Rebalance dynamic adjustment method and service system applied to multi-source system

Abstract

The invention provides a rebalance dynamic adjustment method applied to a multi-source system and a service system, and relates to the technical field of smart energy, and the method comprises the steps: collecting a real-time operation data set of the multi-source energy system, covering energy supply, load demand and system operation state features, carrying out the dynamic supply and demand deviation feature extraction, and carrying out the dynamic adjustment of the dynamic supply and demand deviation features; a feature set containing supply, demand fluctuation and system stability deviation is generated, then a pre-trained multi-source collaborative prediction model is called, and a multi-source collaborative prediction result containing energy supply and load demand trend prediction features is generated based on the feature set; and dynamic adjustment strategy matching is carried out according to a multi-source collaborative prediction result and real-time operation data, and a strategy set of energy output adjustment, energy storage charging and discharging, load priority distribution and the like is obtained. And finally, executing an adjustment strategy and obtaining feedback data, optimizing parameters of the multi-source collaborative prediction model based on the adjustment strategy, triggering closed-loop adjustment operation, and realizing rebalance dynamic adjustment of the multi-source system.

Description

Technical Field

[0001] The present invention relates to the field of smart energy technology, and in particular to a rebalancing dynamic adaptation method and service system applied to a multi-source system. Background Art

[0002] In the field of smart energy, with the widespread application of multi-source energy systems, their efficient and stable operation faces many challenges. Traditional multi-source energy system management methods are often difficult to adapt to complex and changing operating environments, and have obvious limitations in practical applications.

[0003] Most of the existing multi-source energy system management methods use simple statistical analysis methods, which makes it difficult to effectively extract dynamic supply and demand deviation characteristics from massive data, and cannot accurately identify key information such as supply fluctuations, demand fluctuations, and system stability deviations, and thus cannot provide strong support for subsequent decision-making. In terms of prediction, the existing prediction models are mostly predictions for single energy sources or simple scenarios. There is a lack of effective prediction methods for the coordinated operation of multi-source energy systems, and it is impossible to comprehensively consider the mutual influence and synergy between multiple energy sources, resulting in inaccurate prediction results and difficulty in reflecting the real trend of energy supply and load demand. In addition, traditional methods are often based on experience and fixed rules, lack the ability to dynamically match and adjust according to real-time prediction results and system operating status, and cannot generate a multi-faceted collaborative adaptation strategy set, making it difficult to achieve comprehensive optimization of all aspects of the system. Summary of the invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a rebalancing dynamic adaptation method applied to a multi-source system, the method comprising: Collecting a real-time operating data set of a multi-source energy system, wherein the real-time operating data set includes energy supply characteristics, load demand characteristics, and system operating status characteristics; Performing dynamic supply and demand deviation feature extraction processing on the real-time operation data set to generate a dynamic supply and demand deviation feature set including supply fluctuation features, demand fluctuation features and system stability deviation; Calling a pre-trained multi-source collaborative prediction model to generate a multi-source collaborative prediction result based on the dynamic supply and demand deviation feature set, wherein the multi-source collaborative prediction result includes energy supply trend prediction features and load demand trend prediction features; Perform dynamic adaptation strategy matching processing on the multi-source collaborative prediction result and the real-time operation data set to generate a multi-source collaborative adaptation strategy set, wherein the multi-source collaborative adaptation strategy set includes an energy output adjustment strategy, an energy storage charging and discharging strategy, and a load priority allocation strategy; An adaptation strategy in the multi-source collaborative adaptation strategy set is executed and adaptation execution feedback data is obtained, and parameters of the multi-source collaborative prediction model are optimized based on the adaptation execution feedback data to trigger a closed-loop adaptation operation.

[0005] On the other hand, an embodiment of the present invention also provides a multi-source energy debugging service system, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0006] Based on the above aspects, the embodiment of the present application realizes efficient, accurate and adaptive dynamic adaptation of the multi-source energy system, and significantly improves the overall performance and operation efficiency of the multi-source energy system. Specifically, by collecting the real-time operation data set of the multi-source energy system, on this basis, the dynamic supply and demand deviation feature extraction processing is carried out, the supply fluctuation, demand fluctuation and system stability deviation are identified, and the pre-trained multi-source collaborative prediction model is called. The multi-source collaborative prediction results are generated based on the dynamic supply and demand deviation feature set, and the dynamic adaptation strategy matching processing is carried out according to the multi-source collaborative prediction results and the real-time operation data set, and a multi-source collaborative adaptation strategy set including energy output adjustment, energy storage charging and discharging, and load priority allocation is generated, so as to realize the comprehensive and collaborative optimization of all aspects of the system. Finally, the multi-source collaborative adaptation strategy is executed and the adaptation execution feedback data is obtained. Based on the adaptation execution feedback data, the multi-source collaborative prediction model is parameter optimized to trigger the closed-loop adaptation operation, forming a complete feedback optimization mechanism, so that the system can continuously self-learn and improve, and adapt to the ever-changing operating environment. As a result, the complex and changeable operation problems of the multi-source energy system are effectively solved, the energy utilization efficiency is improved, and the stability and reliability of the system are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a schematic diagram of the execution flow of the rebalancing dynamic adaptation method applied to a multi-source system provided by an embodiment of the present invention.

[0008] Figure 2 is a schematic diagram of exemplary hardware and software components of a multi-source energy debugging service system provided by an embodiment of the present invention; Figure numbers: 100 - multi-source energy debugging service system; 110 - network port; 120 - processor; 130 - bus; 140 - storage medium; 150 - I / O interface. DETAILED DESCRIPTION

[0009] The present invention will be described in detail below with reference to the accompanying drawings. Figure 11 is a flow chart of a rebalancing dynamic adaptation method applied to a multi-source system provided by an embodiment of the present invention. The rebalancing dynamic adaptation method applied to a multi-source system is introduced in detail below.

[0010] Step S110 , collecting a real-time operation data set of the multi-source energy system, wherein the real-time operation data set includes energy supply characteristics, load demand characteristics, and system operation status characteristics.

[0011] In this embodiment, there is a multi-source energy system in the energy park, which integrates multiple energy supply methods and various load demands, aiming to achieve efficient distribution and stable supply of energy. In detail, in the energy park, the real-time operation data set covers energy supply characteristics, load demand characteristics and system operation status characteristics. For example, in terms of energy supply, the photovoltaic power generation system starts working from sunrise every morning, and its power usually changes with the change of light intensity. For example, at 9 o'clock in the morning, the photovoltaic power generation power data is 500 kilowatts, and it rises to 600 kilowatts at 10 o'clock. The above data forms a photovoltaic power generation power data sequence. The wind power generation system outputs power according to the size of the wind speed. During a certain period of time, when the wind speed is relatively stable, the wind power generation power is maintained at about 800 kilowatts. As the wind speed changes, the power will also fluctuate accordingly, forming a wind power generation power data sequence. The energy storage system performs charging and discharging operations based on the energy supply and demand situation. Its output power data records the power value and time of each charging and discharging. For example, at a certain moment, the energy storage system outputs electrical energy to the system at a power of 200 kilowatts, thus forming an output power data sequence of the energy storage system.

[0012] In terms of load demand characteristics, there are many factories in the energy park, and the industrial load demand varies with production plans and equipment operation. When a factory conducts large-scale production in the morning, the industrial load demand data reaches 1,500 kilowatts at 10 o'clock. After that, as some equipment is shut down for maintenance, the load gradually decreases, forming an industrial load demand data sequence. In the commercial area, the business hours of the shops and the use of electrical equipment determine the commercial load demand. At around 12 noon, the commercial load demand data reaches 800 kilowatts, forming a commercial load demand data sequence. The electricity consumption in the residential area shows obvious periodicity. The peak period of electricity consumption is from 7 to 10 pm. The residential load demand data may rise from 600 kilowatts to 900 kilowatts during this period, forming a residential load demand data sequence.

[0013] In terms of system operation status characteristics, the grid frequency deviation data sequence reflects the difference between the actual operating frequency of the grid and the standard frequency. For example, during a certain period of time, the grid frequency deviation data fluctuates between -0.1Hz and 0.2Hz, and the above data is recorded to form a grid frequency deviation data sequence. The voltage deviation data sequence reflects the deviation of the grid voltage from the rated voltage. For example, at a certain moment, the voltage deviation data is +5V, and the voltage deviation data at different moments is recorded to form a voltage deviation data sequence. That is, through data acquisition equipment and sensors, the above-mentioned types of data are collected in real time to form a real-time operation data set of the multi-source energy system.

[0014] Step S120, performing dynamic supply and demand deviation feature extraction processing on the real-time operation data set to generate a dynamic supply and demand deviation feature set including supply fluctuation features, demand fluctuation features and system stability deviation.

[0015] In this embodiment, the photovoltaic power generation data sequence in the energy supply feature can be subjected to time series fluctuation analysis, such as calculating the instantaneous power change rate at a time interval of 10 minutes. For example, from 9:00 to 9:10 in the morning, the photovoltaic power generation power increased from 500 kilowatts to 550 kilowatts, and its instantaneous power change rate was (550-500) ÷ 500 × 100% = 10%, and this 10% can be understood as part of the first supply fluctuation sub-feature. For the wind power generation data sequence, calculate its power standard deviation. Assuming that in a certain hour, the wind power generation data is 780 kilowatts, 820 kilowatts, 800 kilowatts and other data, the power standard deviation obtained by calculation is 20 kilowatts, and this 20 kilowatts can be understood as the second supply fluctuation sub-feature. For the energy storage system output power data sequence, count its charge and discharge switching frequency. In one day, the energy storage system charges and discharges 15 times, so the charge and discharge switching frequency is 15 times / day, which can be used as the third supply fluctuation sub-feature.

[0016] In addition, the industrial load demand data sequence in the load demand feature can be processed for demand mutation detection. For example, by analyzing the data, it is detected that at 11 a.m., due to the start of a large equipment, the industrial load demand suddenly rises from 1,200 kilowatts to 1,800 kilowatts. This 11 o'clock is the mutation time point. All such mutation time points are recorded to form the first demand fluctuation sub-feature. For the commercial load demand data sequence, its load fluctuation amplitude is analyzed. For example, within one day, the commercial load demand ranges from the lowest 500 kilowatts to the highest 900 kilowatts, and the load fluctuation amplitude is 900-500=400 kilowatts, which is used as the second demand fluctuation sub-feature. For the residential load demand data sequence, by analyzing the data of multiple days, it is found that the residential electricity consumption shows a periodic fluctuation pattern with the peak electricity consumption from 7 to 10 p.m. every day and the low electricity consumption from 2 to 5 a.m., which constitutes the third demand fluctuation sub-feature.

[0017] Furthermore, the stability index calculation can be performed on the grid frequency deviation data sequence in the system operation status characteristics, and the grid frequency deviation data within a certain period of time is accumulated. For example, within 8 hours, the grid frequency deviation data are -0.1Hz, 0Hz, 0.1Hz, etc., and the cumulative offset is -0.1+0+0.1=0Hz, which is used as the first stability deviation sub-feature. For the voltage deviation data sequence, find out its instantaneous fluctuation extreme value. At a certain moment, the voltage deviation reaches +8V, which is the second stability deviation sub-feature. Finally, all the above sub-features are merged to generate a dynamic supply and demand deviation feature set that includes supply fluctuation characteristics, demand fluctuation characteristics and system stability deviation.

[0018] Step S130, calling a pre-trained multi-source collaborative prediction model, and generating a multi-source collaborative prediction result based on the dynamic supply and demand deviation feature set, wherein the multi-source collaborative prediction result includes energy supply trend prediction features and load demand trend prediction features.

[0019] In this embodiment, the dynamic supply and demand deviation feature set can be input into the encoder layer of the multi-source collaborative prediction model, and the encoder layer can perform multi-dimensional feature fusion processing on the dynamic supply and demand deviation feature set. For example, supply fluctuation characteristics such as the instantaneous power change rate of photovoltaic power generation, the standard deviation of wind power generation, and the charging and discharging switching frequency of the energy storage system are integrated with demand fluctuation characteristics such as the sudden change time point of industrial load demand, the fluctuation amplitude of commercial load demand, and the periodic fluctuation pattern of residential load demand, as well as system stability deviation characteristics such as the cumulative offset of grid frequency deviation and the instantaneous fluctuation extreme value of voltage deviation. After fusion, a fused feature vector containing supply and demand coupling characteristics and system stability association characteristics is generated.

[0020] Next, the time series prediction layer of the multi-source collaborative prediction model is called to perform time series expansion processing on the fused feature vector. For example, the energy supply trend and load demand trend in the next 24 hours can be predicted. Among them, the energy supply trend prediction features include the photovoltaic power generation power prediction curve, which predicts that in the next 24 hours, the photovoltaic power generation power will reach a peak of 1,200 kilowatts around 12 noon and then gradually decrease; the wind power generation power prediction curve shows that from 8 to 10 pm, as the wind speed increases, the wind power generation power will increase from 700 kilowatts to 900 kilowatts; the energy storage system dispatchable capacity prediction curve shows that at 4 pm, the energy storage system dispatchable capacity will reach 500 kilowatt-hours. The load demand trend forecast features include the industrial load demand forecast curve, which predicts that the industrial load demand will increase from 1,600 kW to 1,800 kW between 2 p.m. and 4 p.m.; the commercial load demand forecast curve shows that the commercial load demand will increase from 700 kW to 9 p.m.; and the residential load demand forecast curve shows that the residential load demand will increase from 800 kW to 1,000 kW between 8 p.m. and 10 p.m.

[0021] Then, the energy supply trend prediction characteristics and the load demand trend prediction characteristics can be used to calculate the supply and demand matching degree. For example, it is calculated that at 9 o'clock in the evening, there is a supply and demand gap of 200 kilowatts between energy supply and load demand, which is the supply and demand gap prediction value; at the same time, it is predicted that around 11 o'clock in the evening, energy supply and load demand will reach a balance, which is the supply and demand balance time point. By combining the above results, a multi-source collaborative prediction result containing energy supply trend prediction characteristics and load demand trend prediction characteristics can be generated.

[0022] Step S140, performing dynamic adaptation strategy matching processing based on the multi-source collaborative prediction result and the real-time operation data set to generate a multi-source collaborative adaptation strategy set, wherein the multi-source collaborative adaptation strategy set includes an energy output adjustment strategy, an energy storage charging and discharging strategy, and a load priority allocation strategy.

[0023] In this embodiment, the supply-demand gap forecast value can be matched with a plurality of preset supply-demand balance threshold intervals. Assume that the preset threshold intervals are: [-300 kW, -100 kW], [-100 kW, 100 kW], [100 kW, 300 kW], etc. The supply-demand gap forecast value at 9 pm is 200 kW, which belongs to the target threshold interval of [100 kW, 300 kW].

[0024] Then, according to the strategy generation rules corresponding to the target threshold interval, the set of dispatchable energy supply sources can be screened from the energy supply characteristics. For example, since the photovoltaic power generation system has basically stopped working at 9 o'clock in the evening, the wind power generation system still has a certain adjustment space, and the energy storage system can also perform charging and discharging operations, the set of dispatchable energy supply sources includes wind power generation systems and energy storage systems. Based on the current state of charge and dispatchable capacity in the energy storage charging and discharging strategy, assuming that the current state of charge of the energy storage system is 60% and the dispatchable capacity is 400 kWh, it is calculated that the maximum charging and discharging power of the energy storage system is 150 kW.

[0025] Next, according to the preset industry priority weights in the load priority allocation strategy, industrial loads have a higher priority due to production continuity requirements; commercial loads are second; and residential loads have a relatively lower priority. The industrial load demand data series, commercial load demand data series, and residential load demand data series are prioritized for load reduction. For example, when load reduction is required, residential loads are prioritized, followed by commercial loads, and the stable operation of industrial loads is guaranteed as much as possible.

[0026] Finally, based on the set of dispatchable energy supply sources, maximum charging and discharging power, and load reduction priority ranking, a set of multi-source coordinated adaptation strategies is generated. The energy output adjustment strategy is to appropriately increase the output power of the wind power generation system, for example, to increase the wind power generation power from the current 750 kilowatts to 850 kilowatts; the energy storage charging and discharging strategy is to allow the energy storage system to discharge at a power of 100 kilowatts; the load priority allocation strategy is to appropriately reduce the residential load when necessary, such as reducing the residential load by 100 kilowatts.

[0027] Step S150, executing the adaptation strategy in the multi-source collaborative adaptation strategy set and obtaining adaptation execution feedback data, and optimizing the parameters of the multi-source collaborative prediction model based on the adaptation execution feedback data to trigger a closed-loop adaptation operation.

[0028] In this embodiment, a power adjustment instruction in the energy output adjustment strategy can be sent to the photovoltaic inverter controller of the multi-source energy system. Since the power of the photovoltaic power generation system is relatively low at this time, the wind power generation system is mainly adjusted. After receiving the power adjustment instruction, the wind power generation system increases the power from 750 kilowatts to 850 kilowatts, and the actual adjustment response data of the photovoltaic power generation power is obtained (although the photovoltaic power generation power is not adjusted, the relevant data is recorded for overall feedback).

[0029] In addition, the charge and discharge rate instructions in the energy storage charge and discharge strategy can be sent to the bidirectional converter of the energy storage system, and the energy storage system discharges at a power of 100 kilowatts to obtain the actual charge and discharge power data of the energy storage system. For example, within 1 hour after executing the instruction, the energy storage system actually discharges at an average power of 95 kilowatts.

[0030] In addition, a load reduction instruction in the load priority allocation strategy can be sent to the load management terminal to reduce the residential load and obtain actual load reduction data, such as actually reducing the residential load by 80 kilowatts.

[0031] Furthermore, the actual adjustment response data, the actual charging and discharging power data, and the actual load reduction data can be combined to generate the adaptation execution feedback data. The strategy execution deviation index between the adaptation execution feedback data and the multi-source collaborative prediction results is calculated. For example, in the multi-source collaborative prediction results, the energy storage system is expected to discharge at a power of 100 kilowatts, the actual average power is 95 kilowatts, and the deviation is 5 kilowatts; the residential load is expected to be reduced by 100 kilowatts, and the actual reduction is 80 kilowatts, with a deviation of 20 kilowatts, etc. The strategy execution deviation index is obtained through comprehensive calculation.

[0032] Then, the parameters of the multi-source collaborative prediction model are optimized based on the adaptation execution feedback data to trigger the closed-loop adaptation operation. Then, the strategy execution deviation index is input into the feedback learning layer of the multi-source collaborative prediction model, and the gradient change of the feature fusion weight matrix of the encoder layer is calculated according to the strategy execution deviation index, and the feature fusion weight matrix is ​​updated based on the gradient descent algorithm. For example, when it is detected that the weight setting of certain supply fluctuation characteristics and demand fluctuation characteristics is unreasonable during fusion, resulting in a deviation between the prediction result and the actual execution situation, the weight matrix is ​​adjusted so that the above characteristics can more accurately reflect the actual situation when fused.

[0033] Finally, according to the time dimension deviation component in the strategy execution deviation index, adjust the window length and sliding step in the time window expansion parameters of the time series prediction layer. If the time dimension deviation component increases, it means that the time accuracy of the prediction decreases. Shorten the window length and increase the sliding step, for example, shorten the window length from the original 6 hours to 4 hours, and increase the sliding step from 1 hour to 2 hours; if the time dimension deviation component decreases, extend the window length and reduce the sliding step.

[0034] Therefore, the updated multi-source collaborative prediction results can be regenerated according to the adjusted feature fusion weight matrix and time window expansion parameters, and the updated multi-source collaborative prediction results can be matched with the real-time operation data set for secondary dynamic adaptation strategy matching to generate an optimized multi-source collaborative adaptation strategy set. By repeatedly executing the optimized multi-source collaborative adaptation strategy set and iteratively updating the parameters of the multi-source collaborative prediction model until the strategy execution deviation index is lower than the preset deviation threshold, the efficient and stable operation of the multi-source energy system can be achieved.

[0035] Based on the above steps, the embodiment of the present application realizes efficient, accurate and adaptive dynamic adaptation of the multi-source energy system, and significantly improves the overall performance and operation efficiency of the multi-source energy system. Specifically, by collecting the real-time operation data set of the multi-source energy system, on this basis, the dynamic supply and demand deviation feature extraction processing is carried out, the supply fluctuation, demand fluctuation and system stability deviation are identified, and the pre-trained multi-source collaborative prediction model is called. The multi-source collaborative prediction results are generated based on the dynamic supply and demand deviation feature set, and the dynamic adaptation strategy matching processing is carried out according to the multi-source collaborative prediction results and the real-time operation data set, and a multi-source collaborative adaptation strategy set including energy output adjustment, energy storage charging and discharging, and load priority allocation is generated, so as to realize the comprehensive and collaborative optimization of all aspects of the system. Finally, the multi-source collaborative adaptation strategy is executed and the adaptation execution feedback data is obtained. Based on the adaptation execution feedback data, the multi-source collaborative prediction model is parameter optimized to trigger the closed-loop adaptation operation, forming a complete feedback optimization mechanism, so that the system can continuously self-learn and improve, and adapt to the ever-changing operating environment. As a result, the complex and changeable operation problems of the multi-source energy system are effectively solved, the energy utilization efficiency is improved, and the stability and reliability of the system are enhanced.

[0036] In a possible implementation, step S120 includes: Step S121, performing time series fluctuation analysis on the photovoltaic power generation data sequence, wind power generation data sequence and energy storage system output power data sequence in the energy supply characteristics, extracting the instantaneous power change rate of the photovoltaic power generation data sequence as the first supply fluctuation sub-feature, extracting the power standard deviation of the wind power generation data sequence as the second supply fluctuation sub-feature, and extracting the charge and discharge switching frequency of the energy storage system output power data sequence as the third supply fluctuation sub-feature.

[0037] In this embodiment, for the photovoltaic power generation data sequence, the power value is recorded every 15 minutes starting from 8 o'clock in the morning. The power at 8 o'clock is 300 kilowatts, and the power at 8:15 is 350 kilowatts. To calculate the instantaneous power change rate, the power at 8 o'clock is subtracted from the power at 8 o'clock, that is, 350 kilowatts minus 300 kilowatts, and the power change is 50 kilowatts. Then the power change is divided by the power at 8 o'clock, that is, 50 kilowatts divided by 300 kilowatts, which is approximately equal to 0.167, which is converted into a percentage of 16.7%. This 16.7% is the instantaneous power change rate of the photovoltaic power generation data sequence during this period, which is used as the first supply fluctuation sub-feature.

[0038] For the wind power data series, on a certain day, the wind power is recorded once every hour, and the data are 750 kW, 800 kW, 780 kW, 820 kW, and 790 kW respectively. First, calculate the average value of this set of data, add the above data, 750+800+780+820+790=3940 kW, and then divide it by the number of data 5, and the average value is 3940 kW divided by 5, which is 788 kW. Then calculate the square of the difference between each data and the average value, (750-788)²=(-38)²=1444, (800-788)²=12²=144, (780-788)²=(-8)²=64, (820-788)²=32²=1024, (790-788)²=2²=4. Next, we calculate the average of the square values, (1444+144+64+1024+4)÷5=2680÷5=536. Finally, we take the square root of the average value and get the power standard deviation of about 23.15 kilowatts, which is the second supply fluctuation sub-characteristic.

[0039] For the energy storage system output power data sequence, the number of charge and discharge switching of the energy storage system can be counted in one day. For example, from 7 am to 7 pm, the energy storage system is recorded to have 12 charge and discharge switchings. Then the charge and discharge switching frequency is 12 times / 12 hours, that is, 1 time / hour, which is the third supply fluctuation sub-feature.

[0040] Step S122, performing demand mutation detection processing on the industrial load demand data sequence, commercial load demand data sequence and residential load demand data sequence in the load demand characteristics, determining the mutation time point set of the industrial load demand data sequence as the first demand fluctuation sub-feature, determining the load fluctuation amplitude of the commercial load demand data sequence as the second demand fluctuation sub-feature, and determining the periodic fluctuation pattern of the residential load demand data sequence as the third demand fluctuation sub-feature.

[0041] In this embodiment, for an industrial load demand data sequence, for example, at 10 a.m., a factory suddenly increases its load demand from 1,000 kilowatts to 1,500 kilowatts due to the start of a new production line. The time point of 10 a.m. is recorded. Then, at 2 p.m., due to the shutdown of some equipment, the load demand drops from 1,400 kilowatts to 1,000 kilowatts. The time point of 2 p.m. is also recorded. Thus, the above mutation time points can be grouped together as the first demand fluctuation sub-feature.

[0042] For the commercial load demand data series, the highest and lowest values ​​of the commercial load demand are recorded every day in a week. For example, the lowest value on Monday is 600 kW and the highest value is 900 kW; the lowest value on Tuesday is 580 kW and the highest value is 880 kW, etc. By comparing the above data, the maximum fluctuation range of the commercial load demand in a week is determined, that is, the highest value minus the lowest value, for example, 900 kW minus 580 kW equals 320 kW, and this 320 kW is the second demand fluctuation sub-feature.

[0043] For the residential load demand data series, we can analyze the residential load demand data for each day of a month. For example, we can find that the peak electricity consumption period is from 7 to 10 pm every day, and the load demand gradually increases from 500 kilowatts at 7 pm to 800 kilowatts at 10 pm, and then gradually decreases. The electricity consumption period is from 2 to 4 am, and the load demand remains at around 300 kilowatts. This cyclical fluctuation pattern that repeats every day serves as the third demand fluctuation sub-feature.

[0044] Step S123, calculating the stability index of the grid frequency deviation data sequence and the voltage deviation data sequence in the system operation status characteristics, obtaining the cumulative offset of the grid frequency deviation as the first stability deviation sub-characteristic, and obtaining the instantaneous fluctuation extreme value of the voltage deviation as the second stability deviation sub-characteristic.

[0045] In this embodiment, for the power grid frequency deviation data sequence, the power grid frequency deviation value can be recorded every half an hour. For example, in one day, the recorded data is -0.05Hz, 0Hz, 0.03Hz, -0.02Hz, etc. By accumulating and adding the above data, -0.05+0+0.03+(-0.02)=-0.04Hz, this -0.04Hz is the cumulative offset of the power grid frequency deviation, which is used as the first stability deviation sub-feature.

[0046] For the voltage deviation data sequence, the voltage deviation value is monitored in real time on a certain day, and the maximum value recorded is +6V and the minimum value is -4V, where +6V is the instantaneous fluctuation extreme value of the voltage deviation, which serves as the second stability deviation sub-feature.

[0047] Step S124, combining the first supply fluctuation sub-feature, the second supply fluctuation sub-feature, the third supply fluctuation sub-feature, the first demand fluctuation sub-feature, the second demand fluctuation sub-feature, the third demand fluctuation sub-feature, the first stability deviation sub-feature and the second stability deviation sub-feature to generate a dynamic supply and demand deviation feature set.

[0048] In this embodiment, the first supply fluctuation sub-feature (the instantaneous power change rate of the photovoltaic power data sequence is 16.7%), the second supply fluctuation sub-feature (the power standard deviation of the wind power data sequence is 23.15 kilowatts), the third supply fluctuation sub-feature (the charge and discharge switching frequency of the energy storage system output power data sequence is 1 time / hour), the first demand fluctuation sub-feature (the mutation time point set of the industrial load demand data sequence, such as 10 am, 2 pm, etc.), the second demand fluctuation sub-feature (the load fluctuation amplitude of the commercial load demand data sequence is 320 kilowatts), the third demand fluctuation sub-feature (the periodic fluctuation mode of the residential load demand data sequence, with a peak from 7 pm to 10 pm and a trough from 2 am to 4 am), the first stability deviation sub-feature (the cumulative offset of the grid frequency deviation is -0.04 Hz) and the second stability deviation sub-feature (the instantaneous fluctuation extreme value of the voltage deviation is +6V) can be combined to generate a dynamic supply and demand deviation feature set.

[0049] In a possible implementation, step S130 includes: Step S131, input the dynamic supply and demand deviation feature set into the encoder layer of the multi-source collaborative prediction model, perform multi-dimensional feature fusion processing on the dynamic supply and demand deviation feature set through the encoder layer, and generate a fusion feature vector including supply and demand coupling features and system stability correlation features.

[0050] For example, in the energy supply fluctuation characteristics, the instantaneous power change rate of the photovoltaic power data series is 15%, the power standard deviation of the wind power data series is 25 kilowatts, and the charge and discharge switching frequency of the energy storage system output power data series is 1.5 times / hour; in the load demand fluctuation characteristics, the mutation time point set of the industrial load demand data series is 11 a.m. and 3 p.m., the load fluctuation amplitude of the commercial load demand data series is 350 kilowatts, and the periodic fluctuation pattern of the residential load demand data series is 7 to 10 p.m. every night as the peak electricity consumption, and 2 to 4 a.m. as the trough electricity consumption; in the system stability deviation characteristics, the cumulative offset of the grid frequency deviation is -0.06Hz, and the instantaneous fluctuation extreme value of the voltage deviation is +8V.

[0051] The encoder layer performs multi-dimensional feature fusion processing on the above dynamic supply and demand deviation features. For example, the relationship between energy supply and load demand, as well as the impact of system stability on them can be comprehensively considered. For example, the instantaneous power change rate of photovoltaic power generation is combined with the mutation time point of industrial load demand to analyze whether there is a potential correlation between the two. It may be found that when photovoltaic power generation power rises rapidly, the startup of certain industrial equipment will cause a sudden change in industrial load demand. Through this correlation analysis, partial information containing supply and demand coupling characteristics is generated. At the same time, the cumulative offset of the grid frequency deviation is associated with the standard deviation of wind power generation power, considering the impact of system stability on energy supply fluctuations, thereby generating system stability correlation features. Therefore, the above fused information together constitutes a fused feature vector.

[0052] Step S132, calling the time series prediction layer of the multi-source collaborative prediction model, performing time series expansion processing on the fused feature vector, and generating energy supply trend prediction features and load demand trend prediction features within a future preset time window. The energy supply trend prediction features include photovoltaic power generation power prediction curves, wind power generation power prediction curves, and energy storage system dispatchable capacity prediction curves, and the load demand trend prediction features include industrial load demand prediction curves, commercial load demand prediction curves, and residential load demand prediction curves.

[0053] In this embodiment, the next 24 hours are used as a preset time window to generate energy supply trend prediction characteristics and load demand trend prediction characteristics.

[0054] In terms of the characteristics of energy supply trend prediction, for the photovoltaic power prediction curve, by analyzing historical data and current weather conditions, time and other factors, it is predicted that in the next 24 hours, the photovoltaic power will change with the change of sunshine intensity. For example, there will be a certain power output at around 7 o'clock in the morning, and as the sunshine increases, the power will gradually increase. It is expected to reach a peak of 1,300 kilowatts at around 12 noon, and then as the sunshine decreases, the power will gradually decrease, and basically stop generating electricity at around 7 o'clock in the evening.

[0055] The wind power generation prediction curve takes into account factors such as the current wind speed, wind direction, and the operating status of wind power generation equipment. It is predicted that in the next 24 hours, due to changes in meteorological conditions, the wind speed will increase from 9 to 11 p.m., and the wind power generation will gradually increase from 800 kilowatts at 9 p.m. to 950 kilowatts at 11 p.m., and then gradually decrease as the wind speed decreases.

[0056] The dispatchable capacity prediction curve of the energy storage system is generated based on the current state of charge of the energy storage system, the historical data of charging and discharging, and the forecast of future energy supply and demand. Assume that the current state of charge of the energy storage system is 55% and the dispatchable capacity is 450 kWh. Taking into account the changes in future energy supply and load demand, it is predicted that in the next 24 hours, the energy storage system will be charged when there is excess photovoltaic power generation and discharged when the load demand is peak and the energy supply is insufficient. It is expected that from 4 to 6 pm, the energy storage system will charge at a power of 120 kilowatts, so that the dispatchable capacity will reach 580 kWh at 6 pm; from 8 to 10 pm, the energy storage system will discharge at a power of 150 kilowatts, and the dispatchable capacity will drop to 330 kWh at 10 pm.

[0057] In terms of load demand trend prediction characteristics, the industrial load demand prediction curve is generated based on the production plans and equipment operation arrangements of each factory in the park. For example, a large factory plans to carry out large-scale production activities from 2 pm to 6 pm. It is expected that the industrial load demand will gradually increase from 1,600 kilowatts at 2 pm to 1,900 kilowatts at 4 pm. After that, as some production links are completed, the load demand will drop to 1,700 kilowatts at 6 pm.

[0058] The commercial load demand forecast curve combines the business hours of shops in the commercial area and the usage patterns of electrical equipment. It is expected that from 7 to 9 p.m., as the customer flow in shopping malls, restaurants and other commercial places increases, the commercial load demand will gradually increase from 750 kilowatts at 7 p.m. to 900 kilowatts at 9 p.m., and then gradually decrease as shops close one after another.

[0059] The residential load demand forecast curve is based on the residents' daily electricity consumption habits. 7 to 10 p.m. is the peak period for residents' electricity consumption. It is expected that the residential load demand will gradually increase from 700 kilowatts at 7 p.m. to 950 kilowatts at 9 p.m., and then drop to 600 kilowatts at 11 p.m. as residents fall asleep.

[0060] Step S133, calculating the supply-demand matching degree between the energy supply trend prediction characteristics and the load demand trend prediction characteristics, and generating a multi-source collaborative prediction result including a supply-demand gap prediction value and a supply-demand balance time point.

[0061] For example, at 9 p.m., the photovoltaic power generation is basically 0 kW, the wind power generation is 900 kW, the power output corresponding to the dispatchable capacity of the energy storage system is 120 kW, and the total energy supply is 900+120=1020 kW; while the industrial load demand is 1750 kW, the commercial load demand is 850 kW, and the residential load demand is 900 kW, so the total load demand is 1750+850+900=3500 kW. Subtracting the two, the supply-demand gap forecast value is 3500-1020=2480 kW.

[0062] On this basis, we continue to analyze the energy supply and load demand at each time point to find the time point when supply and demand are balanced. After calculation and analysis, we found that around 3 a.m., due to the significant reduction in industrial load demand, commercial load demand is basically 0, and residential load demand is also at a low level. At this time, wind power generation still has a certain power output, and the energy storage system also has a certain dispatchable capacity. Energy supply and load demand are balanced. This 3 a.m. is the time point of supply and demand balance. Combining information such as the supply and demand gap forecast value and the supply and demand balance time point, a multi-source collaborative forecast result containing energy supply trend forecast characteristics and load demand trend forecast characteristics is generated.

[0063] In a possible implementation, step S140 includes: Step S141, matching the supply-demand gap forecast value with a plurality of preset supply-demand balance threshold intervals to determine the target threshold interval to which the current supply-demand gap belongs.

[0064] Assume that the preset threshold intervals are: [-3000 kW, -2000 kW], [-2000 kW, -1000 kW], [-1000 kW, 1000 kW], [1000 kW, 2000 kW], [2000 kW, 3000 kW], etc. The supply-demand gap forecast value at 9 pm is 2480 kW, which belongs to the target threshold interval of [2000 kW, 3000 kW].

[0065] Step S142, according to the strategy generation rule corresponding to the target threshold interval, a set of dispatchable energy supply sources is screened from the energy supply characteristics, and the maximum charge and discharge power of the energy storage system is determined based on the current state of charge and dispatchable capacity in the energy storage charge and discharge strategy.

[0066] For example, at 9 o'clock in the evening, the photovoltaic power generation system basically stops working and cannot be dispatched; although the wind power generation system has a certain power output, the room for improvement is limited; and the energy storage system still has a certain dispatchable capacity. Therefore, the set of dispatchable energy supply sources is mainly the energy storage system.

[0067] Assuming that the current state of charge of the energy storage system is 40% and the dispatchable capacity is 300 kWh, taking into account factors such as the safe operation and charging and discharging efficiency of the energy storage system, after a series of calculations and evaluations (for example, combining the rated power of the energy storage system, historical charging and discharging data, and current environmental factors such as temperature), it is determined that the maximum charging and discharging power of the energy storage system is 180 kW.

[0068] Step S143, according to the preset industry priority weights in the load priority allocation strategy, the industrial load demand data sequence, the commercial load demand data sequence and the residential load demand data sequence are sorted by load reduction priority.

[0069] For example, the preset industrial load has the highest priority due to the production continuity requirement; the commercial load is second; and the residential load has a relatively low priority. Therefore, when it is necessary to reduce the load, the residential load is considered first, followed by the commercial load, and the stable operation of the industrial load is guaranteed as much as possible.

[0070] Step S144, generating a multi-source collaborative adaptation strategy set including an energy output adjustment strategy, an energy storage charging and discharging strategy, and a load priority allocation strategy based on the dispatchable energy supply source set, the maximum charging and discharging power, and the load reduction priority ranking.

[0071] For example, based on the dispatchable energy supply source set (energy storage system), maximum charge and discharge power (180 kilowatts) and load reduction priority ranking, a multi-source coordinated adaptation strategy set including energy output adjustment strategy, energy storage charge and discharge strategy and load priority allocation strategy is generated. In terms of energy output adjustment strategy, since the dispatchable energy at this time is mainly the energy storage system, the strategy is to let the energy storage system discharge at a maximum charge and discharge power of 180 kilowatts to increase energy supply. The energy storage charge and discharge strategy clearly states that in the next period of time, it will continue to discharge at a power of 180 kilowatts until the charge state of the energy storage system reaches the safe lower limit or the supply and demand gap is effectively alleviated. The load priority allocation strategy is to give priority to appropriate reduction of residential loads, such as reducing 200 kilowatts of residential loads during the peak period of electricity consumption from 9 to 10 pm; if there is still a supply and demand gap, consider reducing some commercial loads, such as reducing 150 kilowatts of commercial loads, to ensure that industrial loads can operate normally and ensure that production activities in the park are not greatly affected. Through the formulation and implementation of such a series of strategies, a complete set of multi-source coordinated adaptation strategies has been formed to cope with the real-time supply and demand imbalance in the multi-source energy system and ensure the stable supply and efficient use of energy.

[0072] In a possible implementation, step S150 includes: Step S151 : sending a power adjustment instruction in the energy output adjustment strategy to a photovoltaic inverter controller of the multi-source energy system to obtain actual adjustment response data of photovoltaic power generation.

[0073] In this embodiment, since the photovoltaic power generation system has basically stopped working at this time, although the power adjustment instruction is sent, the photovoltaic power generation power cannot be effectively adjusted. The actual adjustment response data of the photovoltaic power generation power finally obtained shows that the photovoltaic power generation power remains at the current low level, for example, 50 kilowatts, and has not changed due to the instruction.

[0074] Step S152, sending the charge and discharge rate instruction in the energy storage charge and discharge strategy to the bidirectional converter of the energy storage system to obtain actual charge and discharge power data of the energy storage system.

[0075] In this embodiment, after receiving the charge and discharge speed instruction, the energy storage system starts to operate at the set charge and discharge power of 180 kW. In the next hour, the actual charge and discharge power data of the energy storage system is obtained through real-time monitoring equipment. After recording and calculation, the energy storage system actually discharges at an average power of 175 kW during this hour. The specific calculation process is: the discharge power of the energy storage system is recorded every 10 minutes, which are 170 kW, 172 kW, 178 kW, 176 kW, 174 kW, and 175 kW respectively. The above data are added to obtain a total of 1025 kW, and then divided by the number of records 6, and the average discharge power is 1025 kW ÷ 6 ≈ 175 kW.

[0076] Step S153: sending the load reduction instruction in the load priority allocation strategy to the load management terminal to obtain actual load reduction amount data.

[0077] For example, according to the load priority allocation strategy, the residential load is reduced first. After receiving the instruction, the load management terminal regulates the residential load. According to statistics, the residential load is actually reduced by 180 kilowatts. During the reduction process, the changes in electricity consumption of each residential user are recorded, and the actual load reduction amount is obtained by summarizing the above data.

[0078] Step S154, combining the actual adjustment response data, the actual charging and discharging power data and the actual load reduction data to generate adaptation execution feedback data, and calculating a strategy execution deviation index between the adaptation execution feedback data and the multi-source collaborative prediction result.

[0079] For example, the actual adjustment response data (the photovoltaic power generation power is maintained at 50 kilowatts), the actual charging and discharging power data (the energy storage system discharges at an average power of 175 kilowatts), and the actual load reduction data (the actual reduction of residential load is 180 kilowatts) are combined to generate the adaptive execution feedback data. Next, the strategy execution deviation index between the adaptive execution feedback data and the multi-source collaborative prediction results is calculated. In the multi-source collaborative prediction results, the energy storage system is expected to discharge at a power of 180 kilowatts, the actual average power is 175 kilowatts, and the deviation is 180-175=5 kilowatts; the residential load is expected to be reduced by 200 kilowatts, and the actual reduction is 180 kilowatts, with a deviation of 200-180=20 kilowatts. Taking into account the deviations in these two aspects, the strategy execution deviation index is calculated through a specific calculation method (for example, after assigning weights according to the importance of different factors, assuming that the energy storage system discharge power deviation weight is 0.6 and the residential load reduction deviation weight is 0.4). The specific calculation is: the impact value of the energy storage system discharge power deviation is 5 kW × 0.6 = 3, and the impact value of the residential load reduction deviation is 20 kW × 0.4 = 8. The sum of the two gives a strategy execution deviation index of 3+8=11.

[0080] Step S155, inputting the strategy execution deviation index into the feedback learning layer of the multi-source collaborative prediction model, and dynamically adjusting the feature fusion weight matrix of the encoder layer and the time window expansion parameter of the timing prediction layer through the feedback learning layer.

[0081] For example, the strategy execution deviation indicator 11 can be input into the feedback learning layer of the multi-source collaborative prediction model. The feedback learning layer starts to dynamically adjust the feature fusion weight matrix of the encoder layer and the time window expansion parameter of the time series prediction layer.

[0082] For example, through analysis, it was found that in the previous feature fusion process, some feature weights related to the energy storage system and load demand were not set accurately, resulting in deviations between the prediction results and the actual execution. For example, the feature of the charge and discharge switching frequency of the energy storage system has a low weight during fusion, which does not fully reflect its impact on the overall supply and demand situation. The feature fusion weight matrix is ​​updated based on the gradient descent algorithm. The specific operation is to gradually adjust the weights of each feature according to the deviation index and the pre-set learning rate (for example, the learning rate is 0.01). Assuming that the original weight of the charge and discharge switching frequency feature of the energy storage system is 0.2, after calculation, according to the gradient descent algorithm, the new weight is 0.2+0.01× (the impact value of the deviation on the feature weight). Through complex calculations (involving the algorithm and data relationship within the model), the new weight is 0.23. Similar adjustments are made to other feature weights to complete the update of the feature fusion weight matrix.

[0083] For the time window expansion parameters of the time series prediction layer, they are adjusted according to the time dimension deviation component in the strategy execution deviation index. Analysis shows that there are certain deviations in the time dimension when predicting the charging and discharging power and load reduction of the energy storage system, indicating that the time window expansion parameter settings are not reasonable. The time dimension deviation component in the strategy execution deviation index shows that the time accuracy of the prediction needs to be improved. The current time window length is 6 hours and the sliding step is 1 hour. Due to the increase in the time dimension deviation component, the window length is shortened and the sliding step is increased according to the rules. The window length is shortened to 4 hours and the sliding step is increased to 2 hours. After this adjustment, the time series prediction layer can more accurately capture the changing trends of energy supply and load demand in the time dimension.

[0084] Step S156, regenerates the updated multi-source collaborative prediction result according to the adjusted feature fusion weight matrix and time window expansion parameter, and performs secondary dynamic adaptation strategy matching processing on the updated multi-source collaborative prediction result and the real-time operation data set to generate an optimized multi-source collaborative adaptation strategy set.

[0085] In this embodiment, the updated multi-source collaborative forecasting results are more accurate in terms of energy supply trend forecasting and load demand trend forecasting. For example, the forecast of the dispatchable capacity of the energy storage system is closer to the actual situation, and the forecast of industrial, commercial and residential load demand in different time periods is also more accurate.

[0086] Assume that after calculation and analysis, the new supply-demand gap forecast value is in another target threshold interval [-1000 kW, -500 kW]. According to the strategy generation rules corresponding to the new target threshold interval, the set of dispatchable energy supply sources is re-screened from the energy supply characteristics. At this time, it is found that in addition to the energy storage system, the wind power generation system also has a certain dispatchable space. Based on the current state of charge and dispatchable capacity in the energy storage charging and discharging strategy, the maximum charge and discharge power of the energy storage system is re-determined. After calculation (considering factors such as the current state of charge of the energy storage system, charging and discharging efficiency, and remaining life), the maximum charge and discharge power of the energy storage system is determined to be 150 kW. At the same time, according to the preset industry priority weights in the load priority allocation strategy, the industrial load demand data series, commercial load demand data series, and residential load demand data series are again prioritized for load reduction. Since the supply-demand gap is relatively small at this time, the load reduction strategy may be adjusted, such as appropriately reducing the amount of residential load reduction and increasing the regulation of commercial loads.

[0087] Then, based on the re-screened dispatchable energy supply source set (energy storage system and wind power generation system), the newly determined maximum charging and discharging power (150 kilowatts for energy storage system) and the re-ordered load reduction priority, an optimized multi-source coordinated adaptation strategy set is generated. The energy output adjustment strategy is to appropriately increase the output power of the wind power generation system, while controlling the energy storage system to charge and discharge at a power of 150 kilowatts; the energy storage charging and discharging strategy clearly states that in the next period of time, according to the energy supply and demand situation, the charging and discharging of the energy storage system should be reasonably arranged; the load priority allocation strategy is to give priority to reducing a certain amount of commercial load, such as reducing 100 kilowatts of commercial load, and appropriately reduce the reduction of residential load, such as reducing 100 kilowatts of residential load, to ensure the stable operation of industrial load.

[0088] Step S157 , repeatedly executing the optimized multi-source collaborative adaptation strategy set and iteratively updating the parameters of the multi-source collaborative prediction model until the strategy execution deviation index is lower than a preset deviation threshold.

[0089] In this embodiment, after each execution of the multi-source collaborative adaptation strategy, new adaptation execution feedback data is obtained and a new strategy execution deviation index is calculated. The new strategy execution deviation index is input into the feedback learning layer of the multi-source collaborative prediction model again, and the feature fusion weight matrix of the encoder layer and the time window expansion parameter of the timing prediction layer are further adjusted. This process is repeated until the strategy execution deviation index is lower than the preset deviation threshold. For example, the preset deviation threshold is 5. After multiple iterative adjustments, the strategy execution deviation index gradually decreases and eventually decreases to 4, meeting the requirement of being lower than the preset deviation threshold. At this point, under the action of the optimized strategy, the multi-source energy system can operate more stably and efficiently, and realize the reasonable allocation of energy and the balance of supply and demand.

[0090] In a possible implementation, step S155 includes: Step S1551, calculating the gradient change of the feature fusion weight matrix of the encoder layer according to the strategy execution deviation index, and updating the feature fusion weight matrix based on the gradient descent algorithm.

[0091] In this embodiment, the strategy execution deviation index is obtained by comprehensively analyzing the deviations between the actual execution conditions and the predicted results in various aspects. For example, after executing the multi-source coordinated adaptation strategy once, the strategy execution deviation index is calculated, which reflects the deviations of multiple factors on the energy supply side and the load demand side, including the differences between the actual values ​​and the predicted values ​​of photovoltaic power generation, wind power generation, energy storage system charging and discharging, industrial load, commercial load, and residential load.

[0092] When calculating the gradient change of the feature fusion weight matrix of the encoder layer, it is necessary to analyze the contribution of each feature to the deviation. Taking the instantaneous power change rate of the photovoltaic power data sequence in the energy supply feature as an example, if the change of the actual photovoltaic power is different from the predicted situation, it is necessary to see whether the weight setting of the feature in the feature fusion weight matrix is ​​reasonable. Assume that in the previous prediction, the weight of this feature is 0.3. By analyzing the strategy execution deviation index, it is found that the prediction deviation of photovoltaic power has a certain impact on the overall deviation. In the specific calculation, the change of the actual photovoltaic power is compared with the predicted situation to obtain a deviation value related to the feature. For example, the actual photovoltaic power changes faster than the predicted value in a certain period of time, resulting in a certain deviation. Through a complex calculation process (involving the correlation analysis of the deviation value with the overall strategy execution deviation index, and considering the role of the feature in the entire model and other factors), the gradient change of the feature weight is 0.05.

[0093] A similar analysis is performed for the power standard deviation feature of the wind power data series. For example, assuming that its original weight is 0.25, after comparing the fluctuation of actual wind power generation with the predicted results, the gradient change of the feature weight is calculated to be -0.03. This means that in the current situation, the weight of this feature in the fusion process may be too high and needs to be appropriately reduced.

[0094] For the charge-discharge switching frequency feature of the energy storage system output power data sequence, the original weight was 0.2. After analyzing the difference between the actual charge-discharge switching situation and the prediction, the gradient change was calculated to be 0.04. This means that the impact of this feature on the prediction result is greater than the previously set weight, and the weight needs to be increased.

[0095] For the feature of the mutation time point set of the industrial load demand data sequence in the load demand feature, the weight is 0.15. By comparing the difference between the actual industrial load mutation time and the predicted time, the gradient change is calculated to be 0.02.

[0096] The load fluctuation amplitude characteristics of the commercial load demand data series, with a weight of 0.1, calculate the gradient change to be -0.01.

[0097] The periodic fluctuation pattern characteristics of the residential load demand data series, with a weight of 0.1, calculate the gradient change to be 0.01.

[0098] The cumulative offset feature of the grid frequency deviation in the system stability deviation feature has a weight of 0.05, and the calculated gradient change is -0.005.

[0099] The instantaneous fluctuation extreme value feature of the voltage deviation has a weight of 0.05, and the calculated gradient change is 0.005.

[0100] The feature fusion weight matrix is ​​updated based on the gradient descent algorithm. Taking the instantaneous power change rate feature weight of the photovoltaic power generation data sequence as an example, the updated weight is the original weight 0.3 plus the gradient change 0.05, that is, 0.3+0.05=0.35.

[0101] The power standard deviation feature weight of the wind power data series is updated to 0.25+(-0.03)=0.22.

[0102] The characteristic weight of the charge-discharge switching frequency of the energy storage system output power data sequence is updated to 0.2+0.04=0.24.

[0103] The characteristic weight of the mutation time point set of the industrial load demand data series is updated to 0.15+0.02=0.17.

[0104] The load fluctuation amplitude characteristic weight of the commercial load demand data series is updated to 0.1+(-0.01)=0.09.

[0105] The characteristic weight of the periodic fluctuation pattern of the residential load demand data series is updated to 0.1+0.01=0.11.

[0106] The cumulative offset feature weight of the grid frequency deviation is updated to 0.05+(-0.005)=0.045.

[0107] The instantaneous fluctuation extreme value characteristic weight of voltage deviation is updated to 0.05+0.005=0.055.

[0108] This completes the update of the feature fusion weight matrix of the encoder layer, allowing the model to more accurately reflect the actual situation when fusing various features.

[0109] Step S1552: According to the time dimension deviation component in the strategy execution deviation index, adjust the window length and sliding step size in the time window expansion parameters of the time series prediction layer.

[0110] The adjustment of the time window extension parameter satisfies the following conditions: When the time dimension deviation component increases, the window length is shortened and the sliding step size is increased.

[0111] When the time dimension deviation component decreases, the window length is extended and the sliding step size is reduced.

[0112] In this embodiment, the time dimension deviation component in the strategy execution deviation index reflects the degree of deviation between the predicted time and the actual situation on the time axis.

[0113] Assume that after a certain execution of a strategy, the time dimension deviation component of the calculated strategy execution deviation indicator increases. This indicates that the time accuracy of the prediction has decreased, and the time window expansion parameters of the time series prediction layer need to be adjusted. For example, the current time window length is 8 hours and the sliding step is 1 hour. Since the time dimension deviation component increases, shorten the window length and increase the sliding step according to the rules. Shorten the window length to 6 hours and increase the sliding step to 2 hours.

[0114] Specifically, when analyzing the time series data of energy supply and load demand, it was found that when the prediction was made with an 8-hour window length and a 1-hour sliding step length, many short-term change trends were not accurately captured, resulting in a large deviation between the predicted time and the actual situation. Shortening the window length to 6 hours can focus more on short-term changes in energy supply and demand and improve the accuracy of forecasts of recent trends. Increasing the sliding step length to 2 hours can adapt to data changes more quickly and adjust forecast results in a timely manner while ensuring a certain time coverage range.

[0115] On the contrary, if the time dimension deviation component in the strategy execution deviation indicator decreases, it means that the time accuracy of the forecast has improved. For example, if the original time window length is 6 hours and the sliding step is 2 hours, the time dimension deviation component decreases, so the window length is extended and the sliding step is reduced. The window length is extended to 8 hours and the sliding step is reduced to 1 hour. This adjustment is because in this case, a longer window length can better capture the trend of energy supply and demand over a longer time range, while a smaller sliding step can analyze the changes in data more carefully, further improving the accuracy of the forecast.

[0116] By dynamically adjusting the window length and sliding step size in the time window expansion parameters of the time series prediction layer according to the time dimension deviation component in the strategy execution deviation index, the multi-source collaborative prediction model can better adapt to the dynamic changes of the multi-source energy system in the energy park, improve the accuracy and timeliness of the prediction, and ensure the stable and efficient operation of the energy system in the energy park. In the process of continuous repetition, the parameters of the model are continuously optimized, and the deviation between the prediction results and the actual situation is continuously reduced, ultimately achieving the precise regulation and optimized operation of the multi-source energy system.

[0117] In a possible implementation, before step S110, the method further includes a pre-training step of the multi-source collaborative prediction model: Step S210, obtaining a historical energy operation data set, wherein the historical energy operation data set includes historical energy supply characteristics, historical load demand characteristics, and historical system stability characteristics.

[0118] In this embodiment, in terms of historical energy supply characteristics, photovoltaic power generation data for the past year are collected. Photovoltaic power generation is recorded every 15 minutes from sunrise to sunset every day. For example, on a certain day in spring, the photovoltaic power generation is 100 kilowatts at 7 a.m. and rises to 120 kilowatts at 7:15 a.m. As the sunshine increases, the power changes continuously, reaching a peak of 800 kilowatts at 12 noon, and then gradually decreases. The photovoltaic power generation data for each day of the year are sorted in chronological order to form a photovoltaic power generation data record. The same is true for wind power generation data. Wind power generation fluctuates greatly under different seasons and weather conditions. For example, on a windy day in summer, the wind power generation power rises rapidly from 500 kilowatts to 700 kilowatts within 1 hour, and the long-term recorded wind power generation data is integrated. The output power data of the energy storage system records the power and time of each charge and discharge. For example, during a low power consumption period at night, the energy storage system is charged at a power of 150 kilowatts for 3 hours. The above data is fully recorded to constitute the historical energy supply characteristics.

[0119] In terms of historical load demand characteristics, for industrial load data, there are many factories in the park, and the production plans and equipment operation conditions of different factories are different, and the industrial load demand also changes accordingly. A large factory has a production period from 8 am to 5 pm on weekdays. The industrial load demand gradually increases from 1,000 kilowatts at 8 am to 1,500 kilowatts at 11 am, and remains at a high level from 2 pm to 4 pm, and then gradually decreases. Collect the industrial load demand data of each factory in the past year to form an industrial load data record. In terms of commercial load data, the business hours of shops in commercial areas are relatively concentrated, such as 9 am to 9 pm, and the peak hours are from 12 noon to 2 pm and 6 pm to 8 pm. The load demand gradually increases from 300 kilowatts at 9 am to 600 kilowatts at 12 pm, and the changes in commercial loads in a year are recorded. The residential load data shows obvious daily periodicity. The peak time for electricity consumption is from 7 to 10 p.m., rising from 400 kilowatts at 7 a.m. to 600 kilowatts at 9 a.m., and the trough time is from 2 to 5 a.m., maintaining at around 150 kilowatts. The above data are collected to form the historical load demand characteristics.

[0120] In terms of historical system stability characteristics, the grid frequency deviation data records the difference between the actual grid operating frequency and the standard frequency every 30 minutes. For example, in a certain week, the recorded grid frequency deviation data is -0.05Hz, 0Hz, 0.03Hz, etc. The voltage deviation data monitors the deviation between the grid voltage and the rated voltage in real time. For example, in a certain month, the maximum value of the recorded voltage deviation is +8V and the minimum value is -5V. The above data is collated to form the historical system stability characteristics.

[0121] Step S220: standardize the historical energy operation data set to generate a standardized training data set.

[0122] In a possible implementation, step S220 includes: Step S221 , performing maximum and minimum value normalization processing on the photovoltaic power data, wind power data and energy storage system output power data in the historical energy supply characteristics to obtain normalized energy supply characteristics.

[0123] Taking photovoltaic power generation data as an example, in the data of the past year, the maximum power value is 1000 kilowatts and the minimum power value is 50 kilowatts. For the photovoltaic power generation value at a certain moment, such as 300 kilowatts, normalization is performed. The calculation process is to subtract the minimum value of 50 kilowatts from the power value to get 300-50=250 kilowatts, and then divide it by the difference between the maximum and minimum values, that is, 1000-50=950 kilowatts, and the normalized result is 250÷950≈0.263. The same calculation is performed on all photovoltaic power generation data to obtain the normalized photovoltaic power generation data. Wind power generation data and energy storage system output power data are also normalized according to this method, and finally the normalized energy supply characteristics are obtained.

[0124] Step S222, performing Z-score normalization processing on the industrial load data, commercial load data and residential load data in the historical load demand characteristics to obtain standardized load demand characteristics.

[0125] Taking industrial load data as an example, first calculate the average value of industrial load data. Assuming that the total industrial load data collected in a year is 5475000 kilowatts, and the number of records is 35040 times (calculated as 365 days a year and 96 records per day), the average value is 5475000÷35040≈156.25 kilowatts. Then calculate the standard deviation, subtract the average value from each industrial load data value and square it, add the above square values, divide it by the number of data, and finally square it. Assume that the standard deviation is about 200 kilowatts after calculation. For the industrial load value at a certain moment, such as 180 kilowatts, Z-score standardization is performed. The calculation process is to subtract the average value of 156.25 kilowatts from this value to get 180-156.25=23.75 kilowatts, and then divide it by the standard deviation of 200 kilowatts, and the standardized result is 23.75÷200=0.119. By processing all industrial load data in the same way, standardized industrial load data can be obtained. Commercial load data and residential load data are also standardized in this way to obtain standardized load demand characteristics.

[0126] Step S223, performing sliding window mean processing on the grid frequency deviation data and voltage deviation data in the historical system stability characteristics to obtain smoothed system stability characteristics.

[0127] For the power grid frequency deviation data, the sliding window size is set to 10 data points (i.e. 5 hours). For example, starting from the first data point, the first 10 power grid frequency deviation data are -0.05Hz, 0Hz, 0.03Hz, -0.02Hz, 0.01Hz, -0.01Hz, 0Hz, 0.02Hz, -0.03Hz, and 0.04Hz. Add the above data to get -0.05+0+0.03+(-0.02)+0.01+(-0.01)+0+0.02+(-0.03)+0.04=0.02Hz, and then divide it by the number of data 10 to get the sliding window mean of 0.02÷10=0.002Hz. Slide the window backward by one data point, take the second to eleventh data points for the same calculation, and so on, to get a series of sliding window means, which constitute the smoothed power grid frequency deviation data. For the voltage deviation data, the sliding window size is also set to 10 data points, and the calculation is performed according to the above method to obtain the smoothed voltage deviation data, and then the smoothed system stability characteristics are obtained.

[0128] Step S224, aligning the normalized energy supply characteristics, the standardized load demand characteristics and the smoothed system stability characteristics in time series and merging them to generate a standardized training data set.

[0129] For example, the normalized photovoltaic power generation, wind power generation, and energy storage system output power data at the same time point are combined with the standardized industrial load, commercial load, and residential load data, as well as the smoothed grid frequency deviation and voltage deviation data to form a complete data record. All of the above data records are sorted in chronological order to eventually generate a standardized training data set.

[0130] Step S230: constructing an initial multi-source collaborative prediction model, wherein the initial multi-source collaborative prediction model includes a feature encoder, a timing predictor, and a feedback corrector.

[0131] In this embodiment, the feature encoder is responsible for feature fusion of the input data, integrating different types of energy supply, load demand and system stability features. The time series predictor performs time series analysis based on the fused features to predict future energy supply and load demand trends. The feedback corrector adjusts and optimizes the initial multi-source collaborative prediction model based on the difference between the prediction results and the actual data.

[0132] Step S240, input the standardized training data set into the initial multi-source collaborative prediction model for iterative training until the error rate between the prediction result output by the initial multi-source collaborative prediction model and the actual historical operation data is lower than a preset threshold, thereby obtaining a pre-trained multi-source collaborative prediction model.

[0133] And, step S240 includes: Step S241, dividing the standardized training data set into a training set and a validation set, wherein the training set is used to update the parameters of the initial multi-source collaborative prediction model, and the validation set is used to evaluate the generalization ability of the model.

[0134] Assuming a 7:3 ratio, 70% of the data is randomly selected from the standardized training data set as the training set to update the parameters of the initial multi-source collaborative prediction model; the remaining 30% is used as the validation set to evaluate the generalization ability of the initial multi-source collaborative prediction model.

[0135] Step S242: In each iterative training process, the feature encoder is called to perform feature fusion on the data in the training set to generate a training fusion feature vector.

[0136] For example, for a certain set of training data, the feature encoder will fuse the normalized photovoltaic power data, wind power data, and energy storage system output power data with the standardized industrial load data, commercial load data, residential load data, and smoothed grid frequency deviation data and voltage deviation data. For example, weights can be assigned to each feature based on the importance of different features, and then the above features are combined according to the weights to generate a training fusion feature vector.

[0137] Step S243: calling the time series predictor to generate a training prediction result based on the training fusion feature vector.

[0138] In this embodiment, the time series predictor analyzes the time series information in the training fusion feature vector, combines the changing trend of historical data, and predicts the energy supply and load demand in the future. For example, it predicts the photovoltaic power generation, wind power generation, energy storage system dispatchable capacity, and industrial load, commercial load, and residential load demand in the next hour.

[0139] Step S244, calling the feedback corrector to calculate the model loss function according to the difference between the training prediction result and the actual historical operation data, and updating the parameters of the feature encoder and the timing predictor based on the back propagation algorithm.

[0140] Assume that the predicted photovoltaic power generation is 600 kilowatts, while the photovoltaic power generation in the actual historical operation data is 580 kilowatts. The difference between the two is 600-580=20 kilowatts. The differences between all predicted values ​​and the actual values ​​are calculated comprehensively, such as by using methods such as mean square error, squaring the above difference values, summing them up, and averaging them. For example, suppose that after calculation, the sum of the mean square errors between all energy supply and load demand forecast values ​​and the actual values ​​is 100 (this is just an example to illustrate the calculation concept, and the actual calculation is more complicated). This value is the value of the model loss function.

[0141] Step S245 , when the error rate on the validation set does not decrease after multiple consecutive iterations, terminate the training and save the current model parameters as a pre-trained multi-source collaborative prediction model.

[0142] In this embodiment, the back propagation algorithm calculates the contribution of each parameter to the loss from back to front based on the value of the model loss function, and then adjusts the parameters based on the contribution. For example, if it is found that the weight setting of a feature in the feature encoder causes a large prediction error, the weight will be adjusted based on the calculation results of the back propagation algorithm. Assuming that the original weight of a feature is 0.3, after calculation and adjustment, the new weight may become 0.35. Such adjustments are made to all parameters in the feature encoder and the timing predictor to reduce the value of the model loss function and improve the prediction accuracy of the model.

[0143] During the training process, the above steps are repeated continuously, and the model is optimized in each iteration. At the same time, the validation set is used to evaluate the generalization ability of the model. When the error rate on the validation set does not decrease for multiple consecutive iterations, it means that the model has reached a relatively stable state, and further training may not bring significant improvement. At this point, terminate the training and save the current model parameters as a pre-trained multi-source collaborative prediction model.

[0144] Figure 2 The schematic diagram shows exemplary hardware and software components of a multi-source energy debugging service system 100 that can implement the concept of the present application according to some embodiments of the present application. For example, the processor 120 can be used in the multi-source energy debugging service system 100 and used to perform the functions in the present application.

[0145] The multi-source energy debugging service system 100 can be a general-purpose server or a special-purpose server, both of which can be used to implement the rebalancing dynamic adaptation method for a multi-source system of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0146] For example, the multi-source energy debugging service system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and storage media 140 in different forms, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the multi-source energy debugging service system 100 may also include program instructions stored in ROM, RAM, or other types of non-temporary storage media, or any combination thereof. The method of the present application can be implemented according to the above program instructions. The multi-source energy debugging service system 100 also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0147] For ease of explanation, only one processor is described in the multi-source energy debugging service system 100. However, it should be noted that the multi-source energy debugging service system 100 in the present application may also include multiple processors, so the steps performed by one processor described in the present application may also be performed jointly or individually by multiple processors. For example, if the processor of the multi-source energy debugging service system 100 executes step A and step B, it should be understood that step A and step B may also be performed jointly by two different processors or individually in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor execute steps A and B together.

[0148] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the rebalancing dynamic adaptation method applied to a multi-source system as described above is implemented.

[0149] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, various features are sometimes combined into one embodiment, drawing or description thereof.

Claims

1. A rebalancing dynamic adaptation method applied to a multi-source system, characterized in that: The method comprises: Collecting a real-time operating data set of a multi-source energy system, wherein the real-time operating data set includes energy supply characteristics, load demand characteristics, and system operating status characteristics; Performing dynamic supply and demand deviation feature extraction processing on the real-time operation data set to generate a dynamic supply and demand deviation feature set including supply fluctuation features, demand fluctuation features and system stability deviation; Calling a pre-trained multi-source collaborative prediction model to generate a multi-source collaborative prediction result based on the dynamic supply and demand deviation feature set, wherein the multi-source collaborative prediction result includes energy supply trend prediction features and load demand trend prediction features; Perform dynamic adaptation strategy matching processing on the multi-source collaborative prediction result and the real-time operation data set to generate a multi-source collaborative adaptation strategy set, wherein the multi-source collaborative adaptation strategy set includes an energy output adjustment strategy, an energy storage charging and discharging strategy, and a load priority allocation strategy; An adaptation strategy in the multi-source collaborative adaptation strategy set is executed and adaptation execution feedback data is obtained, and parameters of the multi-source collaborative prediction model are optimized based on the adaptation execution feedback data to trigger a closed-loop adaptation operation.

2. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 1, characterized in that: The step of extracting dynamic supply and demand deviation features from the real-time operation data set to generate a dynamic supply and demand deviation feature set including supply fluctuation features, demand fluctuation features and system stability deviation includes: Performing time series fluctuation analysis on the photovoltaic power generation data sequence, wind power generation data sequence and energy storage system output power data sequence in the energy supply characteristics, extracting the instantaneous power change rate of the photovoltaic power generation data sequence as the first supply fluctuation sub-feature, extracting the power standard deviation of the wind power generation data sequence as the second supply fluctuation sub-feature, and extracting the charge and discharge switching frequency of the energy storage system output power data sequence as the third supply fluctuation sub-feature; Performing demand mutation detection processing on the industrial load demand data sequence, the commercial load demand data sequence and the residential load demand data sequence in the load demand characteristics, determining the mutation time point set of the industrial load demand data sequence as the first demand fluctuation sub-characteristic, determining the load fluctuation amplitude of the commercial load demand data sequence as the second demand fluctuation sub-characteristic, and determining the periodic fluctuation mode of the residential load demand data sequence as the third demand fluctuation sub-characteristic; Calculating the stability index of the grid frequency deviation data sequence and the voltage deviation data sequence in the system operation state characteristics, obtaining the cumulative offset of the grid frequency deviation as the first stability deviation sub-characteristic, and obtaining the instantaneous fluctuation extreme value of the voltage deviation as the second stability deviation sub-characteristic; The first supply fluctuation sub-feature, the second supply fluctuation sub-feature, the third supply fluctuation sub-feature, the first demand fluctuation sub-feature, the second demand fluctuation sub-feature, the third demand fluctuation sub-feature, the first stability deviation sub-feature and the second stability deviation sub-feature are combined to generate a dynamic supply and demand deviation feature set.

3. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 2, characterized in that: The calling of the pre-trained multi-source collaborative prediction model to generate a multi-source collaborative prediction result based on the dynamic supply and demand deviation feature set includes: Inputting the dynamic supply-demand deviation feature set into the encoder layer of the multi-source collaborative prediction model, performing multi-dimensional feature fusion processing on the dynamic supply-demand deviation feature set through the encoder layer, and generating a fusion feature vector including supply-demand coupling features and system stability correlation features; Calling the time series prediction layer of the multi-source collaborative prediction model to perform time series expansion processing on the fused feature vector to generate energy supply trend prediction features and load demand trend prediction features within a future preset time window; The energy supply trend prediction characteristics include photovoltaic power generation prediction curve, wind power generation prediction curve and energy storage system dispatchable capacity prediction curve, and the load demand trend prediction characteristics include industrial load demand prediction curve, commercial load demand prediction curve and residential load demand prediction curve; The energy supply trend prediction characteristics and the load demand trend prediction characteristics are used to calculate the supply-demand matching degree, and a multi-source collaborative prediction result including a supply-demand gap prediction value and a supply-demand balance time point is generated.

4. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 3, characterized in that: The performing dynamic adaptation strategy matching processing according to the multi-source collaborative prediction result and the real-time operation data set to generate a multi-source collaborative adaptation strategy set includes: Matching the supply-demand gap forecast value with a plurality of preset supply-demand balance threshold intervals to determine the target threshold interval to which the current supply-demand gap belongs; According to the strategy generation rule corresponding to the target threshold interval, a set of dispatchable energy supply sources is selected from the energy supply characteristics, and the maximum charge and discharge power of the energy storage system is determined based on the current state of charge and dispatchable capacity in the energy storage charge and discharge strategy; According to the preset industry priority weights in the load priority allocation strategy, the industrial load demand data sequence, the commercial load demand data sequence and the residential load demand data sequence are prioritized for load reduction; A multi-source collaborative adaptation strategy set including an energy output adjustment strategy, an energy storage charging and discharging strategy, and a load priority allocation strategy is generated based on the dispatchable energy supply source set, the maximum charging and discharging power, and the load reduction priority ranking.

5. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 3, characterized in that: The executing the adaptation strategy in the multi-source collaborative adaptation strategy set and obtaining adaptation execution feedback data includes: Sending a power adjustment instruction in the energy output adjustment strategy to a photovoltaic inverter controller of the multi-source energy system to obtain actual adjustment response data of photovoltaic power generation power; Sending a charge and discharge rate instruction in the energy storage charge and discharge strategy to the bidirectional converter of the energy storage system to obtain actual charge and discharge power data of the energy storage system; Sending a load reduction instruction in the load priority allocation strategy to a load management terminal to obtain actual load reduction amount data; The actual adjustment response data, the actual charging and discharging power data and the actual load reduction amount data are combined to generate adaptation execution feedback data, and a strategy execution deviation index between the adaptation execution feedback data and the multi-source collaborative prediction result is calculated.

6. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 5, characterized in that: The performing parameter optimization on the multi-source collaborative prediction model based on the adaptation execution feedback data to trigger a closed-loop adaptation operation includes: Inputting the strategy execution deviation index into the feedback learning layer of the multi-source collaborative prediction model, and dynamically adjusting the feature fusion weight matrix of the encoder layer and the time window expansion parameter of the time series prediction layer through the feedback learning layer; Regenerate an updated multi-source collaborative prediction result according to the adjusted feature fusion weight matrix and time window expansion parameter, and perform a secondary dynamic adaptation strategy matching process on the updated multi-source collaborative prediction result and the real-time operation data set to generate an optimized multi-source collaborative adaptation strategy set; Repeat the optimized multi-source collaborative adaptation strategy set and iteratively update the parameters of the multi-source collaborative prediction model until the strategy execution deviation index is lower than a preset deviation threshold.

7. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 6, characterized in that: The dynamically adjusting the feature fusion weight matrix of the encoder layer and the time window expansion parameter of the time series prediction layer through the feedback learning layer includes: Calculating the gradient change of the feature fusion weight matrix of the encoder layer according to the strategy execution deviation index, and updating the feature fusion weight matrix based on the gradient descent algorithm; According to the time dimension deviation component in the strategy execution deviation indicator, adjusting the window length and sliding step size in the time window expansion parameter of the time series prediction layer; The adjustment of the time window extension parameter satisfies the following conditions: When the time dimension deviation component increases, shortening the window length and increasing the sliding step length; When the time dimension deviation component decreases, the window length is extended and the sliding step size is reduced.

8. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 1, characterized in that: Before collecting the real-time operation data set of the multi-source energy system, the method further includes a pre-training step of the multi-source collaborative prediction model: Acquire a historical energy operation data set, wherein the historical energy operation data set includes historical energy supply characteristics, historical load demand characteristics, and historical system stability characteristics; Standardizing the historical energy operation data set to generate a standardized training data set; Constructing an initial multi-source collaborative prediction model, wherein the initial multi-source collaborative prediction model includes a feature encoder, a timing predictor, and a feedback corrector; The standardized training data set is input into the initial multi-source collaborative prediction model for iterative training until the error rate between the prediction result output by the initial multi-source collaborative prediction model and the actual historical operation data is lower than a preset threshold, thereby obtaining a pre-trained multi-source collaborative prediction model.

9. The rebalancing dynamic adaptation method applied to a multi-source system according to claim 8, characterized in that: The step of standardizing the historical energy operation data set to generate a standardized training data set includes: Performing maximum and minimum value normalization processing on the photovoltaic power generation data, wind power generation data, and energy storage system output power data in the historical energy supply characteristics to obtain normalized energy supply characteristics; Performing Z-score standardization processing on the industrial load data, commercial load data, and residential load data in the historical load demand characteristics to obtain standardized load demand characteristics; Performing sliding window mean processing on the power grid frequency deviation data and voltage deviation data in the historical system stability characteristics to obtain smoothed system stability characteristics; Aligning the normalized energy supply characteristics, the standardized load demand characteristics, and the smoothed system stability characteristics in time series and merging them to generate a standardized training data set; And, inputting the standardized training data set into the initial multi-source collaborative prediction model for iterative training until the error rate between the prediction result output by the initial multi-source collaborative prediction model and the actual historical operation data is lower than a preset threshold, thereby obtaining a pre-trained multi-source collaborative prediction model, including: Dividing the standardized training data set into a training set and a validation set, wherein the training set is used to update the parameters of the initial multi-source collaborative prediction model, and the validation set is used to evaluate the generalization ability of the model; In each iterative training process, the feature encoder is called to perform feature fusion on the data in the training set to generate a training fusion feature vector; Calling the time series predictor to generate a training prediction result based on the training fusion feature vector; Calling the feedback corrector to calculate the model loss function according to the difference between the training prediction result and the real historical operation data, and updating the parameters of the feature encoder and the timing predictor based on the back propagation algorithm; When the error rate on the validation set does not decrease for multiple consecutive iterations, the training is terminated and the current model parameters are saved as a pre-trained multi-source collaborative prediction model.

10. A multi-source energy debugging service system, characterized in that: The multi-source energy debugging service system includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the rebalancing dynamic adaptation method applied to a multi-source system as described in any one of claims 1 to 9 above.

Images