Distributed energy equipment adjustment method based on neural network, electronic equipment and storage medium

Abstract

The invention provides a distributed energy equipment adjustment method based on a neural network, electronic equipment and a storage medium, and relates to the technical field of distributed energy equipment adjustment, and the method specifically comprises the steps: obtaining an electric power time matrix sequence of a historical region in combination with the season, weather condition and temperature of a t moment; the LSTM recurrent neural network [pi] 1 is trained; the total real-time power consumption power per 30 minutes in the next day is predicted by using the pi 1; combining seasons, weather conditions, average temperature and the like of k days to obtain a historical distributed energy equipment charging total energy consumption sequence W; the LSTM recurrent neural network pi < 2 > is trained; the total charging demand electric energy of the distributed energy equipment in a certain day is predicted by using pi 2; and calculating an optimal charging output power curve of the distributed energy equipment. According to the technical scheme, the problem that in the prior art, the distributed energy equipment peak conflicts with the life electricity consumption peak, and equipment resources cannot be fully utilized is solved.

Description

Technical Field

[0001] This invention relates to the field of distributed energy equipment regulation technology, specifically to a distributed energy equipment regulation method, electronic device, and storage medium based on neural networks. Background Technology

[0002] Distributed energy refers to systems that produce, store, and use energy in a decentralized and discrete manner, as opposed to traditional centralized energy systems. Common distributed energy devices include photovoltaic and wind power-related batteries and electric vehicles. As an important basic electrical device, distributed energy systems are experiencing rapid growth due to the rapid development of distributed microgrid systems and the ongoing electrification process aimed at low carbon emissions. The number of distributed photovoltaic, wind power, and energy storage devices is constantly increasing, as is the demand for residential power supply devices such as electric vehicle charging equipment and energy substitution devices for fossil fuels. In many areas, the control between distributed generation equipment and energy-consuming equipment remains disordered. Many energy-consuming devices require large amounts of power and long operating times, with energy consumption highly correlated with users' work and rest schedules. Furthermore, the peak power consumption of distributed generation equipment is inversely correlated with the peak electricity consumption of residential users, leading to a overlap between peak charging times and peak residential electricity consumption, thus placing enormous pressure on regional transformers.

[0003] Existing control systems do not consider transformer load in terms of power utilization. They only adjust energy efficiency by modifying equipment utilization rates and equipment usage time, which cannot resolve the conflict between peak demand from distributed energy devices and peak residential electricity consumption. Furthermore, the forecasting of distributed energy device demand is based solely on the number of devices and their operating hours, without considering important influencing factors such as weekdays, temperature, and seasons.

[0004] Therefore, there is a need for a neural network-based distributed energy device regulation method, electronic device, and storage medium that can fully utilize equipment resources and reduce power loss. Summary of the Invention

[0005] The main objective of this invention is to provide a method for regulating distributed energy equipment, an electronic device, and a storage medium based on neural networks, in order to solve the problem in the prior art where peak hours of distributed energy equipment conflict with peak hours of residential electricity consumption, and the equipment resources cannot be fully utilized.

[0006] To achieve the above objectives, the present invention provides a method for regulating distributed energy devices based on neural networks, specifically including the following steps:

[0007] S1, combined with the season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, road conditions, and total real-time power consumption P at time t. t The historical regional power consumption time matrix sequence M is obtained.

[0008] S2, the first LSTM recurrent neural network π1 is trained using the historical regional power consumption time matrix sequence M.

[0009] S3 uses π1 to predict the total real-time power consumption every 30 minutes of the next day.

[0010] S4, combined with the season, weather conditions, average temperature, whether it is a working day, number of distributed energy devices without traffic restrictions, and total charging energy demand of distributed energy devices on day k, yields the historical total charging energy consumption sequence W of distributed energy devices.

[0011] S5 uses the historical total energy consumption sequence of distributed energy devices to train the second LSTM recurrent neural network π2.

[0012] S6 uses π2 to predict the total electrical energy demand for charging distributed energy devices on a certain day.

[0013] S7, taking a charging device with two charging positions as an example, calculate the charging time interval and charging time of the two charging positions.

[0014] S8. Based on the charging time interval and charging time obtained in step S7, construct the optimal output power curve L of the area transformer. e Based on steps S1 to S6, predict the total actual power consumption curve L for the next 24 hours. k Total electricity demand for charging distributed energy devices (W) per day k Calculate the optimal output power curve for charging distributed energy devices.

[0015] Furthermore, step S1 specifically involves defining a state matrix φ by combining the parameters of season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions at time t. t Let P be the total real-time power consumption at time t. t ; every 30 minutes for φ t and P t A record is made to obtain the historical regional power consumption time matrix sequence M = {m0, m1, ... m}. k ,...},in, This is the historical regional power consumption time matrix sequence for day k. The state matrix φ at the nth time of each day t ;t∈[0,47].

[0016] Further, step S2 specifically involves training the first LSTM recurrent neural network π1 using the historical regional power consumption time matrix M, where [φ t ,t] is the input parameter of π1, [Pt [t] represents the output parameters of π1. Adjust the neuron parameters so that π1 completes [φ] t ,t]to[P t The mapping of , t].

[0017] Furthermore, step S3 specifically includes the following steps:

[0018] S3.1, based on the weather and temperature information released by the meteorological bureau for the next day, as well as whether it is a weekday and the number of distributed energy devices not subject to traffic restrictions, a status information sequence [φ′] is constructed. t Using π1, the total real-time power consumption for every 30 minutes in the next day is predicted to be [P′t,t′];

[0019] S3.2, [P′ t ,t′] and the corresponding real-time total power consumption sequence [P t The error of t] is used to update and optimize π1, resulting in the optimized neural network π′1. Subsequently, the prediction of real-time power consumption is completed by the optimized neural network π′1.

[0020] Furthermore, step S4 specifically involves defining a state matrix φ by combining the season, weather conditions, temperature, whether it is a weekday, the number of distributed energy devices, and road condition parameters for day k. k Let W be the total electrical energy demand for charging distributed energy devices on day k. k The historical total energy consumption sequence for charging distributed energy devices is obtained as W = {w0, w1, ... w}. k , ...}.

[0021] Further, step S5 specifically involves: training the second LSTM recurrent neural network π2 using the historical total energy consumption sequence W of distributed energy devices, where [φ k [k] represents the network input parameters, [w] represents the input parameters of the network. k [k] represents the network's output parameters. By adjusting the neuron parameters, π2 completes [φ] k ,k]to[W k The mapping of [k].

[0022] Furthermore, step S6 specifically includes the following steps:

[0023] S6.1, based on the weather and average temperature released by the meteorological bureau for the next day, whether it is a weekday, the number of distributed energy devices without traffic restrictions, road conditions, and other known status information [φ′ k Using π2, the total daily charging energy consumption of distributed energy devices is predicted to be [w′]. k,k′];

[0024] S6.2, Predicted total energy consumption for charging distributed energy devices [w′] k [k′] and the corresponding actual total energy consumption for charging distributed energy devices [w k The error of [k] is used to update and optimize π2, resulting in a new network π′2. Thereafter, the prediction of daily charging energy consumption is completed by the new neural network π′2, and this process is repeated daily.

[0025] Furthermore, step S7 specifically includes the following steps:

[0026] S7.1, Assume a charging device is connected to a main charging station and a secondary charging station via a switch. When a distributed energy device is connected to the charging device, if the charging station is located at the main charging station, the initial charging time of the main charging station is... The estimated departure time of the main charging station is The system will ask the user if they accept flexible charging. If the user does not accept flexible charging, the system will determine whether to start charging directly based on whether the current power exceeds the limit, and calculate the charging end time of the main charging station.

[0027] S7.2 If the user accepts flexible charging, the system will continue to ask the user for their expected departure time. The charging of the distributed energy device will be completed between the current time and the expected departure time. The specific charging time will be allocated based on the predicted power curve, the charging time interval, and the charging time. At the same time, a charging discount will be provided to users who accept flexible charging to encourage them to accept flexible charging.

[0028] S7.3, if the charging position is located in the secondary charging position, the initial charging time of the secondary charging position is... The estimated departure time of the secondary charging station is First, ask the user at the secondary charging station if they accept flexible charging. If both the primary and secondary charging stations accept flexible charging, calculate the charging time interval and charging time for each station. The starting time of the charging time interval for the secondary charging station is after the charging completion time of the primary charging station.

[0029] Furthermore, step S7 also includes the following steps:

[0030] S7.4 If the main charging station accepts flexible charging and the secondary charging station does not accept flexible charging, the secondary charging station will start charging immediately after the main charging station finishes charging; the total charging time of the main and secondary charging stations and their charging time intervals are calculated with the expected departure time of the main charging station as the latest time.

[0031] S7.5 If the main charging station does not accept flexible charging and the secondary charging station accepts flexible charging, then the charging end time of the main charging station is taken as the start time of the charging time interval of the secondary charging station, and the expected departure time of the secondary charging station is taken as the end time of the charging time interval of the secondary charging station. Calculate the charging time interval of the secondary charging station and its charging time.

[0032] S7.6 If neither the main nor the auxiliary charging station accepts flexible charging, the distributed energy devices of the main and auxiliary charging stations will be charged in sequence according to the transformer status. That is, if the actual output power of the transformer is redundant relative to the maximum output power, the charging task will be carried out. If the actual output power is not redundant relative to the maximum output power, the charging time will be delayed until the output power of the transformer is redundant.

[0033] S7.7, obtain the charging time range and charging time of the main and secondary charging stations, and adjust accordingly based on requirements. The specific timing of the adjustment is as follows, among which These are the charging times for the main and secondary charging stations, respectively. These are the charging time intervals for the main and auxiliary charging stations, respectively.

[0034] Furthermore, step S7 also includes the following steps:

[0035] S7.8, during the charging time allocation process, if the predicted total charging demand w of distributed energy devices is... k Greater than This represents the next moment, where L v (t) is the optimal output power for charging distributed energy devices. Even if the current regional transformer cannot meet the charging needs of all distributed energy devices, it will issue a warning to users that the charging function is insufficient, so as to remind users to charge in other locations.

[0036] Further, step S8 specifically involves: obtaining the power p when the transformer has the highest transmission efficiency based on the charging time interval and charging time obtained in step S7. e Construct the optimal output power curve L of the regional transformer e ={P0 e , ..., P t e , ..., P 47 e}, L e With L k ={P0′,...,P t ′,...,P 47 ′} have the same dimensions, and P0 e P1 e , ..., P47 e All equal to p e L k To predict the L24-hour total power consumption based on the predicted real-time power curve, specifically using a certain point in time on the current day as the time origin. k and w k The optimal output power curve L for charging distributed energy devices is then obtained. v =L e -L k Based on the charging time intervals, charging power, and charging time of the distributed energy devices that accept flexible charging, the charging time of the distributed energy devices is rationally allocated until the L is fully charged. v Curve, where P t e P represents the optimal output power of the transformer at time t. t ′ represents the predicted power value of the electrical equipment at time t.

[0037] The present invention also provides a distributed energy device regulation electronic device based on neural network, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a distributed energy device regulation method based on neural network.

[0038] The present invention also provides a non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements a neural network-based method for regulating distributed energy devices.

[0039] The present invention has the following beneficial effects:

[0040] This invention obtains regional electricity load based on a more sophisticated deep learning time series prediction algorithm model pre-trained with data, and allocates the time and sequence of distributed energy devices according to the predicted peak load and the actual transformer capacity. This fully utilizes equipment resources, reduces energy loss, and ensures electricity safety while ensuring that regional transformers operate at optimal efficiency. Attached Figure Description

[0041] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0042] Figure 1 A schematic diagram of the charging structure of the main and secondary charging positions in step S7 of the present invention is shown.

[0043] Figure 2 A schematic diagram of the charging time interval and charging time period calculation method of the main and auxiliary charging positions in step S7 of the present invention is shown.

[0044] Figure 3 The optimal output power curve for charging distributed energy devices is shown. Detailed Implementation

[0045] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Example 1

[0047] A method for regulating distributed energy devices based on neural networks specifically includes the following steps:

[0048] S1, combined with the season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, road conditions, and total real-time power consumption P at time t. t The historical regional power consumption time matrix sequence M is obtained.

[0049] S2, the first LSTM recurrent neural network π1 is trained using the historical regional power consumption time matrix sequence M.

[0050] S3 uses π1 to predict the total real-time power consumption every 30 minutes of the next day.

[0051] S4, combined with the season, weather conditions, average temperature, whether it is a working day, number of distributed energy devices without traffic restrictions, and total charging energy demand of distributed energy devices on day k, yields the historical total charging energy consumption sequence W of distributed energy devices.

[0052] S5 uses the historical total energy consumption sequence of distributed energy devices to train the second LSTM recurrent neural network π2.

[0053] S6 uses π2 to predict the total electrical energy demand for charging distributed energy devices on a certain day.

[0054] S7, such as Figure 1 As shown, taking a charging device with two charging positions as an example, the charging time interval and charging time of the two charging positions are calculated.

[0055] S8. Based on the charging time interval and charging time obtained in step S7, construct the optimal output power curve L of the area transformer. eBased on steps S1 to S6, predict the total actual power consumption curve L for the next 24 hours. k Total electricity demand for charging distributed energy devices (W) per day k Calculate the optimal output power curve for charging distributed energy devices.

[0056] Specifically, step S1 involves defining a state matrix φ by combining parameters such as season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions at time t. t Let P be the total real-time power consumption at time t. t ; every 30 minutes for φ t and P t A record is made to obtain the historical regional power consumption time matrix sequence M = {m0, m1, ... m}. k ,...},in, This is the historical regional power consumption time matrix sequence for day k. Let φt be the state matrix at the nth time of each day; t∈[0,47].

[0057] Specifically, step S2 involves training the first LSTM recurrent neural network π1 using the historical regional power consumption time matrix M, where [φ t ,t] is the input parameter of π1, [P t [t] represents the output parameters of π1. Adjust the neuron parameters so that π1 completes [φ] t ,t]to[P t The mapping of , t].

[0058] Specifically, step S3 includes the following steps:

[0059] S3.1, based on the weather and temperature information released by the meteorological bureau for the next day, as well as whether it is a weekday and the number of distributed energy devices not subject to traffic restrictions, a status information sequence [φ′] is constructed. t Using π1, the total real-time power consumption for every 30 minutes in the next day is predicted, resulting in the predicted total real-time power consumption sequence [P′]. t ,t′];

[0060] S3.2, [P′ t ,t′] and the corresponding real-time total power consumption sequence [P t The error of t] is used to update and optimize π1, resulting in the optimized neural network π′1. Subsequently, the prediction of real-time power consumption is completed by the optimized neural network π′1.

[0061] Specifically, step S4 involves defining a state matrix φ by combining the parameters of season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions for day k. k Let w be the total electrical energy demand for charging distributed energy devices on day k. k The historical total energy consumption sequence for charging distributed energy devices is obtained as W = {w0, w1, ... w}. k , ...}.

[0062] Specifically, step S5 involves training the second LSTM recurrent neural network π2 using the historical total energy consumption sequence W of distributed energy devices, where [φ k [k] represents the network input parameters, [w] represents the input parameters of the network. k [k] represents the network's output parameters. By adjusting the neuron parameters, π2 completes [φ] k ,k]to[w k The mapping of [k].

[0063] Specifically, step S6 includes the following steps:

[0064] S6.1, based on the weather and average temperature released by the meteorological bureau for the next day, whether it is a weekday, the number of distributed energy devices without traffic restrictions, road conditions, and other known status information [φ′ k Using π2, the total daily charging energy consumption of distributed energy devices is predicted to be [w′]. k ,k′];

[0065] S6.2, Predicted total energy consumption for charging distributed energy devices [w′] k [k′] and the corresponding actual total energy consumption for charging distributed energy devices [w k The error of [k] is used to update and optimize π2, resulting in a new network π′2. Thereafter, the prediction of daily charging energy consumption is completed by the new neural network π′2, and this process is repeated daily.

[0066] Specifically, such as Figure 2 As shown, step S7 specifically includes the following steps:

[0067] S7.1, Assume a charging device is connected to a main charging station and a secondary charging station via a switch. When a distributed energy device is connected to the charging device, if the charging station is located at the main charging station, the initial charging time of the main charging station is... The estimated departure time of the main charging station is The system will ask the user whether they accept flexible charging through the human-computer interaction interface. If the user does not accept flexible charging, the system will determine whether to start charging directly based on whether the current power exceeds the limit, and calculate the charging end time of the main charging station.

[0068] S7.2, if the user accepts flexible charging, then continue to ask the user for their expected departure time. The charging of the distributed energy equipment will be completed between the current time and the expected departure time. The specific charging time is based on the predicted power curve (e.g., ...). Figure 3 As shown in the figure, the charging time interval and charging time are allocated; at the same time, charging discounts are provided to users who accept flexible charging to guide users to accept flexible charging; the predicted power curve is predicted every 30 minutes every day, and 48 data points can be obtained every day. The predicted power value is on the y-axis and the x-axis is from 0 to 47, thus forming a power curve.

[0069] S7.3, if the charging position is located in the secondary charging position, the initial charging time of the secondary charging position is... The estimated departure time of the secondary charging station is First, ask the user at the secondary charging station if they accept flexible charging. If both the primary and secondary charging stations accept flexible charging, calculate the charging time interval and charging time for each station. The starting time of the charging time interval for the secondary charging station is after the charging completion time of the primary charging station.

[0070] Specifically, step S7 also includes the following steps:

[0071] S7.4 If the main charging station accepts flexible charging and the secondary charging station does not accept flexible charging, the secondary charging station will start charging immediately after the main charging station finishes charging, and the main charging station will start flexible charging as early as possible; the total charging time of the main and secondary charging stations and their charging time intervals are calculated with the expected departure time of the main charging station as the latest time.

[0072] S7.5 If the main charging station does not accept flexible charging and the secondary charging station accepts flexible charging, then the charging end time of the main charging station is taken as the start time of the charging time interval of the secondary charging station, and the expected departure time of the secondary charging station is taken as the end time of the charging time interval of the secondary charging station. Calculate the charging time interval of the secondary charging station and its charging time.

[0073] S7.6 If neither the main nor the auxiliary charging station accepts flexible charging, the distributed energy devices of the main and auxiliary charging stations will be charged in sequence according to the transformer status. That is, if the actual output power of the transformer is redundant relative to the maximum output power, the charging task will be carried out. If the actual output power is not redundant relative to the maximum output power, the charging time will be delayed until the output power of the transformer is redundant.

[0074] S7.7, obtain the charging time range and charging time of the main and secondary charging stations, and adjust accordingly based on requirements. The specific timing of the adjustment is as follows, among which These are the charging times for the main and secondary charging stations, respectively. These are the charging time intervals for the main and auxiliary charging stations, respectively.

[0075] Specifically, step S7 also includes the following steps:

[0076] S7.8, during the charging time allocation process, if the predicted total charging demand w of distributed energy devices is... k Greater than This represents the next moment, where L v (t) is the optimal output power for charging distributed energy devices. Even if the current regional transformer cannot meet the charging needs of all distributed energy devices, it will issue a warning to users that the charging function is insufficient, so as to remind users to charge in other locations.

[0077] When a distributed energy device is connected to the charging equipment, the system asks the user if they accept flexible charging. If they do, the system asks for their expected departure time and offers a discount on the charging fee. If they do not accept, no discount is offered. The system then checks if the current transformer load is above the limit. If not, charging begins directly for the user's device. If the load is above the limit, the system enters a charging wait state until the transformer load becomes redundant before charging begins.

[0078] Once the user accepts flexible charging and inputs their expected departure time, the total real-time power curve L for other predicted power consumption is used. k The charging time period of the device is allocated according to the charging duration, charging start time and expected departure time, so that the charging work is completed before the expected end time and is distributed as much as possible during the off-peak hours.

[0079] To address the low utilization rate of charging equipment caused by devices remaining in their charging positions after charging is complete, each charging device is equipped with two charging interfaces (main and auxiliary) and two charging slots (main and auxiliary). If both charging slots are idle, the interface connected to the charging device first becomes the main charging slot, and the interface connected later becomes the auxiliary charging slot. If the main charging slot is occupied and the auxiliary charging slot is idle, the charging device connects to the auxiliary charging slot, calculates the charging time for the main charging slot, and displays the estimated charging time for the main charging slot. If this charging time meets the user's needs, the distributed energy device at the auxiliary charging slot will begin charging after the main charging slot has finished charging. The switching between the main and auxiliary charging interfaces is controlled by relays and other devices.

[0080] Specifically, step S8 involves: obtaining the power p when the transformer has the highest transmission efficiency based on the charging time interval and charging time obtained in step S7. e Construct the optimal output power curve L of the regional transformer e ={P0 e , ..., P te , ..., P 47 e}, L e With L k ={P0′,...,P t ′,...,P 47 ′} have the same dimensions, and P0 e P1 e , ..., P 47 e All equal to p e L k Based on the predicted real-time total power consumption curve; specifically, using 12:00 noon as the time origin, predict the L for the next 24 hours. k and w k The optimal output power curve L for charging distributed energy devices is... v =L e -L k Based on the charging time intervals, charging power, and charging time of the distributed energy devices accepting flexible charging, the charging time of the distributed energy devices is rationally allocated. Rational allocation means maximizing the transformer's output power at each moment and maintaining it at its optimal output power. The transformer is most efficient when it outputs maximum power. The total output power of the transformer is adjusted by regulating the charging time of numerous distributed devices.

[0081] Until L is filled v Curve, where P t e P represents the optimal output power of the transformer at time t. t ′ represents the predicted power value of the electrical equipment at time t.

[0082] This invention is applicable to areas most suitable for installing distributed energy charging equipment without adding or replacing transformers. It considers the impact of distributed energy equipment on regional electricity consumption, especially the impact of high-power charging on regional transformers. By predicting real-time regional electricity consumption, charging planning for distributed energy equipment can be carried out in advance, enabling more rational power allocation while meeting the charging needs of distributed energy equipment and improving the operating efficiency of regional transformers.

[0083] Example 2

[0084] The present invention also provides a distributed energy device regulation electronic device based on neural network, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a distributed energy device regulation method based on neural network.

[0085] This invention utilizes an LSTM (Long Short-Term Memory) recurrent neural network, taking parameterized variables such as weather, temperature, workday conditions, and the number of unrestricted distributed energy devices from historical data as inputs to the neural network. It predicts the energy demand for charging distributed energy devices in the near future and the real-time power of other electrical loads at various times, providing a data foundation for subsequent regulation.

[0086] This invention adjusts the time period based on the real-time power of other electrical devices at various times throughout the day, as well as the time interval and consumption of currently connected distributed energy sources, in order to improve the overall efficiency of charging equipment and enhance the safety of regional power equipment.

[0087] This invention addresses the problem of low equipment time utilization by using a switching device to switch between primary and secondary positions without adding any equipment.

[0088] Example 3

[0089] The present invention also provides a non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements a neural network-based method for regulating distributed energy devices.

[0090] This invention utilizes an LSTM (Long Short-Term Memory) recurrent neural network, taking parameterized variables such as weather, temperature, workday conditions, and the number of unrestricted distributed energy devices from historical data as inputs to the neural network. It predicts the energy demand for charging distributed energy devices in the near future and the real-time power of other electrical loads at various times, providing a data foundation for subsequent regulation.

[0091] This invention adjusts the time period based on the real-time power of other electrical devices at various times throughout the day, as well as the time interval and consumption of currently connected distributed energy sources, in order to improve the overall efficiency of charging equipment and enhance the safety of regional power equipment.

[0092] This invention addresses the problem of low equipment time utilization by using a switching device to switch between primary and secondary positions without adding any equipment.

[0093] Example 4

[0094] A method for regulating distributed energy devices based on neural networks specifically includes the following steps:

[0095] S1, combined with the season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, road conditions, and total real-time power consumption P at time t. t The historical regional power consumption time matrix sequence M is obtained.

[0096] S2, the first LSTM recurrent neural network π1 is trained using the historical regional power consumption time matrix sequence M.

[0097] S3 uses π1 to predict the total real-time power consumption every 30 minutes of the next day.

[0098] S4, combined with the season, weather conditions, average temperature, whether it is a working day, number of distributed energy devices without traffic restrictions, and total charging energy demand of distributed energy devices on day k, yields the historical total charging energy consumption sequence W of distributed energy devices.

[0099] S5 uses the historical total energy consumption sequence of distributed energy devices to train the second LSTM recurrent neural network π2.

[0100] S6 uses π2 to predict the total electrical energy demand for charging distributed energy devices on a certain day.

[0101] S7, such as Figure 1 As shown, taking a charging device with two charging positions as an example, the charging time interval and charging time of the two charging positions are calculated.

[0102] S8. Based on the charging time interval and charging time obtained in step S7, construct the optimal output power curve L of the area transformer. e Based on steps S1 to S6, predict the total actual power consumption curve L for the next 24 hours. k Total electricity demand for charging distributed energy devices (W) per day k Calculate the optimal output power curve for charging distributed energy devices.

[0103] Specifically, step S1 involves defining a state matrix φ by combining parameters such as season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions at time t. t Let P be the total real-time power consumption at time t. t ; every 30 minutes for φ t and P t A record is made to obtain the historical regional power consumption time matrix sequence M = {m0, m1, ... m}. k ,...},in, This is the historical regional power consumption time matrix sequence for day k. The state matrix φ at the nth time of each day t ;t∈[0,47].

[0104] Specifically, step S2 involves training the first LSTM recurrent neural network π1 using the historical regional power consumption time matrix M, where [φ t ,t] is the input parameter of π1, [P t[t] represents the output parameters of π1. Adjust the neuron parameters so that π1 completes [φ] t ,t]to[P t The mapping of , t].

[0105] Specifically, step S3 includes the following steps:

[0106] S3.1, based on the weather and temperature information released by the meteorological bureau for the next day, as well as whether it is a weekday and the number of distributed energy devices not subject to traffic restrictions, a status information sequence [φ′] is constructed. t Using π1, the total real-time power consumption for every 30 minutes in the next day is predicted, resulting in the predicted total real-time power consumption sequence [P′]. t ,t′];

[0107] S3.2, [P′ t ,t′] and the corresponding real-time total power consumption sequence [P t The error of t] is used to update and optimize π1, resulting in the optimized neural network π′1. Subsequently, the prediction of real-time power consumption is completed by the optimized neural network π′1.

[0108] Specifically, step S4 involves defining a state matrix φ by combining the parameters of season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions for day k. k Let w be the total electrical energy demand for charging distributed energy devices on day k. k The historical total energy consumption sequence for charging distributed energy devices is obtained as W = {w0, w1, ... w}. k , ...}.

[0109] Specifically, step S5 involves training the second LSTM recurrent neural network π2 using the historical total energy consumption sequence W of distributed energy devices, where [φ k [k] represents the network input parameters, [w] represents the input parameters of the network. k [k] represents the network's output parameters. By adjusting the neuron parameters, π2 completes [φ] k ,k]to[w k The mapping of [k].

[0110] Specifically, step S6 includes the following steps:

[0111] S6.1, based on the weather and average temperature released by the meteorological bureau for the next day, whether it is a weekday, the number of distributed energy devices without traffic restrictions, road conditions, and other known status information [φ′ k Using π2, the total daily charging energy consumption of distributed energy devices is predicted to be [w′]. k,k′];

[0112] S6.2, Predicted total energy consumption for charging distributed energy devices [w′] k [k′] and the corresponding actual total energy consumption for charging distributed energy devices [w k The error of [k] is used to update and optimize π2, resulting in a new network π′2. Thereafter, the prediction of daily charging energy consumption is completed by the new neural network π′2, and this process is repeated daily.

[0113] Specifically, such as Figure 2 As shown, step S7 specifically includes the following steps:

[0114] S7.1, Assume a charging device is connected to a main charging station and a secondary charging station via a switch. When a distributed energy device is connected to the charging device, if the charging station is located at the main charging station, the initial charging time of the main charging station is... The estimated departure time of the main charging station is The system will ask the user whether they accept flexible charging through the human-computer interaction interface. If the user does not accept flexible charging, the system will determine whether to start charging directly based on whether the current power exceeds the limit, and calculate the charging end time of the main charging station.

[0115] S7.2, if the user accepts flexible charging, then continue to ask the user for their expected departure time. The charging of the distributed energy equipment will be completed between the current time and the expected departure time. The specific charging time is based on the predicted power curve (e.g., ...). Figure 3 As shown in the figure, the charging time interval and charging time are allocated; at the same time, charging discounts are provided to users who accept flexible charging to guide users to accept flexible charging; the predicted power curve is predicted every 30 minutes every day, and 48 data points can be obtained every day. The predicted power value is on the y-axis and the x-axis is from 0 to 47, thus forming a power curve.

[0116] S7.3, if the charging position is located in the secondary charging position, the initial charging time of the secondary charging position is... The estimated departure time of the secondary charging station is First, ask the user at the secondary charging station if they accept flexible charging. If both the primary and secondary charging stations accept flexible charging, calculate the charging time interval and charging time for each station. The starting time of the charging time interval for the secondary charging station is after the charging completion time of the primary charging station.

[0117] Specifically, step S7 also includes the following steps:

[0118] S7.4 If the main charging station accepts flexible charging and the secondary charging station does not accept flexible charging, the secondary charging station will start charging immediately after the main charging station finishes charging, and the main charging station will start flexible charging as early as possible; the total charging time of the main and secondary charging stations and their charging time intervals are calculated with the expected departure time of the main charging station as the latest time.

[0119] S7.5 If the main charging station does not accept flexible charging and the secondary charging station accepts flexible charging, then the charging end time of the main charging station is taken as the start time of the charging time interval of the secondary charging station, and the expected departure time of the secondary charging station is taken as the end time of the charging time interval of the secondary charging station. Calculate the charging time interval of the secondary charging station and its charging time.

[0120] S7.6 If neither the main nor the auxiliary charging station accepts flexible charging, the distributed energy devices of the main and auxiliary charging stations will be charged in sequence according to the transformer status. That is, if the actual output power of the transformer is redundant relative to the maximum output power, the charging task will be carried out. If the actual output power is not redundant relative to the maximum output power, the charging time will be delayed until the output power of the transformer is redundant.

[0121] S7.7, obtain the charging time range and charging time of the main and secondary charging stations, and adjust accordingly based on requirements. The specific timing of the adjustment is as follows, among which These are the charging times for the main and secondary charging stations, respectively. These are the charging time intervals for the main and auxiliary charging stations, respectively.

[0122] Specifically, step S7 also includes the following steps:

[0123] S7.8, during the charging time allocation process, if the predicted total charging demand w of distributed energy devices is... k Greater than This represents the next moment, where L v (t) is the optimal output power for charging distributed energy devices. Even if the current regional transformer cannot meet the charging needs of all distributed energy devices, it will issue a warning to users that the charging function is insufficient, so as to remind users to charge in other locations.

[0124] When a distributed energy device is connected to the charging equipment, the system asks the user if they accept flexible charging. If they do, the system asks for their expected departure time and offers a discount on the charging fee. If they do not accept, no discount is offered. The system then checks if the current transformer load is above the limit. If not, charging begins directly for the user's device. If the load is above the limit, the system enters a charging wait state until the transformer load becomes redundant before charging begins.

[0125] Once the user accepts flexible charging and inputs their expected departure time, the total real-time power curve L for other predicted power consumption is used. k The charging time period of the device is allocated according to the charging duration, charging start time and expected departure time, so that the charging work is completed before the expected end time and is distributed as much as possible during the off-peak hours.

[0126] To address the low utilization rate of charging equipment caused by devices remaining in their charging positions after charging is complete, each charging device is equipped with two charging interfaces (main and auxiliary) and two charging slots (main and auxiliary). If both charging slots are idle, the interface connected to the charging device first becomes the main charging slot, and the interface connected later becomes the auxiliary charging slot. If the main charging slot is occupied and the auxiliary charging slot is idle, the charging device connects to the auxiliary charging slot, calculates the charging time for the main charging slot, and displays the estimated charging time for the main charging slot. If this charging time meets the user's needs, the distributed energy device at the auxiliary charging slot will begin charging after the main charging slot has finished charging. The switching between the main and auxiliary charging interfaces is controlled by relays and other devices.

[0127] Specifically, step S8 involves: obtaining the power p when the transformer has the highest transmission efficiency based on the charging time interval and charging time obtained in step S7. e Construct the optimal output power curve L of the regional transformer e ={P0 e , ..., P t e , ..., P 47 e}, L e With L k ={P0′,...,P t ′,...,P 47 ′} have the same dimensions, and P0 e P1 e , ..., P 47 e All equal to p e L k Based on the predicted real-time total power consumption curve; specifically, using 8:00 AM as the time origin, predict the L for the next 24 hours. k and w k The optimal output power curve L for charging distributed energy devices is... v =L e -L kBased on the charging time intervals, charging power, and charging time of the distributed energy devices accepting flexible charging, the charging time of the distributed energy devices is rationally allocated. Rational allocation means maximizing the transformer's output power at each moment and maintaining it at its optimal output power. The transformer is most efficient when it outputs maximum power. The total output power of the transformer is adjusted by regulating the charging time of numerous distributed devices.

[0128] Until L is filled v Curve, where P t e P represents the optimal output power of the transformer at time t. t ′ represents the predicted power value of the electrical equipment at time t.

[0129] This invention is applicable to areas most suitable for installing distributed energy charging equipment without adding or replacing transformers. It considers the impact of distributed energy equipment on regional electricity consumption, especially the impact of high-power charging on regional transformers. By predicting real-time regional electricity consumption, charging planning for distributed energy equipment can be carried out in advance, enabling more rational power allocation while meeting the charging needs of distributed energy equipment and improving the operating efficiency of regional transformers.

[0130] Example 5

[0131] A method for regulating distributed energy devices based on neural networks specifically includes the following steps:

[0132] S1, combined with the season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, road conditions, and total real-time power consumption P at time t. t The historical regional power consumption time matrix sequence M is obtained.

[0133] S2, the first LSTM recurrent neural network π1 is trained using the historical regional power consumption time matrix sequence M.

[0134] S3 uses π1 to predict the total real-time power consumption every 30 minutes of the next day.

[0135] S4, combined with the season, weather conditions, average temperature, whether it is a working day, number of distributed energy devices without traffic restrictions, and total charging energy demand of distributed energy devices on day k, yields the historical total charging energy consumption sequence W of distributed energy devices.

[0136] S5 uses the historical total energy consumption sequence of distributed energy devices to train the second LSTM recurrent neural network π2.

[0137] S6 uses π2 to predict the total electrical energy demand for charging distributed energy devices on a certain day.

[0138] S7, such as Figure 1 As shown, taking a charging device with two charging positions as an example, the charging time interval and charging time of the two charging positions are calculated.

[0139] S8. Based on the charging time interval and charging time obtained in step S7, construct the optimal output power curve L of the area transformer. e Based on steps S1 to S6, predict the total actual power consumption curve L for the next 24 hours. k Total electricity demand for charging distributed energy devices on day k (w) k Calculate the optimal output power curve for charging distributed energy devices.

[0140] Specifically, step S1 involves defining a state matrix φ by combining parameters such as season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions at time t. t Let P be the total real-time power consumption at time t. t ; every 30 minutes for φ t and P t A record is made to obtain the historical regional power consumption time matrix sequence M = {m0, m1, ... m}. k ,...},in, This is the historical regional power consumption time matrix sequence for day k. The state matrix φ at the nth time of each day t ;t∈[0,47].

[0141] Specifically, step S2 involves training the first LSTM recurrent neural network π1 using the historical regional power consumption time matrix M, where [φ t ,t] is the input parameter of π1, [P t [t] represents the output parameters of π1. Adjust the neuron parameters so that π1 completes [φ] t ,t]to[P t The mapping of , t].

[0142] Specifically, step S3 includes the following steps:

[0143] S3.1, based on the weather and temperature information released by the meteorological bureau for the next day, as well as whether it is a weekday and the number of distributed energy devices not subject to traffic restrictions, a status information sequence [φ′] is constructed. t Using π1, the total real-time power consumption for every 30 minutes in the next day is predicted to be [P′t,t′];

[0144] S3.2, [P′ t,t′] and the corresponding real-time total power consumption sequence [P t The error of t] is used to update and optimize π1, resulting in the optimized neural network π′1. Subsequently, the prediction of real-time power consumption is completed by the optimized neural network π′1.

[0145] Specifically, step S4 involves defining a state matrix φ by combining the parameters of season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions for day k. k Let w be the total electrical energy demand for charging distributed energy devices on day k. k The historical total energy consumption sequence for charging distributed energy devices is obtained as W = {w0, w1, ... w}. k , ...}.

[0146] Specifically, step S5 involves training the second LSTM recurrent neural network π2 using the historical total energy consumption sequence W of distributed energy devices, where [φ k [k] represents the network input parameters, [w] represents the input parameters of the network. k [k] represents the network's output parameters. By adjusting the neuron parameters, π2 completes [φ] k ,k]to[w k The mapping of [k].

[0147] Specifically, step S6 includes the following steps:

[0148] S6.1, based on the weather and average temperature released by the meteorological bureau for the next day, whether it is a weekday, the number of distributed energy devices without traffic restrictions, road conditions, and other known status information [φ′ k Using π2, the total daily charging energy consumption of distributed energy devices is predicted to be [w′]. k ,k′];

[0149] S6.2, Predicted total energy consumption for charging distributed energy devices [w′] k [k′] and the corresponding actual total energy consumption for charging distributed energy devices [w k The error of [k] is used to update and optimize π2, resulting in a new network π′2. Thereafter, the prediction of daily charging energy consumption is completed by the new neural network π′2, and this process is repeated daily.

[0150] Specifically, such as Figure 2 As shown, step S7 specifically includes the following steps:

[0151] S7.1, Assume a charging device is connected to a main charging station and a secondary charging station via a switch. When a distributed energy device is connected to the charging device, if the charging station is located at the main charging station, the initial charging time of the main charging station is... The estimated departure time of the main charging station is The system will ask the user whether they accept flexible charging through the human-computer interaction interface. If the user does not accept flexible charging, the system will determine whether to start charging directly based on whether the current power exceeds the limit, and calculate the charging end time of the main charging station.

[0152] S7.2, if the user accepts flexible charging, then continue to ask the user for their expected departure time. The charging of the distributed energy equipment will be completed between the current time and the expected departure time. The specific charging time is based on the predicted power curve (e.g., ...). Figure 3 As shown in the figure, the charging time interval and charging time are allocated; at the same time, charging discounts are provided to users who accept flexible charging to guide users to accept flexible charging; the predicted power curve is predicted every 30 minutes every day, and 48 data points can be obtained every day. The predicted power value is on the y-axis and the x-axis is from 0 to 47, thus forming a power curve.

[0153] S7.3, if the charging position is located in the secondary charging position, the initial charging time of the secondary charging position is... The estimated departure time of the secondary charging station is First, ask the user at the secondary charging station if they accept flexible charging. If both the primary and secondary charging stations accept flexible charging, calculate the charging time interval and charging time for each station. The starting time of the charging time interval for the secondary charging station is after the charging completion time of the primary charging station.

[0154] Specifically, step S7 also includes the following steps:

[0155] S7.4 If the main charging station accepts flexible charging and the secondary charging station does not accept flexible charging, the secondary charging station will start charging immediately after the main charging station finishes charging, and the main charging station will start flexible charging as early as possible; the total charging time of the main and secondary charging stations and their charging time intervals are calculated with the expected departure time of the main charging station as the latest time.

[0156] S7.5 If the main charging station does not accept flexible charging and the secondary charging station accepts flexible charging, then the charging end time of the main charging station is taken as the start time of the charging time interval of the secondary charging station, and the expected departure time of the secondary charging station is taken as the end time of the charging time interval of the secondary charging station. Calculate the charging time interval of the secondary charging station and its charging time.

[0157] S7.6 If neither the main nor the auxiliary charging station accepts flexible charging, the distributed energy devices of the main and auxiliary charging stations will be charged in sequence according to the transformer status. That is, if the actual output power of the transformer is redundant relative to the maximum output power, the charging task will be carried out. If the actual output power is not redundant relative to the maximum output power, the charging time will be delayed until the output power of the transformer is redundant.

[0158] S7.7, obtain the charging time range and charging time of the main and secondary charging stations, and adjust accordingly based on requirements. The specific timing of the adjustment is as follows, among which These are the charging times for the main and secondary charging stations, respectively. These are the charging time intervals for the main and auxiliary charging stations, respectively.

[0159] Specifically, step S7 also includes the following steps:

[0160] S7.8, during the charging time allocation process, if the predicted total charging demand wk of the distributed energy devices is greater than... This represents the next moment, where L v (t) is the optimal output power for charging distributed energy devices. Even if the current regional transformer cannot meet the charging needs of all distributed energy devices, it will issue a warning to users that the charging function is insufficient, so as to remind users to charge in other locations.

[0161] When a distributed energy device is connected to the charging equipment, the system asks the user if they accept flexible charging. If they do, the system asks for their expected departure time and offers a discount on the charging fee. If they do not accept, no discount is offered. The system then checks if the current transformer load is above the limit. If not, charging begins directly for the user's device. If the load is above the limit, the system enters a charging wait state until the transformer load becomes redundant before charging begins.

[0162] Once the user accepts flexible charging and inputs their expected departure time, the total real-time power curve L for other predicted power consumption is used. k The charging time period of the device is allocated according to the charging duration, charging start time and expected departure time, so that the charging work is completed before the expected end time and is distributed as much as possible during the off-peak hours.

[0163] To address the low utilization rate of charging equipment caused by devices remaining in their charging positions after charging is complete, each charging device is equipped with two charging interfaces (main and auxiliary) and two charging slots (main and auxiliary). If both charging slots are idle, the interface connected to the charging device first becomes the main charging slot, and the interface connected later becomes the auxiliary charging slot. If the main charging slot is occupied and the auxiliary charging slot is idle, the charging device connects to the auxiliary charging slot, calculates the charging time for the main charging slot, and displays the estimated charging time for the main charging slot. If this charging time meets the user's needs, the distributed energy device at the auxiliary charging slot will begin charging after the main charging slot has finished charging. The switching between the main and auxiliary charging interfaces is controlled by relays and other devices.

[0164] Specifically, step S8 involves: obtaining the power p when the transformer has the highest transmission efficiency based on the charging time interval and charging time obtained in step S7. e Construct the optimal output power curve L of the regional transformer e ={P0 e , ..., P t e , ..., P 47 e}, L e With L k ={P0′,...,P t ′,...,P 47 ′} have the same dimensions, and P0 e P1 e , ..., P 47 e All equal to p e L k Based on the predicted real-time total power consumption curve; specifically, using 12:00 AM as the time origin, predict the L for the next 24 hours. k and w k The optimal output power curve L for charging distributed energy devices is... v =L e -L k Based on the charging time intervals, charging power, and charging time of the distributed energy devices accepting flexible charging, the charging time of the distributed energy devices is rationally allocated. Rational allocation means maximizing the transformer's output power at each moment and maintaining it at its optimal output power. The transformer's efficiency is highest when it outputs maximum power. The total output power of the transformer is adjusted by adjusting the charging time of numerous distributed devices until L is filled. v Curve, where P t e P represents the optimal output power of the transformer at time t. t ′ represents the predicted power value of the electrical equipment at time t.

[0165] Therefore, by adjusting the charging time of distributed energy devices, the time efficiency of charging devices can be improved, while ensuring that the regional transformers operate at their optimal efficiency, thus avoiding the waste of equipment resources and transformer overload caused by disordered charging.

[0166] This invention is applicable to areas most suitable for installing distributed energy charging equipment without adding or replacing transformers. It considers the impact of distributed energy equipment on regional electricity consumption, especially the impact of high-power charging on regional transformers. By predicting real-time regional electricity consumption, charging planning for distributed energy equipment can be carried out in advance, enabling more rational power allocation while meeting the charging needs of distributed energy equipment and improving the operating efficiency of regional transformers.

[0167] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for regulating distributed energy devices based on neural networks, characterized in that, Specifically, the steps include the following: S1, combined with the season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, road conditions, and total real-time power consumption P at time t. t The historical regional power consumption time matrix sequence M is obtained; S2, the first LSTM recurrent neural network π1 is trained using the historical regional power consumption time matrix sequence M; S3 uses π1 to predict the total real-time power consumption every 30 minutes in the next day; S4, combined with the season, weather conditions, average temperature, whether it is a working day, number of distributed energy devices without traffic restrictions, and total charging energy demand of distributed energy devices on day k, the historical total charging energy consumption sequence W of distributed energy devices is obtained; S5, the second LSTM recurrent neural network π2 is trained using the historical total energy consumption sequence of distributed energy devices charging W; S6, using π2 to predict the total electrical energy demand for charging distributed energy devices on a certain day; S7, taking a charging device with two charging positions as an example, calculate the charging time interval and charging time of the two charging positions; S8. Based on the charging time interval and charging time obtained in step S7, construct the optimal output power curve L of the area transformer. e Based on steps S1 to S6, predict the total actual power consumption curve L for the next 24 hours. k Total electricity demand for charging distributed energy devices on day k (w) k Calculate the optimal output power curve for charging distributed energy devices.

2. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S1 specifically involves defining the state matrix φ by combining the parameters of season, weather conditions, temperature, whether it is a weekday, number of distributed energy devices, and road conditions at time t. t Let P be the total real-time power consumption at time t. t ; every 30 minutes for φ t and P t A record is made to obtain the historical regional power consumption time matrix sequence M = {m0, m1, ... m k ,...},in, This is the historical regional power consumption time matrix sequence for day k. The state matrix φ at the nth time of each day t ;k∈[0,47].

3. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S2 specifically involves training the first LSTM recurrent neural network π1 using the historical regional power consumption time matrix M, where [φ t ,t] are the input parameters of π1, [P t [,t] represents the output parameters of π1. Adjust the neuron parameters so that π1 completes [φ t ,t] to [P t The mapping of ,t].

4. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S3 specifically includes the following steps: S3.1, based on the weather and temperature information released by the meteorological bureau for the next day, as well as whether it is a weekday and the number of distributed energy devices not subject to traffic restrictions, a status information sequence is constructed [φ]. ′ t ,t ′ Using π1, the total real-time power consumption is predicted every 30 minutes for the next day, resulting in the predicted total real-time power consumption sequence [P]. ′ t ,t ′ ]; S3.2, [P i t ,t ′ ] and the corresponding real-time total power consumption sequence [P t The error of ,t] is used to update and optimize π1, resulting in the optimized neural network π. ′ 1. Subsequently, the prediction of real-time power consumption is handled by the optimized neural network π. ′ 1. Completed.

5. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S4 specifically involves defining the state matrix φ by combining the seasonality, weather conditions, temperature, whether it is a weekday, the number of distributed energy devices, and road conditions for day k. k Let w be the total electrical energy demand for charging distributed energy devices on day k. k The historical total energy consumption sequence for charging distributed energy devices, W = {w0, w1, ... w1}, is obtained. k ,...}.

6. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S5 specifically involves training the second LSTM recurrent neural network π2 using the historical total energy consumption sequence W of distributed energy devices, where [φ k [k] represents the input parameters of the network, [w] represents the input parameters of the network. k [,k] represents the network's output parameters. By adjusting the neuron parameters, π2 completes [φ k ,k] to [w k The mapping of [,k].

7. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S6 specifically includes the following steps: S6.1, based on the weather and average temperature released by the meteorological bureau for the next day, whether it is a weekday, the number of distributed energy devices without traffic restrictions, road conditions, and other known status information [φ ′ k ,k ' Using π², the total daily charging energy consumption of distributed energy devices is predicted, resulting in the predicted total charging energy consumption of distributed energy devices as [w]. ' k ,k ' ]; S6.2, Predicted total energy consumption for charging distributed energy devices [w] ' k ,k ' [and the corresponding total energy consumption for charging distributed energy devices [w]] k The error of [,k] is used to update and optimize π2, resulting in a new network π. ′ 2. Subsequently, the prediction of daily charging energy consumption was carried out by a new neural network π. ′ 2. Complete this step and repeat daily.

8. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S7 specifically includes the following steps: S7.1, Assume a charging device is connected to a main charging station and a secondary charging station via a switch. When a distributed energy device is connected to the charging device, if the charging station is located at the main charging station, the initial charging time of the main charging station is... The estimated departure time of the main charging station is The system will ask the user if they accept flexible charging. If the user does not accept flexible charging, the system will determine whether to start charging directly based on whether the current power exceeds the limit, and calculate the charging end time of the main charging station. S7.2 If the user accepts flexible charging, the system will continue to ask the user for their expected departure time. The charging of the distributed energy device will be completed between the current time and the expected departure time. The specific charging time will be allocated based on the predicted power curve, the charging time interval, and the charging time. At the same time, a charging discount will be provided to users who accept flexible charging to encourage them to accept flexible charging. S7.3, if the charging position is located in the secondary charging position, the initial charging time of the secondary charging position is... The estimated departure time of the secondary charging station is First, ask the user at the secondary charging station if they accept flexible charging. If both the primary and secondary charging stations accept flexible charging, calculate the charging time interval and charging time for each station. The starting time of the charging time interval for the secondary charging station is after the charging completion time of the primary charging station.

9. A method for regulating distributed energy devices based on neural networks according to claim 8, characterized in that, Step S7 also includes the following steps: S7.4 If the main charging station accepts flexible charging and the secondary charging station does not accept flexible charging, the secondary charging station will start charging immediately after the main charging station finishes charging; the total charging time of the main and secondary charging stations and their charging time intervals are calculated with the expected departure time of the main charging station as the latest time. S7.5 If the main charging station does not accept flexible charging and the secondary charging station accepts flexible charging, then the charging end time of the main charging station is taken as the start time of the charging time interval of the secondary charging station, and the expected departure time of the secondary charging station is taken as the end time of the charging time interval of the secondary charging station. Calculate the charging time interval of the secondary charging station and its charging time. S7.6 If neither the main nor the auxiliary charging station accepts flexible charging, the distributed energy devices of the main and auxiliary charging stations will be charged in sequence according to the transformer status. That is, if the actual output power of the transformer is redundant relative to the maximum output power, the charging task will be carried out. If the actual output power is not redundant relative to the maximum output power, the charging time will be delayed until the output power of the transformer is redundant. S7.7, obtain the charging time range and charging time of the main and secondary charging stations, and adjust accordingly based on requirements. The specific timing of the adjustment is as follows, among which These are the charging times for the main and secondary charging stations, respectively. These are the charging time intervals for the main and auxiliary charging stations, respectively.

10. A method for regulating distributed energy devices based on neural networks according to claim 9, characterized in that, Step S7 also includes the following steps: S7.8, during the charging time allocation process, if the predicted total charging demand w of distributed energy devices is... k Greater than This represents the next moment, where L v (t) is the optimal output power for charging distributed energy devices. Even if the current regional transformer cannot meet the charging needs of all distributed energy devices, it will issue a warning to users that the charging function is insufficient, so as to remind users to charge in other locations.

11. The method for regulating distributed energy devices based on neural networks according to claim 1, characterized in that, Step S8 specifically involves: based on the charging time interval and charging time obtained in step S7, determining the power p at which the transformer achieves its highest transmission efficiency. e Construct the optimal output power curve L of the regional transformer e ={P0 e ,...,P t e ,...,P 47 e }, L e With L k ={P0 ′ ,...,P t ′ ,...,P 47 ′ } have the same dimensions, and P0 e P1 e ,...,P 47 e All equal to p e ,L k To predict the L24-hour total power consumption based on the predicted real-time power curve, specifically using a certain point in time on the current day as the time origin. k and w k The optimal output power curve L for charging distributed energy devices is then obtained. v =L e -L k Based on the charging time intervals, charging power, and charging time of the distributed energy devices that accept flexible charging, the charging time of the distributed energy devices is rationally allocated until the L is fully charged. v Curve, where P t e P represents the optimal output power of the transformer at time t. t ′ Let be the predicted power value of the electrical equipment at time t.

12. A distributed energy device regulation electronic device based on neural networks, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the neural network-based distributed energy device regulation method as described in claims 1 to 11.

13. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the neural network-based distributed energy device regulation method as described in any one of claims 1 to 11.

Images