hydrogen and electricity loads of new energy vehicles in each area are predicted according to the charging load model and the hydrogen charging load model; the occupancy rate of personnel in residential buildings, office buildings, and shopping malls, and the overall heat dissipation of personnel are obtained, and the load models of residential buildings, office buildings, and shopping malls are established in combination with the geographical information of the area, and the predicted values ​​of cold, hot, and electric loads of buildings are obtained according to the load model prediction. The present invention takes into account factors such as the occupancy rate of personnel, user characteristics, and travel behavior, so that the prediction results are more in line with the actual situation and the prediction accuracy is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of new energy vehicle load prediction, and relates to a method for predicting building-new energy vehicle load taking into account personnel characteristics. Background Art

[0002] At present, most of the electricity load forecasts in cities are to forecast the total electricity load of the city as a whole. According to historical data, different mathematical models are established to predict the future data change trend based on historical laws. Since there are many factors affecting the energy load of buildings, including a large number of uncertain and random factors, the accuracy of the forecast is affected.

[0003] As environmental pollution becomes increasingly prominent, new energy vehicles, especially electric vehicles, have developed rapidly. However, the travel and demand of electric vehicle users are uncertain, which leads to the intermittent, volatile and random characteristics of electric vehicle charging load in the future, which brings great difficulties to the stable operation of the power grid. With the continuous increase in the penetration rate of electric vehicles, the accurate prediction of urban load becomes more and more important. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a method for predicting building-new energy vehicle load taking into account personnel characteristics, taking into account factors such as personnel occupancy rate, user characteristics, travel behavior, etc., so that the prediction results are more in line with the actual situation, improves the prediction accuracy, and solves the problems existing in the prior art.

[0005] The technical solution adopted by the present invention is a method for predicting building-new energy vehicle load taking into account personnel characteristics, which specifically includes the following steps:

[0006] S1: Based on statistical data, obtain user characteristics of a certain area of ​​the city, as well as the number of residences, shopping malls and office buildings; based on the spatiotemporal variables of travel behavior of different user characteristics, establish a mathematical model of spatiotemporal variables corresponding to different user characteristics, and based on the mathematical model of spatiotemporal variables corresponding to different user characteristics, establish a travel chain of personnel; determine whether the means of transportation for personnel travel is a new energy vehicle, if yes, proceed to step S2, if not, proceed to step S3;

[0007] S2: Based on the travel chain of people, the number of new energy vehicles in a certain area is estimated, and the charging load model of electric vehicles and the hydrogen charging load model of hydrogen vehicles are constructed through the Monte Carlo simulation model. The hydrogen and electric loads of new energy vehicles in each region are predicted based on the charging load model and the hydrogen charging load model;

[0008] S3: Based on the travel chain of personnel, obtain the occupancy rate and overall heat dissipation of residents, office buildings and shopping malls, and establish load models for residences, office buildings and shopping malls in combination with the geographic information of the area. According to the load model, the predicted values ​​of the cold, hot and electric loads of the buildings are obtained.

[0009] Furthermore, in step S1, the mathematical models of spatiotemporal variables corresponding to different user characteristics include: probability density functions of weekend / weekday departure times divided by age, gender, number of households, and household income, probability density functions of travel times for different departure-destination intervals, spatial transfer probability matrices between departures and destinations within k time periods, and prediction models for travel distance and stay time.

[0010] Furthermore, the establishment of the travel chain of the personnel: the departure time and the driving time are extracted according to the corresponding probability density function through the Monte Carlo simulation model, the travel destination is extracted according to the spatial transfer probability matrix between the departure place and the destination in the time period corresponding to the departure time, the driving time is obtained according to the corresponding departure-destination driving time probability density function, and the driving distance and the stay time are obtained through the XGBoost prediction model; thereby determining the time when the personnel arrive at the destination and the time when they leave the destination, and in the time period from arrival to departure, the number of personnel is added to the total number of personnel at the destination; the travel chain of all personnel in a certain area for one day is obtained by the same method.

[0011] Furthermore, the spatial transfer probability matrix between the departure point and the destination within the k time period is See formula (1):

[0012]

[0013] Among them, the matrix The i-th row represents the departure point D i , the jth column represents the destination D j , k is the number of time intervals after discretization.

[0014] Furthermore, the prediction method of the driving distance and the stay time is as follows: the travel day type, number of households, gender, age, family income, departure time, driving time, departure place, and destination in the statistical data are used as input features of the XGBoost model, and the corresponding driving distance and stay time are output.

[0015] Furthermore, the specific method of predicting the driving distance and the stay time by the XGBoost model is as follows:

[0016] Isolation forest method was used to identify outliers in the data set;

[0017] Standardize the data set;

[0018] The entire data set is randomly divided into training set and test set in proportion;

[0019] The grid search method is used to search the parameter space of the XGBoost model to obtain the optimal parameters for the test set evaluation;

[0020] The driving distance and stay time are predicted using the trained XGBoost model.

[0021] Furthermore, in step S2, the current charge / hydrogen charge state is determined by the charging load model and the hydrogen charging load model to determine whether charging / hydrogen charging is needed. If necessary, charging / hydrogen charging is started, and the charging / hydrogen charging time does not exceed the end value of the residence time. The charging time and charging load are counted.

[0022] Furthermore, in step S3, the method for determining the indoor rate of new energy vehicle owners in residences, office buildings, and shopping malls is specifically as follows: the travel chain data includes the departure location, travel distance, charging time, departure time, and destination. The stay time in each area every day is obtained according to the travel chain. The indoor time of a single person in each area is the stay time of the person in each area, thereby obtaining the indoor rate of new energy vehicle owners in residences, office buildings, and shopping malls.

[0023] Furthermore, the room rate Occ i,t Represents the distribution of people at different times, calculated according to the following formula:

[0024]

[0025] Among them, N i,t represents the number of people in building i at time t; where N i,max Indicates the maximum number of people in building i.

[0026] Furthermore, the total heat dissipation of personnel includes the cooling load formed by regional human body heat dissipation and the cooling load formed by human body moisture dissipation; the cooling load Q formed by regional human body heat dissipation τ1 for:

[0027]

[0028] In the formula, is the clustering coefficient, q 1,i is the hourly sensible heat dissipation of personnel, ζ is the time when personnel enter the air-conditioned area, τ-ζ is the time from the time when personnel enter the room, X τ-ζ is the cooling load coefficient of the sensible heat dissipation of human body during τ-ζ time, and m represents the number of people in the current area during the current period;

[0029] The cooling load Q generated by the moisture dissipation of human body in the area τ2 for:

[0030]

[0031] In the formula, q 2,i It is the hourly moisture dissipation of personnel.

[0032] The beneficial effects of the present invention are as follows: the present invention determines the scope of personnel activities in the urban area by dividing the building types in the area, thereby establishing a more realistic new energy vehicle charging and hydrogen filling scenario; based on the Monte Carlo simulation of the personnel spatiotemporal behavior model including the travel purpose, travel distance, departure and departure time of the personnel, it can truly reflect the spatiotemporal random characteristics of the charging and hydrogen filling behavior of the personnel; the building energy load considered according to the gender, age and travel characteristics of the personnel in the building can be more in line with the actual situation, and finally the cold, heat, electricity and hydrogen loads of the building-new energy vehicles in the area are predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 It is a load forecasting simulation flow chart of a building-new energy vehicle taking into account personnel characteristics of the present invention.

[0035] Figure 2 This is a flow chart of predicting driving distance and parking time based on the XGBoost model in the present invention.

[0036] Figure 3 This is a diagram showing the effect of residence time when the XGBoost model prediction model is used for prediction in the present invention.

[0037] Figure 4 This is a driving distance time effect diagram when using the XGBoost model prediction model for prediction in the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with 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 making creative work are within the scope of protection of the present invention.

[0039] Example,

[0040] A load forecasting method for buildings and new energy vehicles taking into account the random characteristics of personnel, such as Figure 1 As shown, the specific steps include:

[0041] S1: Based on statistical data, obtain the population situation in a certain area of ​​the city, the number of residences, shopping malls and office buildings. Analyze the spatiotemporal variables of travel behavior with different user characteristics (including departure time, travel purpose, travel time, travel distance, and stay time), and establish a mathematical model of spatiotemporal variables with different user characteristics; study the travel chain modeling method considering the uncertainty of travel behavior variables, and construct travel demand models for different types of users.

[0042] The spatiotemporal probability distribution of a single user's single trip is calculated as follows:

[0043] S11: Departure time and travel time

[0044] The departure time is divided according to the travel day type, family proportion (including the number of households, gender, age), and family income (divided into 24 types in the embodiment); the family proportion (including the number of households, gender, age) in a certain area is determined by the census data, and N personnel data are simulated, where N is determined by the size of the area. The optimal probability density function is fitted based on the departure time of the weekend and the week (Monday to Friday) of the N personnel data, and the departure time probability density function is shown in Table 1.

[0045] Table 1 Probability density function of departure time with different personnel characteristics and corresponding parameter fitting results

[0046]

[0047]

[0048] The driving time is divided according to the departure-destination (residence, office, shopping mall, visiting place or other places) (divided into 25 types in the embodiment, 5 departure places × 5 destinations = 25 types); the driving time distribution of different departure-destination and the corresponding parameter fitting results are shown in Table 2; wherein H represents home (residence), W represents work (office building), S represents shopping (shopping mall), V represents visitors or friends (visiting place), and O represents other trips (other places).

[0049] Table 2 Probability density functions of travel time for different departure-destination groups and corresponding parameter fitting results

[0050]

[0051] S12: Departure and destination selection

[0052] The spatial transfer of different departure points and destinations in different time periods in the travel data is statistically analyzed, and the spatial transfer probability is calculated based on the spatial transfer situation to obtain the travel purpose.

[0053]

[0054] matrix is the spatial transition probability matrix in the k time period. The i-th row represents the departure point D i , the jth column represents the destination D j , k is the number of time intervals after discretization.

[0055] S13: Driving distance and dwell time

[0056] The Xgboost model is used to predict the vehicle / personnel travel distance and stay time, such as Figure 2 As shown, specifically:

[0057] (1) Use the isolation forest method to identify outliers in the data set;

[0058] (2) Standardize the data set;

[0059] (3) The entire data set is randomly divided into training set and test set in a ratio of 7:3 or 8:2;

[0060] (4) Use the grid search method to search the XGBoost model parameter space and obtain the optimal parameters evaluated on the test set;

[0061] (5) Use the trained XGBoost model to predict driving distance and stay time.

[0062] The specific steps are as follows:

[0063] The travel day type (i.e., weekdays and weekends), household ratio (including number of households, gender, age), household income, departure time, travel time, departure place, and destination in the household travel (NHTS) data obtained in step S11 are used as model input features;

[0064] The island forest method is used to screen outliers and missing values ​​in the data set and delete them, and then the data is normalized.

[0065] The data set is standardized, and the standardization process is shown in formula (2).

[0066] x'=(xx min ) / (x max -x min ) (2)

[0067] Where x' is the normalized value of the variable, x is the actual value of the variable, and x min represents the minimum value of the variable, x max Indicates the maximum value of a variable.

[0068] The XGBoost algorithm is used for prediction, the driving distance and the dwell time are output, and the prediction error is calculated as shown in equations (3) to (6).

[0069]

[0070]

[0071]

[0072]

[0073] Formula (3) is the XGBoost model, where is the predicted value of the i-th sample, K is the number of trees, f k (x) is the kth tree function, x i represents the input variable of the i-th sample, and F represents the solution space of the regression tree.

[0074] O(f t ) represents the objective function, y i represents the true value of the i-th sample, T represents the number of leaf nodes, n represents the number of samples, l is the loss function, Ω(f t ) is the complexity of the t-th tree, γ and λ are penalty coefficients, T t is the number of tree child nodes in the tth tree, is the weight of the jth leaf node, see formula (4)-(5). In order to minimize the objective function, a new function (tree) f is added while training the model. t (x i ), and expand it by the second-order Taylor, as shown in formula (6), Represents the predicted value of the i-th sample at the t-1th iteration.

[0075] The grid search method is used to search the parameter space of the XGBoost algorithm to obtain the optimal parameters evaluated on the test set. The model parameters include the maximum depth of the tree, the weight sum of the minimum leaf node samples, the learning rate, and the number of iterations.

[0076] In order to evaluate the performance of the XGBoost prediction model, this paper uses root mean square error (RMSE), mean absolute percentage error (MAPE) and mean absolute error (MAE) as the error evaluation indicators of the model.

[0077] The entire training set is used as the algorithm input, and the XGBoost algorithm is retrained with the searched optimal parameters to obtain the final model.

[0078] Save the trained prediction model and input the travel data into the prediction model to obtain the driving distance and stay time.

[0079] Calculate the departure time of the next trip based on the travel time and the stop time, and repeat the above steps.

[0080] The present invention uses the travel day type, family proportion (including number of households, gender, age), family income, departure time, driving time, departure place, and destination in the household travel (NHTS) data as model input features to obtain the driving distance and stay time, solving the problem of mismatch of travel-related variables caused by the current random extraction based on the Monte Carlo algorithm. The Xgboost algorithm can be used to construct the relationship between input and output, and predict through variables with strong correlation to improve the prediction accuracy, such as the driving distance has a strong correlation with the driving time, and the stay time has a strong correlation with the destination. Machine learning algorithms are usually black box models, which obtain the relationship between input and output through historical data training and have poor interpretability. However, compared with the traditional Monte Carlo extraction only, the correlation between each variable is ignored, which can easily cause mismatch problems.

[0081] Establishment of personnel travel chain: The departure time and travel time are extracted according to the corresponding probability density function through the Monte Carlo simulation model, the travel destination is extracted according to the spatial transfer probability matrix between the departure place and the destination in the time period corresponding to the departure time, the travel time is obtained according to the corresponding departure-destination travel time probability density function, and the travel distance and stay time are obtained through the XGBoost prediction model; thereby determining the time when the personnel arrive at the destination and the time when they leave the destination, and in the time period from arrival to departure, the number of personnel is added to the total number of personnel at the destination; the travel chain of all personnel in a certain area for one day is obtained through the same method.

[0082] Example:

[0083] (1) extracting the departure time and the driving time according to the corresponding probability density function through the Monte Carlo simulation model, such as the departure time is 8:30 on weekdays, the driving time is 20 minutes, and the first departure location;

[0084] (2) Extract the travel destination according to the spatial transfer probability matrix between the departure place and the destination in the time period corresponding to the departure time. If the departure time is 8:30, the travel probability matrix between 8 and 10 o'clock is selected. According to the proportion of extraction from home to various locations shown in Table 3 (such as residence-office is 0.3193), it is assumed that the simulation extracts the residence to the office building.

[0085] Table 3 Probability of different departure-destination from 8:00 to 10:00

[0086] H(Purpose) W (purpose) S(Purpose) V(Purpose) O(Purpose) H(Departure) 0.0168 0.3193 0.2159 0.1139 0.3341 W(Departure) 0.1815 0.5246 0.1475 0.0437 0.1026 S(Departure) 0.3295 0.1635 0.3053 0.0614 0.1403 V(Departure) 0.2896 0.2225 0.1713 0.1398 0.1767 O(Departure) 0.3179 0.1067 0.2848 0.0706 0.2199

[0087] (3) According to the known conditions (age, gender, family income, departure time, travel time, departure-destination of the person), the travel distance and stay time are obtained through the XGBoost prediction model. In the embodiment, the prediction result is that the travel distance is 5 km and the stay time is 4 hours.

[0088] (4) The time of arrival at the office building is 8:50, and the time of departure is 12:50; at this time, the person is counted as being in the office from 8:50 to 12:50, and the number of people in the office building during this period is increased by 1. The departure time is 12:50, and the travel time and the travel purpose at the corresponding time are extracted. The above steps are repeated to obtain the travel situation of the person in one day. The travel chain of all people in a certain area in one day is obtained by the same method.

[0089] Determine whether the person's travel vehicle is a new energy vehicle. If so, proceed to step S2; if not, proceed to step S3.

[0090] S2: Estimate the number of new energy vehicles in the region, and construct a charging load model for electric vehicles and a hydrogen charging load model for hydrogen vehicles based on Monte Carlo simulation, and predict the hydrogen and electricity loads of new energy vehicles in each region according to the charging load model and the hydrogen charging load model; wherein, the number of simulated people is determined according to the size of the region, and the number of simulated people in the embodiment is 1000.

[0091] Construction of electric vehicle charging load model and hydrogen vehicle charging load model: Based on the travel chain of people, the number of new energy vehicles in a certain area is estimated, and the electric vehicle charging load model and hydrogen vehicle charging load model are constructed through the Monte Carlo simulation model. The hydrogen and electricity loads of new energy vehicles in each region are predicted based on the charging load model and the hydrogen charging load model.

[0092] The steps for calculating the electric vehicle charging load are as follows:

[0093] S21: Charging power selection

[0094] In the embodiment of the present invention, the electric vehicle is in a slow charging and fast charging mode with fixed power. In residential and office areas, it is in slow charging mode, with a charging power of 3.5KW and a charging efficiency of 90%; in shopping malls, it is in fast charging mode, with a charging power of 12KW and a charging efficiency of 85%.

[0095] S22: Battery state of charge

[0096] The battery state of charge of an electric vehicle can be calculated from its daily mileage and related parameters of the on-board battery, as shown in Formula 13:

[0097]

[0098] In the formula, SOC t Indicates the current state of charge of the battery of the electric vehicle, d is the mileage, E is the power consumption of the battery per kilometer, B is the battery capacity, SOC t-1 It is the battery state of charge after the last charging. Usually, the user's charging expectation does not change, so the battery state of charge after the electric vehicle is charged is regarded as its charging expectation.

[0099] S23: Charging demand judgment

[0100] Calculate the SOC when arriving at the destination. If it is lower than the threshold or it is the last trip, charge the car. The charging time does not exceed the end value of the stay time.

[0101] In real life, users generally expect 100% charging. To ensure the service life of the battery, users rarely charge the battery after it is exhausted. Therefore, the present invention assumes that the initial state of charge of the electric vehicle is not less than 20%.

[0102] S3: Based on the travel chain data of non-new energy vehicle personnel, obtain the indoor rate of new energy vehicle owners in residences, office buildings, and shopping malls, and the overall heat dissipation of personnel, and combine the geographical information of the area to establish the load model of residences, office buildings, and shopping malls. According to the load model, the predicted values ​​of the cold, hot, and electric loads of the buildings are obtained. Among them, the basic physical information of the building includes the heat transfer coefficient of the wall, the window-to-wall ratio, etc.

[0103] For the calculation of residential building load, it is assumed that there are y1 households in the house. The family population ratio, male-female gender ratio, and age ratio are obtained based on the personnel survey data. The indoor situation of each household in the house is obtained by combining the indoor rate of new energy vehicle owners and the indoor rate of non-new energy vehicle owners. Taking into account the different heat generation caused by different genders and ages of the human body, the cooling, heating, and electricity loads of the residential building are calculated. Taking into account the charging and hydrogen filling needs of new energy vehicles, the cooling, heating, electricity, and hydrogen load model of -residential-new energy vehicle is comprehensively calculated; for the calculation of office and shopping mall building loads, the indoor rate of office and shopping mall personnel and the heating value of personnel are obtained according to the staff stay time and gender ratio, and then the cooling, heating, and electricity loads of the office and shopping mall are calculated. Taking into account the charging and hydrogen filling needs of new energy vehicles, the cooling, heating, electricity, and hydrogen load model of the office and shopping mall is comprehensively calculated.

[0104] Based on the travel chain of people on the same day, the residence time of each area is obtained. The time a single person stays in each area is the residence time of the person in each area. The hourly number of people in each building is obtained through the spatiotemporal distribution results simulated by the travel behavior model. Considering the difference in the number of people in different buildings, the occupancy rate Occ i,t The distribution of people at different times is shown in the following formula:

[0105]

[0106] Among them, N i,t represents the number of people in building i at time t; where N i,max Indicates the maximum number of people in building i.

[0107] The heat dissipation and moisture dissipation of personnel need to be multiplied by the corresponding coefficients, taking into account the differences in gender, age and region. The heat dissipation and moisture dissipation of adult men in each region can be obtained by referring to relevant specifications. Taking the heat dissipation and moisture dissipation of adult men as the standard, the heat dissipation and moisture dissipation of adult women is multiplied by a coefficient of 0.85, and the heat dissipation and moisture dissipation of children is multiplied by a coefficient of 0.75.

[0108] According to the specifications, the heat dissipation q1 and moisture dissipation q2 corresponding to the person can be obtained.

[0109] The cooling load formed by human body heat dissipation can be calculated as follows:

[0110]

[0111] In the formula, Q i,τ1 Indicates the cooling load caused by human body heat dissipation. is the clustering coefficient, q 1,i is the hourly sensible heat dissipation of the person, ζ is the time when the person enters the air conditioner, τ-ζ is the time from the time when the person enters the room, X τ-ζ is the cooling load coefficient of the sensible heat dissipation of the human body during the τ-ζ time.

[0112] If the number of people in the current area during the current period is m, the cooling load caused by human body heat dissipation in this area is:

[0113]

[0114] Similarly, the cooling load formed by the moisture dissipated by the human body can be calculated as follows:

[0115]

[0116] In the formula, Q τ2 represents the cooling load caused by the moisture dissipation of human body in the area, q 2,i It is the hourly moisture dissipation of personnel.

[0117] Use the software EnergyPlus to build residential, office, and shopping mall building models, set relevant building information (such as floor height, room area, window-to-wall ratio, etc.), input the occupancy rate and overall heat dissipation of personnel in each area, input the geographic information of the area, and comprehensively calculate the cooling, heating, and electricity load curves of the buildings in each area.

[0118] Based on the hydrogen and electricity loads of new energy vehicles in each region calculated above, as well as the cooling, heating, and electricity loads of offices, residences, and shopping malls in each region, the cooling, heating, electricity, and hydrogen load curves of each building in the region are obtained by superposition.

[0119] The test results of the embodiment of the present invention for predicting the dwell time and the travel distance are as follows: Figure 3 , 4 As shown, from Figure 3 , 4 It can be seen that the prediction method of the embodiment of the present invention has high prediction accuracy and is of practical application significance.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for predicting building-new energy vehicle load taking into account personnel characteristics, characterized in that: The specific steps include: S1: Based on statistical data, obtain user characteristics of a certain area of ​​the city, as well as the number of residences, shopping malls and office buildings; based on the spatiotemporal variables of travel behavior of different user characteristics, establish a mathematical model of spatiotemporal variables corresponding to different user characteristics, and based on the mathematical model of spatiotemporal variables corresponding to different user characteristics, establish a travel chain of personnel; determine whether the means of transportation for personnel travel is a new energy vehicle, if yes, proceed to step S2, if not, proceed to step S3; S2: Based on the travel chain of people, the number of new energy vehicles in a certain area is estimated, and the charging load model of electric vehicles and the hydrogen charging load model of hydrogen vehicles are constructed through the Monte Carlo simulation model. The hydrogen and electric loads of new energy vehicles in each region are predicted based on the charging load model and the hydrogen charging load model; S3: Based on the personnel travel chain, obtain the occupancy rate and overall heat dissipation of residences, office buildings, and shopping malls. Combined with the geographical information of the area, establish the load model of residences, office buildings, and shopping malls. According to the load model, the predicted values ​​of the building's cooling, heating, and electricity loads are obtained; In step S1, the spatiotemporal variable mathematical models corresponding to different user characteristics include: weekend / weekday departure time probability density functions divided by age, gender, number of households, and household income, travel time probability density functions of different departure-destination, spatial transfer probability matrix between the departure place and the destination within k time periods, and prediction models of travel distance and stay time; Establishment of the personnel travel chain: extract the departure time and travel time according to the corresponding probability density function through the Monte Carlo simulation model, extract the travel destination according to the spatial transfer probability matrix between the departure place and the destination in the time period corresponding to the departure time, obtain the travel time according to the corresponding departure-destination travel time probability density function, and obtain the travel distance and stay time through the XGBoost prediction model; thereby determining the time when the personnel arrive at the destination and the time when they leave the destination, and adding the number of personnel to the total number of personnel at the destination in the time period from arrival to departure; obtain the travel chain of all personnel in a certain area for one day through the same method; The spatial transfer probability matrix between the departure point and the destination within k time periods See formula (1): Among them, the matrix The i-th row represents the departure point D i , the jth column represents the destination D j , k is the number of time intervals after discretization; The method for predicting the driving distance and the stay time is as follows: the type of travel day, number of households, gender, age, household income, departure time, driving time, departure place, and destination in the statistical data are used as input features of the XGBoost model, and the corresponding driving distance and stay time are output; The specific method of predicting the driving distance and residence time by XGBoost model: Isolation forest method was used to identify outliers in the data set; Standardize the data set; The entire data set is randomly divided into training set and test set in proportion; The grid search method is used to search the parameter space of the XGBoost model to obtain the optimal parameters for the test set evaluation; Predict the driving distance and stay time through the trained XGBoost model; In step S2, the current charge / hydrogen charge state is determined by the charging load model and the hydrogen charging load model to determine whether charging / hydrogen charging is required. If necessary, charging / hydrogen charging is started, and the charging / hydrogen charging time does not exceed the end value of the residence time; In step S3, the method for determining the indoor rate of new energy vehicle owners in residences, office buildings, and shopping malls is specifically as follows: the travel chain data includes the departure location, travel distance, charging time, departure time, and destination, and the stay time in each area every day is obtained according to the travel chain. The indoor time of a single person in each area is the stay time of the person in each area, thereby obtaining the indoor rate of new energy vehicle owners in residences, office buildings, and shopping malls; The room rate Occ i,t Represents the distribution of people at different times, calculated according to the following formula: Among them, N i,t represents the number of people in the building at time t; where N i,max Indicates the maximum number of people in a building; The total heat dissipation of personnel includes the cooling load formed by regional human body heat dissipation and the cooling load formed by human body moisture dissipation; the cooling load Q formed by regional human body heat dissipation τ1 for: In the formula, is the clustering coefficient, q 1,i is the hourly sensible heat dissipation of personnel, ζ is the time when personnel enter the air-conditioned area, τ-ζ is the time from the time when personnel enter the room, X τ-ζ is the cooling load coefficient of the sensible heat dissipation of human body during τ-ζ time, and m represents the number of people in the current area during the current period; The cooling load Q generated by the moisture dissipation of human body in the area τ2 for: In the formula, q 2,i It is the hourly moisture dissipation of personnel.

Images