Micro-grid multi-energy optimization method and system integrated with AI algorithm

Abstract

The present application relates to the technical field of energy management, in particular to a micro-grid multi-energy optimization method and system integrated with an AI algorithm, comprising the following steps: collecting node power changes to generate an activity comparison table, identifying differences to adjust scheduling weights to form operation priorities, evaluating energy storage carrying capacity to generate a list, redistributing output according to capacity to generate a distribution table, and calculating path loss rate to generate a priority sequence; in the present application, the load activity difference is determined by introducing the power change median value, and the scheduling weight is adjusted in combination with the direction reversal feature to realize dynamic identification of node priorities; a carrying capacity model is constructed by fusing multiple parameters such as energy storage device capacity, frequency and capacity difference to enhance the accuracy of output evaluation; the energy distribution process is optimized based on the capacity and loss linkage mechanism; and the transmission priority sequence is established by the path input-output difference and operation frequency to effectively improve the accuracy of energy allocation and the response capability of system regulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy management, in particular to a micro-grid multi-energy optimization method and system integrated with an AI algorithm. BACKGROUND

[0002] The technical field of energy management involves intelligent monitoring, scheduling and control of various energy production, distribution, conversion and consumption processes. The core tasks include real-time data acquisition, load forecasting, energy efficiency analysis and optimization control of multi-energy systems such as electricity, heat, renewable energy, etc. The main goal is to achieve energy supply and demand balance, improve energy efficiency, reduce energy costs and promote green and low-carbon development. This technology is widely used in industrial parks, urban energy internet, micro-grids, building energy management systems and other scenarios. It often relies on information technology, automation systems and intelligent algorithms to achieve integrated scheduling and comprehensive optimization management of multi-energy systems. Among them, the traditional micro-grid multi-energy optimization method refers to the coordination and management of various energy resources in the micro-grid, such as photovoltaic, wind power, energy storage and load, etc. during operation and scheduling. It usually uses specific mathematical modeling and calculation methods such as mixed integer linear programming, wavelet transform prediction, price signal response algorithm, Markov chain load modeling to solve multi-energy coordination and optimization problems.

[0003] The existing technology uses mean model processing in the load evaluation link, lacks the ability to sensitively capture the trend of node behavior changes, and has identification lag and control decision deviation problems in high volatility scenarios. In the process of judging energy storage capacity, it relies on a single capacity indicator and does not consider the dynamic relationship of parameters during operation, which can easily cause the scheduling target to mismatch the actual support capacity, leading to execution failure. The path selection is based on fixed topology or preset parameters, ignoring the dynamic changes of performance degradation and resource load in transmission links, which can easily cause local path overload and continuous use of high loss channels, thereby affecting the real-time performance of energy scheduling and system operation efficiency. SUMMARY

[0004] In order to solve the technical problems existing in the prior art, the embodiments of the present application provide a micro-grid multi-energy optimization method integrated with an AI algorithm, which comprises the following steps:

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: a micro-grid multi-energy optimization method integrated with an AI algorithm, comprising the following steps:

[0006] S1: Collect the real-time power value and the average power value of the previous period of the multi-source coupled micro-grid load node within a set time window, calculate the power change ratio, take the median value of the change ratio of all nodes as the active reference value, compare it with each node and record the difference, and generate a load activity comparison table;

[0007] S2: based on the load activity control table, identify the nodes that are active and power direction reversal, perform exponential promotion on the AI scheduling weight and limit the upper limit, identify the nodes that are below the threshold and consistent in direction, proportionally reduce the weight, generate a node running priority table;

[0008] S3: according to the node running priority table, extract the remaining capacity, discharge frequency and charge-discharge capacity variation of the energy storage device, sequentially determine whether it meets the capacity threshold, frequency limit and operation interval, screen the devices that meet the conditions, and generate an energy storage bearability list;

[0009] S4: call the energy storage bearability list, screen the energy storage devices that cannot maintain the target output, limit the output within the capacity range, proportionally distribute the reduced part to the remaining capacity sufficient generator set and energy storage device, and generate an energy real-time distribution table.

[0010] As a further scheme of the present application, the load activity control table includes the node power change ratio median value, the node power change ratio result, and the activity difference index, the node running priority table includes the scheduling weight adjustment coefficient, the node activity classification result, and the power change direction label, the energy storage bearability list includes the capacity utilization evaluation value, the discharge frequency index, and the capacity change mean value, and the energy real-time distribution table includes the output adjustment distribution ratio, the device carrying capacity distribution, and the redundant output distribution result.

[0011] As a further scheme of the present application, the specific steps of S1 are:

[0012] S101: collect the real-time power value and the previous cycle average power value of the multi-source coupled micro-grid load node in the set time window, calculate the numerical proportion between the two, and correspondingly record the real-time value, average value and proportion of the node, to generate a power change ratio data group;

[0013] S102: based on all the proportion values in the power change ratio data group, extract the proportion numerical set of all nodes, sort them by numerical size, calculate the median of the proportion numerical set, and take the median as a unified reference value, to obtain the node activity reference result;

[0014] S103: call the power change ratio data group and the node activity reference result, and perform difference calculation on the node proportion value and the reference value, record the node number, proportion value and difference value, and generate a load activity control table.

[0015] As a further scheme of the present application, the specific steps of S2 are:

[0016] S201: Extract the node power change ratio and the change direction mark of the previous period based on the load activity control table, judge whether the current direction is opposite to the previous period direction, and select the node number with high activity degree higher than the node activity reference value to obtain the direction reversal active node number group;

[0017] S202: Call the direction reversal active node number group, extract the corresponding AI scheduling weight value, perform power operation with a fixed exponential factor for each weight value, limit the result within the set upper limit value, select the node number with low activity degree and consistent change direction, reduce the corresponding weight value according to the proportion factor, and generate the node update scheduling weight value group;

[0018] S203: According to the node updated weight value in the node update scheduling weight value group, arrange all node numbers in descending order according to the weight value, record the arranged node number as a sequence, and generate a node running priority order table.

[0019] As a further scheme of the application, the specific steps of S3 are:

[0020] S301: According to the node running priority order table, obtain the residual capacity, discharge frequency and charge-discharge capacity variation data of the energy storage device, compare the residual capacity with the capacity threshold one by one, and generate a capacity compliance list;

[0021] S302: Call the discharge frequency data in the capacity compliance list, and compare it with the frequency limit value one by one, and exclude the devices not in the limit range to generate a frequency compliance list;

[0022] S303: For the charge-discharge capacity variation data in the frequency compliance list, judge according to the upper and lower limits of the running interval, and only keep the devices within the interval range to generate an energy storage loadability list.

[0023] As a further scheme of the application, the specific steps of S4 are:

[0024] S401: Call the energy storage loadability list, select the energy storage device number with load capacity value less than the target output level, set the actual output value of the corresponding device as the load capacity value, and calculate the difference between the original output and the adjusted output of the device to obtain the total output reduction value group;

[0025] S402: Based on the output reduction total value group, extract the energy storage device and generator set number with residual capacity higher than the set proportion threshold, call the current residual capacity value, perform proportional normalization operation on the device residual capacity value, calculate the normalized residual capacity value, and obtain the allocable device proportion value group;

[0026] S403: According to the allocable device ratio value set and the output reduction total value set, the actual output compensation value that each device should bear on a ratio basis is calculated, and the device number and the corresponding compensation value are combined and recorded uniformly to generate an energy real-time allocation table.

[0027] As a further scheme of the present application, the method further comprises:

[0028] S5: Based on the energy real-time allocation table, the input-output power difference value, the running frequency and the service time of the device in the energy transmission path are extracted, the path loss ratio is calculated, compared with the threshold value, and arranged in ascending order of loss ratio to generate an energy transmission path priority sequence.

[0029] As a further scheme of the present application, the energy transmission path priority sequence includes path loss ratio value, loss threshold value comparison state, and path ranking level.

[0030] As a further scheme of the present application, the specific steps of S5 are:

[0031] S501: Based on the energy real-time allocation table, the input power and output power values of the device in the energy transmission path are obtained, the input power is subtracted from the output power to obtain difference value data, and the running frequency and service time records are associated and stored to generate a power difference value data set;

[0032] S502: The difference value data, running frequency and service time in the power difference value data set are called, the difference value and input power ratio are used to calculate the path loss ratio, and the result is compared with the loss ratio threshold value, only the paths not exceeding the threshold value are retained to generate a loss ratio compliance set;

[0033] S503: According to the loss ratio value in the loss ratio compliance set, the loss ratio values are arranged in ascending order, and the arranged path sequence is recorded in order to generate an energy transmission path priority sequence.

[0034] The micro-grid multi-energy optimization system integrated with AI algorithm comprises:

[0035] The load change comparison module collects real-time power values and previous cycle average power values in a set time window of a multi-source coupled micro-grid load node, calculates the power change ratio and takes the median value as the active reference value in ascending order, and records the difference between the ratio and the active reference value to generate a load activity comparison table;

[0036] The node operation sorting module identifies the active nodes with reversed power change direction based on the load activity comparison table, performs power operation and limits the weight, identifies the nodes with lower activity than the threshold value and consistent direction, and sorts the nodes with reduced weight in proportion to generate a node operation priority order table;

[0037] The energy storage capacity verification module extracts the residual capacity, discharge frequency and charge-discharge capacity variation of the energy storage device according to the node operation priority table, and sequentially determines whether the energy storage device meets the capacity threshold, frequency limit and operation interval, so as to screen the devices meeting the conditions and generate an energy storage loadability list;

[0038] The energy distribution adjustment module calls the energy storage loadability list to screen the energy storage devices with insufficient load capacity, limits the output within the load range, distributes the output difference to the generators and energy storage devices with sufficient capacity in proportion, and generates an energy real-time distribution table;

[0039] The transmission path optimization module extracts the path input-output power difference, operation cycle frequency and service time according to the energy real-time distribution table, calculates the loss ratio of the path by using the difference and input power ratio, compares the loss ratio with the loss threshold, and arranges the paths in ascending order to generate an energy transmission path priority sequence.

[0040] Compared with the prior art, the advantages and positive effects of the present application are that:

[0041] In the present application, the load activity difference is determined by introducing the power change median value, the scheduling weight is adjusted combined with the direction reversal feature, the dynamic identification of node priority is realized, the load capacity, frequency and capacity difference of the energy storage device are fused to construct a load capacity model, the output evaluation accuracy is enhanced, the energy distribution process is optimized based on the capacity and loss linkage mechanism, the transmission priority sequence is established by the path input-output difference and operation frequency, and the accuracy of energy allocation and the response ability of system regulation are effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0043] Figure 1 The step flowchart of the present application is shown in the figure;

[0044] Figure 2 The S1 refinement schematic diagram of the present application is shown in the figure;

[0045] Figure 3 The S2 refinement schematic diagram of the present application is shown in the figure;

[0046] Figure 4 The S3 refinement schematic diagram of the present application is shown in the figure;

[0047] Figure 5 The S4 refinement schematic diagram of the present application is shown in the figure;

[0048] Figure 6 S5 is a refinement of the schematic diagram of the present application;

[0049] Figure 7 S5 is a refinement of the schematic diagram of the present application; DETAILED DESCRIPTION

[0050] The technical solutions in the present application will be described below with reference to the drawings.

[0051] In the embodiments of the present application, the words such as "exemplary", "for example", and the like are used to represent an example, illustration, or description. Any embodiment or design scheme described as "exemplary" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "exemplary" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0052] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding", and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0053] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0054] To make the technical problems, technical solutions, and advantages of the present application clearer, specific embodiments will be described in detail below with reference to the drawings.

[0055] Please refer to Figure 1 The embodiments of the present application provide a micro-grid multi-energy optimization method integrated with an AI algorithm, including the following steps:

[0056] S1: Collect the real-time power values and the average power values of the previous period of the multi-source coupled micro-grid load nodes in a set time window, calculate the power change ratio, and set the median of the power change ratios of all nodes as an active reference value. Compare the node power change ratio with the active reference value one by one and record the differences, and generate a load activity comparison table;

[0057] S2: Based on the load activity comparison table, identify the active and power change direction reversed multi-source coupled micro-grid load node, in the AI scheduling weight parameter pool, perform exponential operation on the scheduling weight to improve and limit in the upper limit range, identify the node whose activity is lower than the threshold and the power change direction remains consistent, proportionally reduce the scheduling weight, and generate a node operation priority table;

[0058] S3: According to the node operation priority table, extract the remaining available capacity percentage value of the energy storage device, the daily cumulative discharge frequency and the start-stop capacity difference of the last charge-discharge, respectively judge whether the capacity meets the scheduling threshold, whether the discharge frequency is within the scheduling frequency limit range and whether the capacity change is within the normal operation interval, combine the three judgment results to filter the energy storage devices that meet the operation conditions, and generate an energy storage loadability list;

[0059] S4: Call the energy storage loadability list, filter the energy storage devices whose carrying capacity is insufficient to maintain the target output level, limit the output within the carrying capacity range, and proportionally distribute the reduced output to the remaining capacity sufficient generator set and energy storage device, and generate an energy real-time distribution table;

[0060] S5: According to the energy real-time distribution table, extract the input power and output power difference of the equipment in the energy transmission path, the running cycle number and the service time, calculate the path loss ratio and compare it with the loss threshold, arrange the path in ascending order according to the loss ratio, and generate an energy transmission path priority sequence.

[0061] The load activity comparison table includes the node power change proportion median, the node power change proportion result, and the activity difference index. The node operation priority table includes the scheduling weight adjustment coefficient, the node activity classification result, and the power change direction label. The energy storage loadability list includes the capacity utilization rate evaluation value, the discharge frequency index, and the capacity change mean value. The energy real-time distribution table includes the output adjustment distribution ratio, the equipment carrying capacity distribution, and the redundant output distribution result. The energy transmission path priority sequence includes the path loss ratio value, the loss threshold comparison state, and the path sorting level.

[0062] Please refer to Figure 2 , the specific steps of S1 are as follows:

[0063] S101: Collect the real-time power value and the previous period average power value of the multi-source coupled micro-grid load node in the set time window, calculate the numerical proportion between the two, and record the real-time value, average value and proportion of the node correspondingly, and generate a power change proportion data group;

[0064] First, a fixed collection time window is determined, for example, distinguished in minutes, and the real-time power of each node is recorded continuously within the window, and the data of each node is read in the form of seconds or minutes, ensuring that the running state can be fully reflected, then the power data in the previous time window is summarized, the average power value of the node is calculated, and each real-time power value in the current window is calculated proportionally with the average value of the previous period, not only the node number is saved, but also the corresponding real-time power value and the average power value of the previous period are saved, and the ratio of the two is calculated as the power change ratio of the node at the corresponding time point, in order to ensure clear data classification, the obtained ratio is compared with the preset interval threshold, usually three intervals are set, the low interval is used to represent that the ratio value is significantly less than the previous period, the normal interval is used to represent that the ratio value is less different from the previous period, and the high interval is used to represent that the ratio value is significantly greater than the previous period, these intervals are set according to historical operation data and load fluctuation rules, for example, the power change range of most nodes in the past longer period is referred to, and reasonable upper and lower limit ranges are formed, and finally the power change ratio data set of each node in the window is formed.

[0065] S102: Based on all the ratio values in the power change ratio data set, the ratio value set of all nodes is extracted, sorted by value, the median of the ratio value set is calculated, and the median is taken as a unified reference value to obtain the node active reference result;

[0066] First, the ratio values obtained by all nodes in the time window are summarized one by one to form a complete ratio set, then the ratio values in the set are sorted by size, and the values are arranged from small to large by comparing them one by one, so as to avoid affecting the subsequent judgment due to disorder, after obtaining the complete order, the ratio value at the middle position is selected as a representative from the sorting result, if the number of ratio values is even, the average result of the middle two ratio values is taken to ensure the balance of the result, the result is the unified reference value, which can eliminate the deviation caused by abnormal data of individual nodes to a certain extent, before setting the reference value, the historical records of the microgrid operation and the fluctuation of each node are combined to analyze in which interval most ratio values are concentrated, and then the median is determined as the reference to obtain the unified reference value of node activity.

[0067] S103: Call the power change ratio data set and the node active reference result to calculate the difference between the node ratio value and the reference value, record the node number, ratio value and difference value, and generate a load activity comparison table;

[0068] Read the records in the power change ratio data set one by one, extract the node number and the ratio value, then perform difference operation on the ratio value and the unified reference value to obtain the difference value of each node. After completing the difference operation, compare the difference value with the set threshold value. If the difference value is greater than the positive threshold value, it is determined to be high; if the difference value is between the positive and negative threshold values, it is determined to be normal; if the difference value is less than the negative threshold value, it is determined to be low. The determination of the threshold value is based on the power fluctuation range and stability characteristics of the node in the historical operation. For example, by comparing the long-term fluctuation range of the ratio value, the nodes with a difference not exceeding the preset value are considered normal, and the nodes exceeding the range are marked as high or low. All node numbers, ratio values and difference values are recorded in the load activity comparison table one by one, so that the table can clearly reflect the deviation of each node under the reference value.

[0069] Please refer to Figure 3 , the specific steps of S2 are:

[0070] S201: Based on the load activity comparison table, extract the node power change ratio and the change direction mark of the previous period, judge whether the current direction is opposite to the previous period direction, and select the node number whose activity is higher than the node activity reference value to obtain the direction reversal active node number group;

[0071] First, read the ratio value and direction mark of each node in the table one by one. The direction mark is obtained by comparing the current power with the power of the last period. If the current ratio is greater than the last period ratio, the direction is marked as rising, and if it is less than the last period ratio, the direction is marked as falling. Then compare the current direction with the last period direction one by one. For example, if the current direction of node N1 is rising and the last period direction is falling, it is judged that the direction is opposite. If the current direction of node N2 is rising and the last period direction is also rising, it is judged that the direction is consistent. On this basis, compare the activity of the node with the activity reference value. If the ratio value of a certain node exceeds the upper limit of the reference value, it is marked as high activity. For example, if the reference value is 1.05 and the node ratio value is 1.20, it is judged as high activity. If the ratio value is lower than the lower limit of the reference value, it is considered as low activity. When dividing the interval, refer to the ratio fluctuation range of most nodes in the past operation data. Set the low interval to be less than 0.9, the normal interval to be 0.9 to 1.1, and the high interval to be greater than 1.1. When screening, only keep the node number that meets both the opposite direction and the activity higher than the reference value. For example, node N1 has opposite direction and ratio value of 1.20, which is recorded in the direction reversal active node number group.

[0072] S202: Call the direction reversal active node number group, extract the corresponding AI scheduling weight value, perform power operation on each weight value with a fixed exponential factor, and limit the result within the set upper limit value. At the same time, select the node number with low activity and consistent change direction, reduce the weight value corresponding to the ratio factor, and generate the node updated scheduling weight value group;

[0073] First, read the node number one by one, and then call the corresponding initial scheduling weight, for example, the weight of node N1 is 0.8, and the weight of node N2 is 0.7, perform the power operation of a fixed exponential factor on each weight value, for example, when the fixed exponential factor is 1.2, raise 0.8 to the power of 1.2, and limit the result within the set upper limit value, the set upper limit value is set with reference to the reasonable distribution range of the weight of each node, for example, set the upper limit to 1.5, to ensure that the weight will not exceed this value, after completing the promotion, it is also necessary to filter out the node number with low activity and consistent direction, for example, the activity of node N3 is lower than 0.9 and the direction is decreased for two consecutive periods, then the initial weight value of the node N3 is reduced by a proportional factor, the proportional factor is set according to the running state, for example, set to 0.85, then multiply the weight of node N3 by 0.85 based on the original value to get the updated value, all weight update results are recorded in the node update scheduling weight value group, so as to form the scheduling order subsequently.

[0074] S203: According to the weight value of the node updated in the node update scheduling weight value group, arrange all node numbers in descending order according to the weight value, record the arranged node number as a sequence, and generate a node running priority order table;

[0075] First, the updated weight value is corresponded to the corresponding node number one by one, for example, the weight of node N1 is updated to 1.2, the weight of node N2 is 0.9, and the weight of node N3 is 0.7, then compare the weight of each node one by one, arrange the node with high weight value in front and the node with low weight value in back, the sorting operation is completed by comparing the weight value of any two nodes, until the whole set is arranged, if the weight value is the same during the sorting process, auxiliary judgment can be made according to the size of the node number or the average level of historical activity, to ensure the uniqueness of the order, finally form a complete node number sequence, and record the sequence, generate a node running priority order table.

[0076] Please refer to Figure 4 , the specific steps of S3 are:

[0077] S301: According to the node running priority order table, obtain the residual capacity, discharge times and charge-discharge capacity variation data of the energy storage device, compare the residual capacity with the capacity threshold one by one, and generate a capacity compliance list;

[0078] First, read the node number in the priority table one by one and call the energy storage device data bound to each node, wherein the remaining capacity is obtained by subtracting the allocated power from the current storage power of the device, the discharge times are read by accumulating the historical record value of the device, and the charge and discharge capacity variation data is obtained by comparing the difference between the current period and the energy storage capacity of the last period. Then, the remaining capacity of each device is compared with the preset capacity threshold one by one. The capacity threshold is set according to the rated capacity of the device and the safety operation requirement, for example, 30% of the rated capacity is taken as the minimum threshold. If it exceeds the threshold, it is considered that the capacity is sufficient, and if it is lower, it is marked as insufficient. During the comparison process, if the remaining capacity of a certain device is 40%, which is higher than the threshold of 30%, it is included in the list, and if the remaining capacity of a certain device is only 25%, it is excluded. After completing the capacity judgment of all devices one by one, the device numbers and related data of the devices meeting the requirements are recorded in the capacity compliance list.

[0079] S302: Call the discharge frequency data in the capacity compliance list and compare it with the frequency limit value one by one to exclude devices not within the limit range and generate a frequency compliance list.

[0080] Read the discharge frequency of each device in the capacity compliance list one by one and compare it directly with the preset frequency limit value. The frequency limit value is set according to the service life of the device and the operation specification. For example, if the device discharge frequency is 15 times under the condition that the daily average allowed discharge is not more than 20 times, it is judged to be within the limit range, and if it exceeds 20 times, it is judged to be out of range and excluded from the list. In order to ensure the completeness of the results, the frequency of all devices meeting the capacity requirements needs to be judged one by one. During this process, the device number and comparison result need to be recorded at the same time, and finally a frequency compliance list is formed which only retains the devices meeting the dual conditions of capacity and frequency.

[0081] S303: For the charge and discharge capacity variation data in the frequency compliance list, judge according to the upper and lower limits of the operation interval, and only retain the devices within the interval range to generate a storage loadability list.

[0082] Read the charge and discharge capacity variation value of each device in the list one by one and compare it directly with the preset upper and lower limits. The upper and lower limits are set according to the device operation stability and the needs of power grid dispatching. For example, the capacity variation is allowed to float within the range of ±15%. If the capacity variation of a certain device is within this range, it is considered to meet the requirements, and if it exceeds, it is excluded. Specifically, for example, the variation of device A is 10%, which is within the interval, and the variation of device B is 20%, which exceeds the upper limit, so it is excluded from the list. The judgment process needs to compare the charge and discharge capacity variation data of each device with the upper and lower limits one by one and record the results. Finally, only the device numbers with reasonable variation data are retained to form a storage loadability list.

[0083] Please refer to Figure 5The specific steps of S4 are as follows:

[0084] S401: Call the energy storage capacity list, filter all energy storage device numbers whose capacity values ​​are less than the target output level, set the actual output value of the corresponding device as the capacity value, and calculate the difference between the original output and the adjusted output of the device to obtain the total output reduction value group.

[0085] First, the equipment number and its load capacity value in the list are read one by one. Then, they are compared with the target output level. If the load capacity value of a certain equipment is lower than the target output level, the equipment is included in the screening results, and the actual output value of the equipment is set as its load capacity value. For example, if the target output of a certain equipment is 100 kilowatts and the load capacity is 80 kilowatts, then the actual output of the equipment is directly set to 80 kilowatts. After the output adjustment is completed, the difference between the original output and the adjusted output is calculated. That is, the original planned output is subtracted from the adjusted output to obtain the difference, which is recorded one by one. In this process, if the original output of the equipment is 100 kilowatts and the adjusted output is 80 kilowatts, the difference is 20 kilowatts. All the differences are accumulated in sequence to form a total output reduction value group, ensuring that the output adjustment result of each piece of equipment is clearly identified.

[0086] S402: Based on the total output reduction value group, extract the numbers of energy storage devices and generator sets whose remaining capacity is higher than the set percentage threshold, call the current remaining capacity value, normalize the remaining capacity value of the device according to the proportion, calculate the normalized remaining capacity value, and obtain the allocable device percentage value group.

[0087] The formula for calculating the normalized residual capacity is as follows:

[0088] ;

[0089] This formula is used to normalize the remaining capacity of energy storage devices and generator sets to form a set of proportional coefficient values ​​for load allocation and scheduling. The logical design of this formula aims to comprehensively consider key factors such as the current capacity (remaining capacity) of individual devices, the group status (average difference), the target scheduling baseline capacity, and current load disturbances, ensuring that the final proportional value possesses both dynamic responsiveness and resource balance. The following provides a detailed explanation of each parameter and the computational structure, and completes a full-process calculation example using real monitoring data.

[0090] in, Representative number is , time is The normalized residual capacity value, Representative number is , time is The current remaining capacity of the device. Represents all devices at any time the minimum residual capacity value of the device, representing the average residual capacity of all devices at time the sum of absolute values of the deviation of residual capacity from its average residual capacity, representing the average residual capacity of all devices at time the average residual capacity of the device, representing the time the target residual capacity reference value set, representing the time the total load disturbance correction factor detected in the time representing the total number of devices participating in the normalization calculation.

[0091] Parameter collection and assignment description and example calculation process:

[0092] Assume that the current dispatch time is 17:30 on the current day, the energy storage accessed is composed of 4 devices numbered 1-4, and the current residual capacity of each device is collected by the EMS platform as follows (sampling period 1 minute):

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] Therefore: ;

[0098] The dispatch center sets the target capacity as "75% of the average residual capacity of the device in the peak period of the previous day" ;

[0099] At the same time, the current total load value is shown by the collector reading as 420 kW, and the average load in this period yesterday was 395 kW, and the disturbance factor is calculated as:

[0100] ;

[0101] Calculate the sum of the group capacity deviation:

[0102] ;

[0103] Recalculate:

[0104] ;

[0105] Substitute all the data into the original formula to calculate the normalized capacity value of device 1:

[0106] ;

[0107] Table 1 Device capacity data and normalized calculation related parameters table:

[0108]

[0109] As shown in Table 1, the normalized value formed by comparing the average deviation and target value of the remaining capacity of each device is used to construct the subsequent scheduling ratio vector.

[0110] Final result meaning and comparison:

[0111] The calculation result is , combined with the normalized proportion standard [0, 1] interval, the value belongs to the medium-low level, which shows that device 1 has certain scheduling ability in the current group, but there is a large gap with the peak value, and its allocation weight needs to be controlled moderately. Through this value participating in the subsequent scheduling function calculation, it will further affect the unit load scheduling ratio.

[0112] Formula innovation point explanation:

[0113] The usefulness of the formula is that by introducing the target capacity offset term and the load disturbance term , the normalization not only depends on the device itself, but also reflects the target scheduling direction and running disturbance characteristics, so as to realize the enhancement of group state sensitivity and the improvement of scheduling stability in the load allocation strategy. This construction method is different from the traditional static normalization model and has stronger dynamic adjustment ability.

[0114] S403: According to the allocable device proportion value group and the total output reduction value group, calculate the actual output compensation value that each device should bear on the basis of proportion, and combine the device number with the corresponding compensation value to record uniformly, and generate an energy real-time allocation table;

[0115] First, read each difference value in the total output reduction value group one by one, and then calculate according to the allocation proportion of the allocable device proportion value group. For example, if the total reduction amount is 60 kilowatts, the proportion of device A is 0.625, and the proportion of device B is 0.375, then device A needs to bear the compensation value of 60x0.625, and device B needs to bear the compensation value of 60x0.375. This calculation is completed for each device in turn to obtain its actual output compensation value. Then, combine the device number with the compensation value piece by piece to form a corresponding relationship. In the result, if device A needs to compensate 37.5 kilowatts and device B needs to compensate 22.5 kilowatts, then A and 37.5 are combined, and B and 22.5 are combined when recording. Finally, all combinations are uniformly summarized into an energy real-time allocation table.

[0116] Please refer to Figure 6 , the specific steps of S5 are:

[0117] S501: Obtain the input power and output power values of the devices in the energy transmission path based on the real-time energy distribution table, subtract the output power from the input power to obtain the difference value data, and store the difference value data in association with the running frequency and service time record to generate a power difference value data set;

[0118] First, read the device number of each path in the distribution table, then call the input power and output power values in the running cycle, perform the operation of input power minus output power to obtain the power difference value, for example, if the input power of a certain energy storage device in the current cycle is 120 kW and the output power is 110 kW, the difference is 10 kW, after completing the difference calculation, the running frequency and service time information of each device are extracted, the running frequency is obtained by the accumulated start-stop times in the daily work record of the device, the service time is read by the accumulated days or hours of the device in use, then the power difference value is stored in association with the running frequency and service time, a one-to-one corresponding data entry is established, for example, the power difference value of device A is 10 kW, the running frequency is 320 times, and the service time is 4 years, then the three are saved together in the storage record, and finally all the records of the devices are summarized to generate a power difference value data set.

[0119] S502: Call the difference value data in the power difference value data set and the running frequency and service time, calculate the path loss ratio using the difference value and input power ratio, and compare the result with the loss ratio threshold value, only keep the paths that do not exceed the threshold value to generate a loss ratio compliance set;

[0120] Path loss ratio index calculation formula:

[0121] ;

[0122] Wherein, represents the path loss ratio index, represents the power difference value of the first path, represents the normalized cumulative running distance ratio factor of the first path, represents the total number of paths participating in the calculation, represents the input power of the first path.

[0123] The formula calculation logic is as follows: first, weight the power loss of each path by distance to form a weighted total loss as the numerator; then sum up the input power of all paths as the denominator; finally calculate the ratio to reflect the relative loss degree of all paths under unit input power.

[0124] Table 2 Path power and distance parameter table:

[0125]

[0126] As shown in Table 2, all parameter values are obtained by monitoring acquisition, there is no dimensional conflict between parameters, and the normalization factor is dimensionless.

[0127] Actual example calculation:

[0128] Numerator part:

[0129] ;

[0130] Denominator part:

[0131] ;

[0132] Final loss ratio calculation result:

[0133] ;

[0134] Threshold setting and judgment:

[0135] The historical data mean , standard deviation , the upper threshold value in the range of 95% is calculated:

[0136] ;

[0137] Current result <Threshold value 0.030, meets the retention condition.

[0138] Formula innovation point description:

[0139] The usefulness of the formula is that the distance normalization factor is introduced to calculate the power loss, which effectively expands the influence expression of the running path distance on energy consumption, enhances the recognition ability of long path high loss situation, and thus improves the sensitivity and efficiency of the energy consumption structure in the path screening strategy.

[0140] S503: According to the loss ratio in the loss ratio meeting set, the loss ratio values are arranged in ascending order, and the arranged path sequence is recorded in order to generate an energy transmission path priority sequence;

[0141] Read each path corresponding to the loss ratio in accordance with the set, and sequentially compare the size, arrange the lower value in front, the higher one in the back, for example, if a path ratio is 0.06, another is 0.08, the former is arranged in front, the sorting operation needs to traverse all paths until the complete sequence is generated, if the same ratio occurs, the device running time or the length of service time can be used for auxiliary sorting, so that the result remains unique, after sorting, the arranged path number is recorded one by one, forming the path order sequence, and the energy transmission path priority sequence is generated.

[0142] Please refer to Figure 7 The micro-grid multi-energy optimization system integrated with an AI algorithm comprises:

[0143] The load change comparison module collects real-time power values and previous cycle average power values in a set time window of a multi-source coupled micro-grid load node, calculates a power change ratio, takes a median value as an active reference value in ascending order, records a difference between the ratio and the active reference value, and generates a load activity comparison table;

[0144] The node operation sorting module identifies active nodes with reversed power change directions based on the load activity comparison table, performs power operation on the running weight and limits the value, identifies nodes with consistent directions and lower activity than a threshold value, sorts the nodes after the running weight is proportionally reduced, and generates a node operation priority order table;

[0145] The energy storage capability verification module extracts the remaining capacity, discharge frequency and charge-discharge capacity variation of the energy storage device according to the node operation priority order table, sequentially determines whether it meets the capacity threshold, frequency limit and operation interval, selects the devices meeting the conditions, and generates an energy storage loadability list;

[0146] The energy distribution adjustment module calls the energy storage loadability list to select energy storage devices with insufficient load capacity, limits the output within the load range, proportionally allocates the output difference to the generator set and the energy storage device with sufficient capacity, and generates an energy real-time distribution table;

[0147] The transmission path optimization module extracts the path input-output power difference, operation cycle frequency and service time according to the energy real-time distribution table, calculates the loss ratio of the path by using the difference and input power ratio, compares the loss ratio with a loss threshold, arranges the paths in ascending order, and generates an energy transmission path priority sequence.

[0148] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A microgrid multi-energy optimization method integrated with an AI algorithm, characterized in that, The method comprises the following steps: S1: collecting real-time power values and previous period average power values of multi-source coupled micro-grid load nodes within a set time window, calculating power change proportions, taking the median of the change proportions of all nodes as an active reference value, comparing the active reference value with each node and recording the differences, and generating a load activity comparison table; S2: based on the load activity comparison table, identifying nodes that are active and have power direction reversal, performing exponential promotion on AI scheduling weights and limiting them to an upper limit, identifying nodes that have activity lower than a threshold value and have consistent directions, proportionally reducing the weights, and generating a node operation priority order table; S3: according to the node operation priority order table, extracting the remaining capacity, discharge frequency and charge-discharge capacity variation of the energy storage device, sequentially determining whether it meets the capacity threshold, frequency limit and operation interval, screening the devices that meet the conditions, and generating an energy storage bearability list; S4: calling the energy storage bearability list, screening energy storage devices that cannot maintain the target output, limiting the output within the capacity range, proportionally distributing the reduced part to the remaining capacity sufficient generator set and energy storage device, and generating an energy real-time distribution table. 2.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 1, wherein, The load activity comparison table includes the median of the node power change proportion, the power change proportion of each node, and the activity difference index. The node operation priority order table includes the scheduling weight adjustment coefficient, the node activity classification result, and the power change direction label. The energy storage bearability list includes the capacity utilization evaluation value, the discharge frequency index, and the capacity variation mean. The energy real-time distribution table includes the output adjustment distribution ratio, the device bearing capacity distribution, and the redundant output distribution result. 3.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 1, wherein, The specific steps of S1 are as follows: S101: collecting real-time power values and previous period average power values of multi-source coupled micro-grid load nodes within a set time window, calculating the numerical proportion between the two, and corresponding recording the real-time value, average value and proportion of the node, generating a power change proportion data set; S102: based on all the proportion values in the power change proportion data set, extracting the proportion numerical set of all nodes, sorting them by numerical size, calculating the median of the proportion numerical set, and taking the median as a unified reference value to obtain the node active reference result; S103: calling the power change proportion data set and the node active reference result, calculating the difference between the node proportion value and the reference value, recording the node number, proportion value and difference value, and generating a load activity comparison table. 4.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 1, wherein, The specific steps of S2 are as follows: S201: based on the load activity comparison table, extracting the node power change proportion and the change direction label of the previous period, judging whether the current direction is opposite to the direction of the previous period, and simultaneously screening the node numbers with activity higher than the node active reference value to obtain the direction reversal active node number group; S202: calling the direction reversal active node number group, extracting the corresponding AI scheduling weight value, performing power operation on each weight value with a fixed exponential factor, limiting the result within a set upper limit value, simultaneously screening the node numbers with low activity and consistent change direction, and proportionally reducing the weight values corresponding to the proportion factor to generate a node updated scheduling weight value group; S203: According to the node update scheduling weight value group, the weight value of the node after the update is updated, all node numbers are arranged in descending order according to the weight value, the arranged node numbers are recorded as a sequence, and a node running priority order table is generated. 5.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 1, wherein, The specific steps of S3 are: S301: According to the node running priority order table, the residual capacity, discharge frequency and charge-discharge capacity variation data of the energy storage device are obtained, the residual capacity is compared with the capacity threshold one by one, and a capacity compliance list is generated; S302: The discharge frequency data in the capacity compliance list is called, and is compared with the frequency limit value one by one, and the devices not within the limit range are screened out, and a frequency compliance list is generated; S303: For the charge-discharge capacity variation data in the frequency compliance list, the upper and lower limits of the running interval are judged, and only the devices within the interval range are retained, and a storage loadability list is generated. 6.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 1, wherein, The specific steps of S4 are: S401: The storage loadability list is called, all storage device numbers with a load capacity value less than the target output level are screened, the actual output value of the corresponding device is set as the load capacity value, the difference between the original output and the adjusted output of the device is calculated, and a total output reduction value group is obtained; S402: Based on the total output reduction value group, the storage devices and generator sets with residual capacity higher than the set proportion threshold are extracted, the current residual capacity value is called, the device residual capacity value is proportionally normalized, the normalized residual capacity value is calculated, and a distributable device proportion value group is obtained; S403: According to the distributable device proportion value group and the total output reduction value group, the actual output compensation value that each device should bear on a proportional basis is calculated, and the device number and the corresponding compensation value are combined and recorded uniformly, and an energy real-time distribution table is generated. 7.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 1, wherein, The method further comprises: S5: Based on the energy real-time distribution table, the input and output power difference value, the running frequency and the service time of the device in the energy transmission path are extracted, the path loss ratio is calculated, compared with the threshold value in ascending order of loss ratio, and an energy transmission path priority sequence is generated. 8.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 7, wherein, The energy transmission path priority sequence includes path loss ratio value, loss threshold comparison state, path ranking level. 9.The microgrid multi-energy optimization method integrating an AI algorithm according to claim 7, wherein, The specific steps of S5 are: S501: Based on the energy real-time distribution table, the input power and output power values of the device in the energy transmission path are obtained, the input power is subtracted from the output power to obtain difference data, and the running frequency and service time records are associated and stored to generate a power difference data set; S502: The difference data in the power difference data set, the running frequency and the service time are called, the difference value and the input power ratio are used to calculate the path loss ratio, and the result is compared with the loss ratio threshold value, only the paths not exceeding the loss ratio threshold value are retained, and a loss ratio compliance set is generated; S503: According to the loss ratio value in the loss ratio compliance set, the paths are arranged in ascending order, and the arranged path sequence is recorded in order to generate an energy transmission path priority sequence.

10. A microgrid multi-energy optimization system integrated with an AI algorithm, characterized in that, The system is used to realize the integrated AI algorithm micro-grid multi-energy optimization method of any one of claims 1-9, and the system comprises: The load change ratio comparison module collects real-time power values and average power values of previous periods in a set time window of a multi-source coupled micro-grid load node, calculates a power change ratio, takes a median value of the ratio in ascending order as an active reference value, records a difference between the ratio and the active reference value, and generates a load activity comparison table; The node operation sequencing module identifies active nodes with reversed power change directions based on the load activity comparison table, performs power operation and limits values, identifies nodes with consistent directions and low activity, sorts nodes after proportional reduction of operation weights, and generates a node operation priority sequence table; The energy storage capability verification module extracts residual capacity, discharge frequency, and charge-discharge capacity variation of energy storage devices according to the node operation priority sequence table, sequentially determines whether the energy storage devices meet capacity thresholds, frequency limits, and operation intervals, selects devices meeting the conditions, and generates an energy storage loadability list; The energy distribution adjustment module calls the energy storage loadability list to select energy storage devices with insufficient load capacity, limits output to a load range, proportionally allocates a difference in output to generators and energy storage devices with sufficient capacity, and generates an energy real-time distribution table; The transmission path optimization module extracts path input-output power difference values, operation cycle numbers, and service times according to the energy real-time distribution table, calculates a loss ratio of the path by using a ratio of the difference value to the input power, compares the loss ratio with a loss threshold, sorts the path in ascending order, and generates an energy transmission path priority sequence.

Images