Micro-grid energy management control system based on energy storage and distributed photovoltaic

Abstract

The invention relates to the technical field of energy management and control, in particular to a micro-grid energy management and control system based on energy storage and distributed photovoltaic, and the system comprises a load prediction scheduling module which collects real-time grid data, analyzes the current load and photovoltaic generating capacity, carries out the time sequence analysis, and generates real-time load prediction data. According to the invention, in the load prediction, the real-time load prediction data generated based on time sequence analysis is combined with the photovoltaic power generation fluctuation, the charging and discharging strategy of the energy storage equipment is dynamically corrected, and the energy supply mismatching caused by the load fluctuation is reduced. According to the causal inference method, a multi-layer causal chain is constructed, key influence factors in photovoltaic operation are identified in combination with real-time load prediction data, real-time adjustment of energy distribution is realized through feedforward correction of fault early warning signals, and equipment overload or energy waste caused by imbalance of energy flow directions is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of energy management control, and particularly to a microgrid energy management and control system based on energy storage and distributed photovoltaics. Background Art

[0002] The technical field of energy management control mainly involves real-time monitoring, regulation, and optimal control in the processes of energy production, distribution, storage, and use. This field covers various energy forms such as electric power, thermal energy, and gas energy. The core lies in achieving the balance and efficient distribution of energy on the supply side, storage side, and load side through means such as data acquisition, prediction analysis, and dynamic regulation.

[0003] In the prior art, in the case of large load fluctuations, the fixed scheduling strategy cannot respond to the changes in load demand in a timely manner, resulting in the failure of energy storage devices to effectively participate in energy regulation, causing energy supply-demand imbalance and a decline in equipment operation efficiency. In terms of fault identification, the prior art mainly relies on single-point monitoring and static threshold judgment, making it difficult to accurately identify the key influencing factors in the photovoltaic system, increasing the concealment of faults, and affecting the reasonable distribution of energy flow. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies existing in the prior art, and a microgrid energy management and control system based on energy storage and distributed photovoltaics is proposed.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: The microgrid energy management and control system based on energy storage and distributed photovoltaics includes:

[0006] A load prediction and scheduling module that collects real-time grid data, analyzes the current load and photovoltaic power generation, performs time series analysis, and generates real-time load prediction data; adjusts the charge and discharge plan of the energy storage device based on the real-time load prediction data, and generates an energy storage scheduling strategy;

[0007] A causal inference fault diagnosis module that uses the real-time load prediction data to construct a causal chain, identifies the key influencing factors in photovoltaic operation, and generates a fault warning signal; performs feedforward correction on the warning signal, adjusts the energy distribution strategy, and generates a corrected energy flow direction;

[0008] A real-time energy optimization module that uses the corrected energy flow direction to execute a model predictive control algorithm to optimize and regulate the energy flow in the microgrid, and generates an optimized energy configuration;

[0009] A system stability monitoring module that monitors the operating state of the entire microgrid based on the optimized energy configuration, performs stability analysis, and generates an operating result; dynamically adjusts the deviation in the operating result to obtain an adjusted system state.

[0010] Preferably, the step of obtaining the real-time load prediction data is as follows:

[0011] Collect real-time grid data, including photovoltaic power generation, load current, and voltage, to obtain a real-time grid operation status data set;

[0012] According to the real-time grid operation status data set, calculate the real-time load correlation evaluation value. The calculation formula is:

[0013]

[0014] where CR is the real-time load correlation evaluation value, is the photovoltaic power generation, I i is the real-time load current, U i is the real-time voltage, P i is the predicted load current, rm is the number of sampling points, and max(P) represents the maximum value of the load current;

[0015] According to the real-time load correlation evaluation value, combined with the time series model, deduce the load demand trend in the future time period to generate real-time load prediction data.

[0016] Preferably, the step of obtaining the energy storage scheduling strategy is as follows:

[0017] Based on the real-time load prediction data, obtain the charge and discharge status of the current energy storage device to obtain an energy storage device status parameter set;

[0018] According to the energy storage device status parameter set, calculate the energy storage regulation adaptability of the energy storage device. The calculation formula is:

[0019]

[0020] where ED is the energy storage regulation adaptability, is the charge and discharge power of each energy storage device, is the real-time load prediction data, is the current load capacity of the energy storage device, T j is the remaining capacity of the energy storage device, N is the total number of energy storage devices, and k is an index value representing different energy storage devices;

[0021] According to the energy storage regulation adaptability, combined with the maximum charge and discharge rate of the energy storage device, optimize the charge and discharge plan of the energy storage device to generate an energy storage scheduling strategy.

[0022] Preferably, the step of obtaining the fault warning signal is as follows:

[0023] Based on the real-time load prediction data, the Pearson correlation coefficient is used to evaluate the correlation between each data, and the causal relationship between the data is determined by combining expert knowledge to construct a causal chain;

[0024] Based on the causal chain, calculate the comprehensive influence index, and the calculation formula is:

[0025]

[0026] where ZI is the comprehensive influence index, is the efficiency of the photovoltaic component, M i is the maintenance index of the photovoltaic component, A i is the absolute value of the difference between the photovoltaic output and the load demand, and n is the number of factors;

[0027] Based on the comprehensive influence index, if the influence index exceeds the preset threshold, it is identified as a fault source and a fault warning signal is generated.

[0028] Preferably, the steps for obtaining the corrected energy flow are as follows:

[0029] Based on the fault warning signal, calculate the difference between the photovoltaic output deviation and the load demand, and generate an energy distribution abnormal parameter set;

[0030] According to the energy distribution abnormal parameter set, analyze the remaining capacity, charge and discharge power and conversion efficiency of the energy storage device, evaluate the ability of the energy storage system to respond to the real-time load demand, and combine the difference between the photovoltaic output and the load demand to generate an energy distribution adjustment strategy;

[0031] According to the energy distribution adjustment strategy, execute the adjustment of the charge and discharge plan of the energy storage device, modify the energy output ratio of the photovoltaic power generation and the energy storage system, and generate the corrected energy flow.

[0032] Preferably, the steps for obtaining the optimized energy configuration are as follows:

[0033] Based on the corrected energy flow, extract the parameters of the photovoltaic power generation amount, the charge and discharge state of the energy storage device and the real-time load demand, and generate an energy flow optimization parameter set;

[0034] According to the energy flow optimization parameter set, calculate the energy flow dynamic adjustment coefficient, and the calculation formula is:

[0035]

[0036] where D is the energy flow dynamic adjustment coefficient, Q i is the real-time photovoltaic power generation power, E i is the current output power of the energy storage device, L i is the real-time load demand, C i is the remaining capacity of the energy storage device, R iis the operating temperature of the photovoltaic device, in °F i is the battery aging factor of the energy storage device, and m is the number of sampling points;

[0037] Based on the energy flow dynamic regulation coefficient, combined with the real-time load demand and the maximum charge and discharge capacity of the energy storage device, allocate the photovoltaic output and the energy storage charge and discharge ratio to generate the optimized energy configuration.

[0038] Preferably, the steps for obtaining the operation result are as follows:

[0039] Based on the optimized energy configuration, call the parameters of the photovoltaic power generation, the charge and discharge status of the energy storage device, and the real-time load demand to generate a set of microgrid operation status parameters;

[0040] According to the set of microgrid operation status parameters, analyze the energy balance of each device under different load levels, extract the energy deviation values of photovoltaic power generation, energy storage output and load consumption, and compare with the real-time working status and standard status of the device to generate a microgrid stability evaluation result;

[0041] Based on the microgrid stability evaluation result, calculate the energy flow consistency in the current operation state, and compare and analyze the operation response delay of each device and the energy compensation situation of load changes to generate an operation result.

[0042] Preferably, the steps for obtaining the adjusted system state are as follows:

[0043] According to the operation result, combined with the output power of the photovoltaic power generation device, the remaining capacity and response rate of the energy storage device, analyze the dynamic balance situation of energy supply and load demand, execute the energy storage charge and discharge plan and adjust the photovoltaic power output to generate a dynamic adjustment control instruction;

[0044] Based on the dynamic adjustment control instruction, analyze the operation status of each device after adjustment, verify the matching situation of the energy output of the photovoltaic power generation and the energy storage device and the load demand, and execute the reallocation of the energy flow direction to generate the adjusted system state.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0046] In the present invention, in load prediction, real-time load prediction data generated based on time series analysis is combined with the fluctuations of photovoltaic power generation to dynamically correct the charge and discharge strategies of energy storage devices, reducing the energy supply mismatch caused by load fluctuations. The causal inference method constructs a multi-layer causal chain, combines real-time load prediction data to identify key influencing factors in photovoltaic operation, and realizes the immediate adjustment of energy distribution through feedforward correction of fault warning signals, avoiding equipment overload or energy waste caused by unbalanced energy flow. Real-time energy optimization is based on the corrected energy flow, executes a model predictive control algorithm, and dynamically adjusts the outputs of photovoltaic and energy storage devices to ensure that the demands on the load side are fully responded to. The system operation state monitoring combines real-time deviation analysis and dynamic adjustment strategies, enabling energy management to maintain continuous optimization during operation, avoiding the limitations of fixed scheduling strategies on system adaptability in traditional methods, and improving the flexibility and stability of energy configuration. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a system flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] Please refer to Figure 1 , the present invention provides a technical solution: The microgrid energy management and control system based on energy storage and distributed photovoltaics includes:

[0050] A load prediction and scheduling module that collects real-time grid data, analyzes the current load and photovoltaic power generation, performs time series analysis, and generates real-time load prediction data; adjusts the charge and discharge plans of energy storage devices based on the real-time load prediction data to generate an energy storage scheduling strategy;

[0051] A causal inference and fault diagnosis module that uses real-time load prediction data to construct a causal chain, identifies key influencing factors in photovoltaic operation, and generates fault warning signals; performs feedforward correction on the warning signals, adjusts the energy distribution strategy, and generates the corrected energy flow;

[0052] A real-time energy optimization module that uses the corrected energy flow to execute a model predictive control algorithm to optimize and adjust the energy flow in the microgrid and generate an optimized energy configuration;

[0053] A system stability monitoring module that monitors the operation state of the entire microgrid based on the optimized energy configuration, performs stability analysis, and generates an operation result; dynamically adjusts the deviation in the operation result to obtain the adjusted system state.

[0054] The steps for obtaining real-time load prediction data are as follows:

[0055] Collect real-time grid data, including photovoltaic power generation, load current, and voltage, to obtain a dataset of the real-time operating state of the grid;

[0056] According to the dataset of the real-time operating state of the grid, calculate the real-time load correlation evaluation value. The calculation formula is:

[0057]

[0058] where CR is the real-time load correlation evaluation value, is the photovoltaic power generation, I i is the real-time load current, U i is the real-time voltage, P i is the predicted load current, rm is the number of sampling points, and max(P) represents the maximum load current;

[0059] According to the real-time load correlation evaluation value, combined with the time series model, deduce the load demand trend in the future time period to generate real-time load prediction data.

[0060] Specifically, when collecting real-time grid data, the recording range of photovoltaic power generation is set between 0 kW and 50 kW and collected once every 5 seconds. The load current is collected synchronously and compared with the preset range of 0 A to 60 A. At the same time, the voltage record is monitored and compared with the range of 0 V to 380 V. Each collection record is accompanied by a collection timestamp and arranged in chronological order. To screen out outliers, the voltage is first checked to see if it exceeds the range of 0 V to 380 V or the load current exceeds the range of 0 A to 60 A. When data outside this range is detected, it is marked as invalid and removed from the subsequent process. The same interval test is also performed on the photovoltaic power generation. For example, if the photovoltaic power generation is negative or exceeds 50 kW, it is considered unreliable and excluded. The collection integrity of the valid data is synchronously recorded and compared with this interval using consecutive multiple collection values to determine if there are obvious fluctuations. Each data item is centrally stored and attached with a unique identifier for convenient subsequent indexing. After multiple verifications, all qualified data is unified to form a dataset of the real-time operating state of the grid.

[0061] The benefit of the formula is that by comprehensively quantifying the differences between photovoltaic power generation, real-time voltage, actual load current, and predicted load current under the same evaluation system, and through the analysis of each sampling point in the numerator part and the normalization operation of the maximum load current in the denominator part, the coupling degree between photovoltaic output and load prediction can be reflected within a certain range, so that the subsequent scheduling strategy can carry out more targeted energy balance in combination with this evaluation value.

[0062] ZQ iThe acquisition steps are as follows: Obtain the photovoltaic power generation value through power detection at the output end of the photovoltaic inverter. The value range is generally between 0 kW and 50 kW. The acquisition method usually involves real-time recording of the current output power value of the inverter and then indexing it with a timestamp. Before execution, the sunshine conditions in the area where the photovoltaic modules are located are verified, and the upper and lower limits of the effective power generation are set. When recording data, invalid points caused by abnormal jitter or signal loss need to be excluded. This parameter can identify whether the daily power generation curve meets expectations by comparing historical data. For example, after measuring 2 kW, 5 kW, and 4 kW respectively within a 10-minute monitoring interval, these three groups of photovoltaic power generation data are respectively corresponding to Then it can be brought into the calculation process during the molecular operation. At this time, if the reading shown on the power detector is negative or exceeds 50 kW, it is directly discarded. At the same time, more sampling points are continuously accumulated according to the actual acquisition frequency, thus forming a complete dataset.

[0063] I i The acquisition steps are as follows: Obtain the real load current value through a current transformer on the load side or other measurement means. The common range is between 0 A and 60 A. After the acquisition is completed, the measured current record is matched with the corresponding photovoltaic output at the corresponding moment, so as to correctly compare with the predicted load current when calculating the molecular part. When obtaining data, a threshold of exceeding 60 A is set and marked. This threshold is established through statistics of on-site electrical equipment. For example, it is found from the actual equipment nameplate and multiple measurement results that the maximum load during daily operation is about 55 A. On this basis, a 5 A margin is left and set as the limit value of 60 A. If 76 A is detected during the acquisition process, it means the data is invalid. Analyze the on-site monitoring log to confirm whether there is an instrument failure or some abnormal impact. Finally, the current data within the range of 0 A to 60 A is assigned to I i to participate in the formula operation.

[0064] U i The acquisition steps are as follows: Detect the line voltage through a voltage sensor. Generally, in industrial or commercial application scenarios, standard line voltages such as 220 V or 380 V can be referred to, but there are often small fluctuations in on-site tests. When collecting voltage data, first determine the sampling frequency. A number of discrete voltage values can be obtained within an observation period. Subsequently, these values are also time-stamped and aligned with the photovoltaic output and load current, and 0 V and excessive voltages are excluded. This voltage threshold can be determined in combination with the rated output of the transformer and local distribution line parameters. For example, within the set range of 0 V to 380 V, when a 400 V record appears in one acquisition, it is positioned as invalid data and investigated. Finally, all qualified voltage values are paired with the photovoltaic power generation and load current at the same time point in the form of U i to calculate the

[0065] P i The acquisition steps of P are as follows: This parameter represents the predicted load current value at the same moment, which is obtained by analyzing historical load data and the electricity consumption pattern in the corresponding time period. The data usually comes from the dispatching center, which performs time series modeling based on the load curves of the past few days or even weeks, and uses parameter calibration methods to deduce the typical load current values in different time periods, and assigns them to the corresponding moments to form the P i sequence. At the same time, during actual use, it will be adjusted in combination with the working hours of local electrical equipment. For example, industrial equipment usually has a larger operating current during the day and a relatively smaller one at night. After each prediction cycle ends, the time series parameters required by the prediction algorithm will be updated to obtain a P closer to the actual situation of the day i . In specific operations, the historical current sampling data of several days will be cleaned and combined with the corresponding time tags, and then the prediction values will be generated through model fitting such as autoregressive plus moving average, and finally the results will be retrieved and assigned to P in the order of the sampling time periods i for correlation calculation by substituting into the formula

[0066] The acquisition steps of rm are as follows: This parameter represents the number of sampling points selected for calculating R. In a complete monitoring activity, a sampling point will be generated in each time period. When the cumulative number reaches the pre-specified quantity, the formula operation can be executed. This quantity can be determined by the system layout and the observation period. For example, if the acquisition is performed once every 5 seconds and lasts for 1 hour, 720 sampling points will be obtained, where rm = 720. When determining this parameter, the hardware acquisition ability and data throughput level will be combined. If the on-site monitoring scale is large, a higher sampling frequency can be set and the database bandwidth can be increased for support. During specific implementation, after all valid samples are recorded, the actual number of completed samples will be counted according to the time stamp or index range

[0067] The acquisition steps of max(P) are as follows: This parameter specifically refers to the maximum value in the predicted load current sequence P i . It is necessary to first obtain a complete predicted current array after time series modeling, and then scan the array to find the maximum current value. For example, if the predicted load current array contains 9A, 10A, 13A, 15A, etc., then max(P) = 15

[0068] Calculation process: In the first step, determine the values of each parameter. For example, let rm = 3 U1 = 220V, U2 = 219V, U3 = 221V, I1 = 11A, I2 = 12A, I3

[0069] = 10A, P1 = 10A, P2 = 11A, P3 = 9A, max(P) = 15A

[0070] In the second step, substitute the numerator for summation calculation. First, calculate and Taking i = 1 as an example, its product is approximately equal to 0.0227. Taking i = 2 as an example, the product is approximately equal to 0.0274. Taking i = 3 as an example, the product is approximately equal to 0.0181. Summing up these three terms gives

[0071] 0.0227 + 0.0274 + 0.0181 = 0.0682;

[0072] In the third step, divide the numerator result by the denominator max(P) = 15 to get

[0073] This result indicates that among the selected sampling points, the deviation between the photovoltaic power generation and the load prediction value is relatively small and at a relatively low correlation level. The corresponding CR value is only about 0.00455, indicating that there is no significant gap between the actual load current and the predicted current during the current period. If a significant increase in CR is observed in the future, the influencing factors can be traced back from the formula to further calibrate the relevant parameters of the load prediction or photovoltaic power generation. After obtaining the real-time load correlation evaluation value, the load demand trend will be deduced in combination with the time series model. We select the ARIMA model and extract thousands of effective samples from the load data records every 5 seconds in the recent week as the training set. First, establish a timestamp for each sampling point and eliminate the invalid records outside the range of 0A to 60A. Subsequently, use autoregression and differencing to identify the parameter order of the sequence. During this process, the AIC criterion will be calculated for each possible order and the optimal order will be obtained by comparison. At the same time, a fluctuation threshold is set to identify the abnormal points beyond the normal load range. This threshold is obtained by statistically analyzing the fluctuation distribution of historical data. For example, after sorting thousands of current data, find the upper boundary of the distribution and combine the normal electricity consumption scenario to calculate with a 2A step. After gradually approaching the critical range where the outlier frequency is greater than one in ten thousand, it is established as the final threshold. After the ARIMA model fitting is completed, the load prediction value for the next time period is generated and combined with the real-time load correlation evaluation value. When a new sample is detected in real time, it is input into the model and updated to complete the rolling prediction of the time series, and finally form the real-time load prediction data for the subsequent control strategy.

[0074] The steps to obtain the energy storage scheduling strategy are as follows:

[0075] Based on the real-time load prediction data, obtain the charge and discharge status of the current energy storage device to obtain the energy storage device status parameter set;

[0076] Calculate the energy storage regulation adaptability of the energy storage device according to the energy storage device state parameter set. The calculation formula is as follows:

[0077]

[0078] where ED is the energy storage regulation adaptability, is the charge and discharge power of each energy storage device, is the real-time load prediction data, is the current load capacity of the energy storage device, T j is the remaining capacity of the energy storage device, N is the total number of energy storage devices, and k is the index value representing different energy storage devices;

[0079] Optimize the charge and discharge plan of the energy storage device according to the energy storage regulation adaptability, and combine with the maximum charge and discharge rate of the energy storage device to generate an energy storage scheduling strategy.

[0080] Specifically, based on the real-time load prediction data, before obtaining the charge and discharge status of the current energy storage device, it is necessary to first extract the load values corresponding to each moment from the obtained prediction data, and then timestamp-correspond these values with the most recent charge and discharge record of the energy storage device to ensure that the two are matched under the same time reference. Then, monitor the actual working mode of the energy storage device. For example, for battery energy storage devices, it is possible to determine whether it is in the discharge state by monitoring its positive current or in the charge state by monitoring its negative current. It is also possible to read the real-time control instructions of the system for the battery to identify the current working mode. During the monitoring process, it is necessary to determine whether the magnitudes of the current and power fall within the pre-established effective range. For example, compare the discharge current with the range of 0A to 300A, and compare the charge current with the range of 0A to 200A. Exceeding these ranges will be determined as abnormal or faulty and marked. At the same time, it is necessary to compare the discharge power with the range of 0kW to 500kW or the charge power with the range of 0kW to 300kW. When an abnormality is detected, a prompt should be added to the system. After confirmation, the charge and discharge data within the normal range should be continuously summarized in chronological order. Such data includes the current power value, charge mode or discharge mode identification, device temperature, single-cell data, and working duration, etc. And these data should be referenced in combination with the internal cycle test results during the previous maintenance or servicing. If it is found here that the temperature or internal cycle test results exceed the specified values, further records will be made. The specified values are usually obtained from the rated parameters of the manufacturer. In the field, the temperature threshold range can be generated by monitoring and statistically analyzing factors such as temperature multiple times. For example, the temperature preset value can be obtained by comparing the average temperature in the normal working area plus a certain upper limit based on the monitoring results for a continuous week. The above-mentioned qualified and effective monitoring data will be directly integrated into the charge and discharge status table of the energy storage device after preliminary cleaning. This table is associated with the load prediction data through a time index. For example, when the load prediction data at the 120th second of the day is 40kW and the corresponding actual charging power is 35kW, these two sets of data will be bound with the same timestamp. After all relevant data is aligned, it can be uniformly packaged to form the energy storage device status parameter set.

[0081] The benefit of the formula is that by comprehensively evaluating the difference degree between the current charge and discharge power of each energy storage device, the predicted load and the actual load capacity of the device, and combining the remaining capacity of the device and the total number of energy storage devices for normalization processing, the dispatching system can quickly obtain the adjustment adaptability of each energy storage device at the current moment, thereby improving the flexibility of dispatching allocation in the subsequent optimization of the charge and discharge plan.

[0082] AE jThe acquisition steps are as follows: This parameter represents the charge and discharge power of the jth energy storage device, which needs to be obtained through real-time monitoring of the energy storage inverter or power metering device. The detection range can be determined by referring to the equipment nameplate and on-site maximum power assessment. For example, for a certain industrial-grade lithium battery energy storage system, its discharge power can reach up to 300kW to 500kW, and the charging power is generally around 300kW. During on-site operation, an independent power metering channel will be established for each energy storage device. By sampling multiple times per second and recording the dynamically changing power values, and then taking the average value of each sampling as the actual power output or input of the device at the current moment. When forming the AE j sequence, extreme values need to be compared. If it exceeds the rated power of the established equipment, it should be registered as abnormal. At the same time, cross-checks are carried out in combination with the working hours and aging degree of each device. Only after the data is fully reliable can these power values be assigned to the corresponding AE j for participating in formula operations. For example, for three devices in a certain energy storage system, the current discharge powers are recorded as 230kW, 245kW, and 220kW respectively, then the corresponding

[0083] The acquisition steps are as follows: This parameter refers to the real-time load prediction data, which needs to extract the data at the same moment or within the same time window from the previously obtained real-time load prediction results and map them to each energy storage device. This is done to match the working point of the current energy storage device. Usually, the prediction data is corresponded to each energy storage device one by one through timestamp alignment or polling query. If the system monitors that the predicted load demand at this moment is high, then the will also increase accordingly. During the data acquisition process, it is necessary to ensure the coherence of the time series and that the prediction model parameters have been updated. When there are multiple prediction values within a cycle, certain interpolation or averaging processing can be carried out to ensure that there is only one unique prediction value at the same time node. For example, within a day, there are 1440 prediction points with a minute cycle. The system queries the load prediction value of 300kW for this period at 08:30, and this 300kW can be attached to all energy storage devices that are online at that time, corresponding to the sequence value.

[0084] The acquisition steps are as follows: This parameter corresponds to the current load capacity of the energy storage device, indicating the maximum power that the energy storage device can provide or absorb at this moment. In the actual field, it may be measured according to the device type and operating environment. For example, when the temperature is too high, the device power will be derated. Therefore, when obtaining this parameter, the device temperature, operating duration, and derating curve published by the manufacturer are usually combined. For example, by querying the device temperature curve within the past 24 hours, if the temperature peak reaches 45°C, then find the corresponding derating section in the manufacturer's manual, reduce the maximum power from 300 kW to 280 kW, and at the same time refer to the real-time current limit and voltage change of the device to finally determine the available maximum power value at this moment and use it as If the system manages multiple energy storage devices simultaneously, calculate the corresponding available power independently for each device and distinguish them during the operation. For example, if the current available powers of three energy storage devices are 280 kW, 250 kW, and 270 kW respectively, assign them to for subsequent solution.

[0085] T j The acquisition steps are as follows: This parameter refers to the remaining capacity of the energy storage device, generally measured in ampere-hours (Ah) or kilowatt-hours (kWh), and may also be adjusted under the conversion relationship between power and voltage. It is necessary to first collect the state information of the energy storage system, including the real-time battery charge, number of cycles, and instantaneous current, and also refer to the attenuation curve of the device under different state of charge conditions. When the system detects that the remaining capacity of the device is too low, the corresponding T j will display a smaller value. In the actual configuration, by comparing the depth of discharge of the device with the SOC (State of Charge) curve provided by the manufacturer multiple times, the calculation formula for the remaining capacity can be obtained. On this basis, combined with the current

[0086] discharge power at this moment and make fine-tuning. For example, when measuring the current remaining capacity of a single energy storage device as 180 kWh, then quantify it as T j = 180. If another device shows a remaining capacity of 270 kWh, then record it as T j

[0087] = 270, so that the differential measurement of energy storage devices with different capacities can be realized during subsequent operations.

[0088] The acquisition steps of N are as follows: This parameter represents the total number of energy storage devices. Each independent device has a corresponding identifier in the system. It is necessary to count all the online and available devices. During on-site deployment, the total number of active energy storage devices is usually determined through device networking registration or maintenance management ledger. For example, in a certain distributed photovoltaic energy storage scenario, 5 energy storage devices may be participating in the scheduling simultaneously, then N = 5. If one of the devices is in the maintenance and out-of-service state and no longer participates in the scheduling, at this time N = 4.

[0089] The steps to obtain k are as follows: This parameter is an index used to distinguish different energy storage devices. In the program, it usually processes a given number of energy storage devices through loops or iterations, and then substitutes their respective operating information into the formula for calculation. When applied in the field, if there are N devices, k processes the devices sequentially from 1 to N. For example, if there are 3 devices in the system, then k can take the sequence {1, 2, 3}, corresponding to and other parameters.

[0090] Calculation process:

[0091] In the first step, assign T j corresponding values to each device. For example, if there are 3 energy storage devices here, let

[0092] In the second step, calculate for each device respectively. Taking j = 1 as an example, AL1 - AC1

[0093] = 300 - 280 = 20, its absolute value is 20. After taking the square root, it is approximately 4.4721, and then divided by T1 = 180 to get Multiply by AE1 = 230 to get 230×0.02484≈5.7132. According to the same steps, calculate for j = 2 and j = 3 respectively to obtain

[0094] In the third step, sum up the above three results: 5.7132 + 6.4180 + 6.0250 = 18.1562, and then divide by the total number of energy storage devices N = 3 to get

[0095] The result shows that the overall energy storage regulation adaptation degree of the three energy storage devices at this moment is 6.05207. A higher value means there is a greater deviation between the device and the load demand or a higher charge-discharge power level, and more refined strategy allocation is required. If it is subsequently observed that ED gradually increases and exceeds a certain empirical threshold, it can be determined that the current dispatching load allocation is approaching the upper limit, and the charge-discharge plan should be adjusted in a timely manner. If ED is close to 0, it indicates that the current device operation is relatively matched with the load prediction. According to the energy storage regulation adaptation degree, optimize the charge-discharge plan of the energy storage device in combination with the maximum charge-discharge rate of the energy storage device. During the implementation process, it is necessary to first obtain the corresponding value for each energy storage device from the regulation adaptation degree and arrange them according to the device index. Then, compare the maximum power that each energy storage device can provide or absorb with the adaptation degree value. If the adaptation degree is relatively high and the remaining capacity is sufficient to support the discharge rate, a higher discharge power will be allocated to the device in the corresponding dispatching logic. At the same time, check whether there is a device with a relatively low adaptation degree but a large available capacity. If such a device is found, its charge-discharge allocation power will be reduced. During the allocation process, truncation control should be performed on devices that exceed the preset rate limit. For example, when the maximum charging rate is set at 300 kW, once it is detected that the allocated power exceeds this value, the upper limit will be 300 kW. This upper limit can be determined based on the rated power range provided by the manufacturer and multiple measurement statistics during stable operation on-site. If the energy storage device used on-site has a relatively high failure probability or thermal management pressure after exceeding 300 kW, the maximum charging rate will be set at 300 kW. Rate limit should also be performed on the discharge, and after all device power allocations are completed, the total power should be rechecked to confirm that it operates within the set safety range. When this step is completed, record the dispatching result according to the device index and compare it with the previously obtained load prediction data. If there is still some surplus power, fine-tuning of the allocation for individual devices can be considered. At the same time, judge whether the device has space for continued charging or discharging based on the remaining capacity value T obtained above. When it is found that the remaining capacity of an individual device is insufficient, it should be temporarily removed from the allocation list or its charging rate should be reduced. The entire process needs to traverse all online energy storage devices and synthesize the final energy storage dispatching strategy after recording the adaptation degree and rate limit.

[0096] The steps for obtaining the fault warning signal are as follows:

[0097] Based on the real-time load prediction data, use the Pearson correlation coefficient to evaluate the correlation between each data, combine expert knowledge to determine the causal relationship between the data, and construct a causal chain;

[0098] Based on the causal chain, calculate the comprehensive influence index. The calculation formula is:

[0099]

[0100] Among them, ZI is the comprehensive influence index, is the photovoltaic element efficiency, M i A is the photovoltaic component maintenance index. i is the absolute value of the difference between PV output and load demand, n is the number of factors;

[0101] Based on the comprehensive impact index, if the impact index exceeds the preset threshold, it is identified as a fault source and a fault warning signal is generated.

[0102] Specifically, based on real-time load forecast data, the Pearson correlation coefficient is used to evaluate the correlation between the data, and the causal relationship between the data is determined in combination with expert knowledge. When conducting the correlation evaluation, the real-time load forecast data is first segmented and compared. For example, the segment interval can be selected as every ten minutes or every half hour as a cycle and the photovoltaic power generation, load demand, voltage and current data are summarized. After generating multiple vectors for each group of data, the Pearson correlation coefficient is calculated in sequence. In comparison with the correlation requirements required for the causal chain in the actual monitoring scenario, the data columns with obvious low correlation or negative correlation are eliminated or marked. Then, based on the years of maintenance records of on-site engineering and technical personnel and the expert knowledge base, each photovoltaic parameter and load indicator are paired one by one for manual analysis and identification of potential causal sequences. If there is a significant deviation in the efficiency data of some photovoltaic components during periodic monitoring, it can be included in the high priority analysis. A list is made, and at the same time, the time periods with large fluctuations in load demand are thoroughly investigated and the corresponding photovoltaic power generation status is confirmed. After comparing the on-site monitoring logs, these deviation data are checked against the supporting records, and known reasons such as seasonal light intensity and temporary fault shutdown are eliminated or confirmed. Finally, the remaining valid data are causally linked, and the sequence of each link in the time dimension is established, and the key nodes and the degree of correlation are marked to form a clear causal chain structure. In the causal chain, it can be shown that some photovoltaic components are not maintained in time, resulting in a decrease in efficiency, which in turn leads to an increase in the difference between photovoltaic output and load demand. The impact path can also be shown that the photovoltaic energy matching degree declines due to temporary startup or shutdown of load-side equipment in certain periods. The above causal chain will serve as an important reference and input for the subsequent calculation of the comprehensive impact index to construct a causal chain.

[0103] The benefit of the formula is that it couples the PV element efficiency, PV element maintenance index and the cube root term of the difference between PV output and load demand in the same evaluation framework for comprehensive superposition, and finally outputs a unified value in the form of square root. This value can quantify potential fault hazards under the joint action of multi-dimensional factors.

[0104] CE iThe acquisition steps are as follows: This parameter is the efficiency of the photovoltaic element, and its value usually ranges from 15% to 22%, which may vary due to different component materials and manufacturing processes. On-site, the current efficiency can be confirmed by directly comparing the actual output power of the photovoltaic module with its rated power under standard test conditions. For example, for a certain type of photovoltaic module, the nominal output power under standard test conditions is 300W, and its standard efficiency is approximately 20% according to the rated power of the module. If it is monitored that the actual output power of this module is only 270W under the same light intensity, the efficiency is approximately 18%. When there are multiple photovoltaic modules in the system, the current efficiency of each module needs to be recorded separately and the average value is obtained. Usually, it is collected multiple times within a certain time period to obtain a relatively stable average value. If the module has dust accumulation, aging or other factors during operation and maintenance, the efficiency will be further reduced. After collecting these data, the efficiency of the photovoltaic element corresponding to each module can be assigned to CE i , for example, in the same batch of modules, efficiency values such as 18%, 17.5%, 19%, 16.8% are observed respectively, and then they are sequentially substituted into the formula to form CE1, CE2, CE3, CE4.

[0105] M i The acquisition steps are as follows: This parameter is the maintenance index of the photovoltaic element, which is used to measure the operation and maintenance quality, cleanliness and working stability of the module. The data can be accumulated during the inspection process. For example, the cleanliness of the module surface is quantified, the surface reflectivity of the module is detected using an optical reflection tester and the change rate of the internal resistance of the module is detected in combination with an internal resistance tester. The effect of the performance recovery of the module after each maintenance is statistically analyzed, and at the same time, whether the maintenance cycle is delayed is recorded. These information are scored and synthesized into a maintenance index value. The common value range can be set from 1 to 10, and the larger the value, the better the maintenance status. When calculating specifically, the scores of surface cleanliness, internal resistance stability and maintenance cycle compliance can be averaged or weighted averaged. If a module has multiple occurrences of not being cleaned in time or other failures, the score of this maintenance index will be reduced. For example, when inspecting 4 modules on the same day, the obtained maintenance indexes are 8.5, 8.0, 9.0, 6.5 respectively, and then they can be mapped to M1, M2, M3, M4 in this formula in turn.

[0106] A iThe acquisition steps are as follows: This parameter represents the absolute value of the difference between the photovoltaic output and the load demand. It is usually obtained by recording the real-time power generation of the photovoltaic module and the real-time power consumption on the load side, taking the difference between the two and taking the absolute value. If the photovoltaic output is greater than the load demand, it is a positive difference, otherwise it is a negative difference. After the absolute value is processed, it can reflect the degree of deviation between the two. In order to ensure the accuracy of the data, it is necessary to collect the photovoltaic power generation output and the real-time load value within the same time reference. When the system is large in scale, the sum of multiple parallel photovoltaic modules and the total load value can also be compared within the same time window. If this difference is too large for many consecutive days, it means that the system may have an abnormality on the load side or the photovoltaic side. For example, if the photovoltaic output is detected to be 280kW and the load demand is 300kW, then A i =|280-300|=20. If there are many components, the output of each component and the load demand are calculated accordingly, thus forming A1, A2, ..., A n The sequence participates in subsequent operations.

[0107] The steps for obtaining n are as follows: it is the number of factors, indicating how many PV modules or factors are included in the calculation of the comprehensive impact index during one operation. In actual scenarios, the system can dynamically determine the value of n based on the number of online PV modules or the temporary shutdown of some modules. For example, 10 PV modules are deployed, but 2 of them are under maintenance and are no longer included in the calculation, then n=8 this time.

[0108] Calculation process:

[0109] The first step is to list each CE separately i ,M i ,A i For example, in one inspection, 4 photovoltaic panels (n=4) are processed and recorded separately. Convert it from percentage to decimal form for calculation: 0.18, 0.175, 0.19, 0.168; maintenance index M1 = 8.5, M2 = 8.0, M3 = 9.0, M4 = 6.5; the absolute value of the difference between photovoltaic output and load demand A1 = 20, A2 = 15, A3 = 35, A4 = 25;

[0110] The second step is to calculate and For example Multiplying the two together gives 0.02118×2.7144≈0.05746. Calculate in the same way Multiplying by about 0.05388, Multiplying by about 0.06906, Multiply by about 0.07566;

[0111] Step 3: Sum up the above 4 items:

[0112] 0.05746 + 0.05388 + 0.06906 + 0.07566 = 0.25606. Take the square root of the sum result to get

[0113] This result indicates that under the combined action of the photovoltaic element efficiency, maintenance indicators, and the difference from the load demand of the 4 components currently being processed, the final comprehensive impact index is approximately 0.5060. When ZI is much larger than a preset threshold (e.g., 30), it indicates that there are quite serious potential faults. When the ZI value is relatively low and close to 0, it indicates that the overall performance of each factor is stable. In this example, 0.5060 is in a very low range compared to the threshold standard of previous operation and maintenance statistics and will not trigger the determination of the fault source. If the subsequent indicators increase significantly and exceed the upper limit set on-site, these components can be listed as key inspection objects based on this result. Based on the comprehensive impact index, if the impact index exceeds the preset threshold, it is marked as the fault source. During the execution process, first retrieve the comprehensive impact index calculated by the current system periodically, and compare it with the threshold range formulated based on historical fault events or component abnormal events. For example, after comparing multiple maintenance cases on-site, the threshold is set in the range of 30 to 35 and actually set to 30. Perform additional weighting processing on certain components or areas. Once it is monitored that the comprehensive impact index has reached 30 or above, these components can be recorded in the pending maintenance list. At the same time, cross-reference the load conditions and photovoltaic output of other components based on the operation data within the same time period to trace the cause of the soaring index. If multiple components exceed this value simultaneously, give priority to checking the components with the most severe efficiency decay or the lowest maintenance score. When it is confirmed that there are obvious output abnormalities or other physical structure damages in the photovoltaic components on-site, corresponding fault warning signals are generated according to the system logic.

[0114] The steps to obtain the corrected energy flow are as follows:

[0115] Based on the fault warning signal, calculate the deviation between the photovoltaic output and the difference in load demand, and generate an energy distribution anomaly parameter set;

[0116] According to the energy distribution anomaly parameter set, analyze the remaining capacity, charge and discharge power, and conversion efficiency of the energy storage device, evaluate the ability of the energy storage system to respond to real-time load demands, and combine the difference between the photovoltaic output and the load demand to generate an energy distribution adjustment strategy;

[0117] According to the energy distribution adjustment strategy, execute the adjustment of the charge and discharge plan of the energy storage device, modify the energy output ratio between photovoltaic power generation and the energy storage system, and generate the corrected energy flow.

[0118] Specifically, based on the fault warning signal, first, the photovoltaic power generation fluctuation information extracted from the fault warning signal is time-correlated with the load demand record to determine the real-time data of the photovoltaic output and the corresponding load demand in the current period. Then, the difference between the two is calculated and the absolute value of the difference is taken. The absolute difference corresponding to each time period is compared with each fault location or fault type to analyze whether there is a situation where the difference exceeds the pre-established difference range, which is usually obtained from the difference distribution during the daily operation of the equipment. For example, after continuously recording the photovoltaic power generation and load demand for two weeks, the normal difference range is statistically obtained from 0 to 30 kW. When the difference exceeds 30 kW, it is considered that an abnormal deviation occurs. At the same time, the electrical characteristic information associated with this deviation is collected and summarized, recording specific situations such as a sudden drop in photovoltaic output or a sudden increase on the load side, as well as supplementary items such as the status, temperature, and voltage of relevant photovoltaic modules. The above information will be merged into a parameter set, and then the occurrence frequency and amplitude of the difference fluctuation are screened one by one, and the difference is segmented and labeled using the same interval division method. For example, a first-level label is added to the paragraph where the difference is between 30 kW and 50 kW, and a second-level label is added to the paragraph where the difference exceeds 50 kW. At the same time, a time dimension is added to each labeled item, and the time duration of continuously maintaining the abnormal value is compared with the preset time threshold, which can be comprehensively obtained based on the instantaneous fluctuation limit of the photovoltaic equipment and the change cycle of the load side. If the difference in a certain period continuously exceeds the preset duration, the relevant photovoltaic modules are further listed as the key objects to be investigated. After all the data is compared, the energy distribution abnormal parameter set is obtained.

[0119] According to the energy distribution anomaly parameter set, read the difference range identifier and the recorded working status of the photovoltaic modules and energy storage system information contained therein respectively. Combine the remaining capacity data and the current charge-discharge power reading records of the energy storage system, and incorporate the conversion efficiency of the energy storage device. These conversion efficiency values are from the regular test session. Every week, the energy storage device undergoes a process of discharging to the set lower limit and then charging to the upper limit. The energy input-output ratio throughout the entire cycle is statistically calculated to obtain the corresponding conversion efficiency. When the remaining capacity of the energy storage device is obtained, interval comparison is carried out. For example, the interval from 100 kWh to 200 kWh is regarded as the interval that can satisfy medium-duration discharge, and the interval from 200 kWh to 400 kWh is regarded as the interval that can satisfy long-duration discharge. Then, refer to the difference sections sorted out previously to match each device one by one to see if it can discharge or absorb energy within the specified duration, and at the same time check whether the current charge-discharge power exceeds a certain predetermined power upper limit. This upper limit is usually set by combining the equipment rated value provided by the manufacturer and the on-site empirical value. Suppose the rated power is set to 400 kW, and through multiple on-site monitoring, it is found that overheating or voltage fluctuations are likely to occur when it exceeds 350 kW, then 350 kW is marked as the predetermined standard. If the charge-discharge power of an energy storage device exceeds 350 kW, the power is adjusted downward in the strategy. Finally, in combination with the time period corresponding to the difference between the photovoltaic output and the load demand, evaluate whether each energy storage device can effectively fill or absorb energy during this time period. Mark the energy storage devices that meet the requirements as the targets that can participate in scheduling, and allocate the available capacity and conversion efficiency to the sub-items of the corresponding time periods. After forming a clear schedulable list, through summarizing and comparing the difference size, the remaining capacity of the energy storage and the expected charge-discharge efficiency, finally generate the energy distribution adjustment strategy.

[0120] According to the energy distribution adjustment strategy, the system will adjust the charge-discharge plan according to the target power and specific time periods allocated to each energy storage device. First, retrieve the remaining power of each energy storage device in the current scheduling cycle in the record and compare it with the available capacity interval set previously. If it has been determined to be able to participate in scheduling, issue commands in sub-periods according to the corresponding charge-discharge time periods in the energy distribution adjustment strategy. During the control command execution stage, detect the difference between the output power of the device and the power generation power on the photovoltaic side again. If the output of some devices exceeds the pre-set power limit interval, lower the allocated power of the device in the command, and track and record the final discharge current or charge current value of this process in real time. When the power generation on the photovoltaic side fluctuates or the load demand is unstable, the remaining capacity and the current power demand will be updated in the next scheduling cycle, and then the energy output ratio will be redistributed, while ensuring that there is no energy storage system exceeding the set rate limit. When all energy storage devices have executed the allocated power output or absorption commands and confirm that the power curve does not conflict with the pre-set safety threshold, synchronize the corresponding power ratio to the overall state of the photovoltaic power generation and energy storage system, and finally generate the corrected energy flow direction.

[0121] The steps for obtaining the optimized energy configuration are as follows:

[0122] Based on the corrected energy flow direction, extract the photovoltaic power generation, the charge and discharge status of the energy storage device, and the real-time load demand parameters to generate an energy flow optimization parameter set;

[0123] According to the energy flow optimization parameter set, calculate the dynamic energy flow regulation coefficient. The calculation formula is:

[0124]

[0125] where D is the dynamic energy flow regulation coefficient, Q i is the real-time photovoltaic power generation, E i is the current output power of the energy storage device, L i is the real-time load demand, C i is the remaining capacity of the energy storage device, R i is the operating temperature of the photovoltaic device, F i is the battery aging factor of the energy storage device, and m is the number of sampling points;

[0126] Based on the dynamic energy flow regulation coefficient, combine the real-time load demand and the maximum charge and discharge capacity of the energy storage device to allocate the photovoltaic output and the energy storage charge and discharge ratio to generate the optimized energy configuration.

[0127] Specifically, based on the corrected energy flow direction, first read the photovoltaic power generation from the records and check it against the actual charge and discharge of the energy storage device. Organize the load demand corresponding to each time period and map it to the photovoltaic power generation and the energy storage charge and discharge status under the same time reference. By comparing the distribution of the photovoltaic power generation within the range of 0 kW to 300 kW, at the same time, compare the output power of the energy storage device with the range of 0 kW to 500 kW and check for any abnormalities exceeding 500 kW. For the load demand, compare the monitored value with the range of 0 kW to 800 kW to identify possible overloads. Merge all the data and index it by time to form a comparable combination of photovoltaic power generation, energy storage charge and discharge, and load demand. In order to take into account the update frequencies of different devices, also check the resolution differences of the timestamps. For example, some devices update data every 10 seconds, and some devices update data every 5 seconds, then use interpolation to unify to a 5-second time base. At the same time, maintain the upper and lower boundary detection of the power values during the interpolation process. If any abnormal values such as negative values or values higher than 800 kW are found in the interpolation results, mark them. Incorporate the marked information into the quality inspection list. After completion, conduct a secondary check on all the valid data on the same time line. After determining that there are no obvious deviations, it constitutes the final energy flow optimization parameter set.

[0128] The advantage of the formula is that by comprehensively considering multi-dimensional factors such as the real-time power generation power of photovoltaic, the difference between the current output power of the energy storage device and the real-time load, the temperature of the photovoltaic device, the remaining capacity of the energy storage device, and the battery aging factor, it uniformly quantifies them into the same dynamic adjustment coefficient. With the fourth root, it can balance the impacts brought by various differences to a certain extent, making the system more sensitive to the energy flow changes under high load or high temperature conditions.

[0129] Q i The acquisition steps of Q are as follows: This parameter represents the real-time power generation power of photovoltaic, and its value usually ranges from 0 kW to 300 kW or higher. The specific range depends on the layout capacity of the on-site photovoltaic modules. In actual applications, it is necessary to set to collect the photovoltaic output every 10 seconds or even 5 seconds. The accurate power value is obtained through the power metering device installed on-site or the output record of the inverter. To ensure data validity, abnormal situations with negative power readings or exceeding the preset rated capacity will be excluded first. Then, the values within the normal range are accumulated and aligned with the time stamp. Finally, the power values at these discrete time points are recorded in sequence to form Q1, Q2, …, Q m , for example, there are several acquisitions in a day. If it is detected that the photovoltaic power at 10:00 is 120 kW and the photovoltaic power at 10:05 is 150 kW, the corresponding parameters can be set as Q1 = 120, Q2 = 150, and continue to supplement the subsequent moments until a complete sequence is formed.

[0130] E i The acquisition steps of E are as follows: This parameter is the current output power of the energy storage device, and the common range is from 0 kW to 500 kW. It is obtained through the power metering device of the energy storage system on-site. Each energy storage device has its power sensor or metering circuit. The operation personnel retrieve these output values multiple times in a day and correspond them to the time. Generally, when the energy storage system discharges, E i being positive represents the output power. When absorbing energy (charging), it can be set to negative or the charge and discharge modes are marked separately. In actual use, the charging power and the discharging power are often stored in different fields respectively, but in the formula, the absolute value of the output power needs to be taken or segmented processing is required. To maintain consistency, it is recommended to record the discharging state with a value greater than 0 as E i , and when calculating the difference with the load for the charging state less than 0, the sign correction is done again. For example, for a certain device, if the discharging power of 200 kW is read at 10:00, then E1 = 200. If it is read that it is charging at a rate of 50 kW at 10:10, it may be recorded as -50 and converted according to the requirements during calculation.

[0131] L iThe acquisition steps are as follows: This parameter represents the real-time load demand, indicating the power required by the on-site load at a certain moment. The data acquisition method is similar to recording the power meter or current-voltage sensor installed at the load end. According to the formula Power = Voltage × Current × Power Factor, the actual load power at this moment can be obtained. The range of this parameter may be higher than the photovoltaic power and the energy storage device power. For example, in industrial applications, it can reach 0 kW to 800 kW. It is necessary to first eliminate invalid data, such as situations far exceeding 800 kW or below 0 kW, and keep the same benchmark as the photovoltaic and energy storage data in combination with the timestamp, so that this difference can be accurately found in the formula. i -L i |This difference. If the load demand at a certain moment is 600 kW and the output power record of the energy storage device is 350 kW, then E i can be set to 350, and L i can be set to 600, and subsequent difference operations can be performed.

[0132] R i The acquisition steps are as follows: This parameter represents the operating temperature of the photovoltaic device, usually in degrees Celsius (°C). It is obtained by arranging temperature sensors on the backplane or near the photovoltaic module and combining the temperature readings of the inverter or the operation and maintenance system. It is collected every fixed time (such as 5 minutes). The common range is from -10 °C to 75 °C. If the temperature acquisition value exceeds 75 °C, an abnormal mark needs to be made and it is necessary to check whether the module is overheated. It should be noted that the temperature value is a non-numerical raw measurement, but it will be used in the formula operation. Therefore, it is necessary to ensure that the basic dimension is in degrees Celsius and the value is within a specific range. For the sake of uniformity, when the temperature of the photovoltaic panel is monitored to be 45 °C at a certain time point, then R i is set to 45, and subsequent square root operations can be directly performed. If the temperature remains around 60 °C for a long time, it will have a greater impact on both the photovoltaic output and the system safety. The operation personnel will use this to judge whether forced cooling is required.

[0133] C i The acquisition steps are as follows: This parameter is the remaining capacity of the energy storage device, measured in kWh. It is necessary to conduct a comparative discharge test every day or record the current state of charge through the built-in control system and combine the rated capacity of the device. For example, if a certain energy storage battery is rated at 500 kWh and the system monitors that 210 kWh remains after one day of operation, then 210 is assigned to C i , so that it can be compared with the output power and load demand at its corresponding timestamp. At the same time, during operation, if the capacity is lower than 30 kWh or close to 0, it will be regarded as the operation limit. For example, when it is specifically detected that the remaining capacity at a certain moment is 150 kWh and the capacity is monitored to continuously decrease to 50 kWh at subsequent moments, they are all recorded in the corresponding C1, C2, ….

[0134] F iThe acquisition steps for this parameter are as follows: This parameter is the battery aging factor of the energy storage device, generally represented by a decimal between 0 and 1. 0 represents no obvious aging, and 1 represents severe aging. On-site, long-term monitoring of the device is required and combined with information such as the number of cycles and the capacity attenuation curve to calculate. For example, during the 300-hour observation stage of continuous device operation, the ratio of the actual available capacity to the rated capacity is recorded multiple times, and it is found that the current aging degree of a certain device is about 0.35, and then 0.35 is assigned to F. i , After evaluating each device in the same batch one by one, a sequence of aging factors is obtained. If a device is used for a longer time, its aging factor can gradually rise to 0.7 or even higher.

[0135] The acquisition steps for m are as follows: It is the number of sampling points, indicating how many discrete time points are sampled in one operation cycle. Generally, it is determined by the sampling frequency and the total monitoring duration of the on-site monitoring system. For example, if samples are taken every 5 minutes within 1 hour, 12 sampling points can be obtained. There are also application scenarios with higher frequencies. If samples are taken every minute, 60 sampling points can be obtained in 1 hour, and m can be determined according to requirements.

[0136] Calculation process:

[0137] First step, collect and list the corresponding Q i , E i , L i , R i , C i , F i values. For example, set m = 3, and select data for three time periods respectively. Record Q1 = 100kW, E1 = 80kW, L1 = 130kW, R1 = 40°C, C1 = 200kWh, F1 = 0.3 at the first time period;

[0138] Second step, record Q2 = 150kW, E2 = 120kW, L2 = 160kW, R2 = 45°C, C2 = 180kWh, F2 = 0.35 at the second time period, and record Q3 = 110kW, E3 = 90kW, L3 = 90kW, R3 = 38°C, C3 = 220kWh, F3 = 0.28 at the third time period;

[0139] Third step, calculate the numerator and denominator for each time period respectively. For example, the numerator for the first time period is

[0140] The denominator is C1 + F1 = 200 + 0.3 = 200.3, and the ratio of the two is approximately 316230 / 200.3 ≈ 1577.76. Similarly, the numerator for the second time period can be calculated The denominator is 180 + 0.35 = 180.35, and the ratio is approximately 402492 / 180.35 ≈ 2231.81. The numerator for the third period The denominator is 220 + 0.28 = 220.28, and the ratio is 0;

[0141] In the fourth step, sum the three values: 1577.76 + 2231.81 + 0 = 3809.57, and then find the fourth root: First, calculate the square root Then take the square root of 61.73 to get

[0142] This result indicates that the energy flow dynamic regulation coefficient is approximately 7.86 at this time. If this value rises significantly to 10 or higher, it means that the imbalance degree between the current photovoltaic power generation, energy storage output, and load demand is relatively large. The system can be re-optimized by reducing the energy storage discharge or adjusting the photovoltaic power generation input, etc. If the coefficient is close to 0, it means that the actual power is highly correlated with the load demand and the temperature burden is small, and the system has a low dynamic regulation pressure on the energy flow. Based on the energy flow dynamic regulation coefficient, it is necessary to read the current values of the real-time load demand and the maximum charge and discharge capabilities of the energy storage devices respectively. For example, compare the load demand range with the interval of 0 kW to 800 kW, and compare the maximum energy storage discharge capacity with the interval of 0 kW to 500 kW. If it is detected that the working temperature of a certain energy storage device is too high, the maximum discharge capacity will be automatically reduced to below 300 kW. After determining the safe discharge or charge capacity of each energy storage device, match the photovoltaic side output power according to the time stamp. If the energy flow dynamic regulation coefficient is high at this time, allocate a higher power ratio to the energy storage device and the photovoltaic output. If the energy flow dynamic regulation coefficient is low, it means that the scheduling can be maintained within a medium power range. When the allocation values of each device all meet the requirement of not exceeding the previously determined maximum capacity range and the system does not show continuous high temperature or voltage fluctuations, the photovoltaic output and energy storage power for the current time period are proportionally allocated and integrated in the record, and finally an optimized energy configuration can be generated.

[0143] The steps to obtain the operation result are as follows:

[0144] Based on the optimized energy configuration, call the parameters of photovoltaic power generation, the charge and discharge status of the energy storage device, and the real-time load demand to generate a set of microgrid operation status parameters;

[0145] According to the set of microgrid operation status parameters, analyze the energy balance of each device under different load levels, extract the energy deviation values of photovoltaic power generation, energy storage output, and load consumption, and compare them with the real-time working status and standard status of the device to generate a microgrid stability assessment result;

[0146] Based on the results of the microgrid stability assessment, calculate the energy flow consistency under the current operating state, and conduct a comparative analysis of the operation response delay of each device and the energy compensation for load changes to generate the operation results.

[0147] Specifically, based on the optimized energy configuration, first, correspond the photovoltaic power generation by time period to the working records of the energy storage device and perform index matching for the load demand in each period. For the photovoltaic power data, point-by-point comparison can be carried out in the range of 0 kW to 300 kW. If it is found to exceed 300 kW, it is marked as abnormal and whether there is a hardware overload condition is checked. At the same time, read the actual charge and discharge power values of the energy storage device every hour on the same day. By comparing the range of 0 kW to 500 kW, determine whether there is a situation of output overlimit. When summarizing these power data, ensure that the acquisition time interval is consistent with that of the load demand. If the storage update times are not unified, use timestamp interpolation to align. When the interpolation result is negative or extremely high, re-check the hardware monitoring log. After confirmation, register the current power corresponding to each device, the photovoltaic power generation, and the real-time load demand one by one in the same parameter table. Then, combine the photovoltaic power, energy storage power, and load demand at the same moment according to the device identification, and mark the records that exceed the predetermined power threshold. The predetermined power threshold can be determined according to the distribution median of the daily monitoring data plus the margin reserved in the device manual. For example, set the photovoltaic power threshold at 300 kW and the energy storage power threshold at 500 kW. Finally, organize all the matched power and demand data into the operation records at the corresponding moments, and combine the adjustment ratio information obtained in the previous steps to form a set of microgrid operation state parameters.

[0148] According to the operating state parameter set of the microgrid, the photovoltaic power generation and the output power of the energy storage device are divided into intervals under different load ranges. For example, the load is divided into four intervals: 0 kW to 200 kW, 200 kW to 400 kW, 400 kW to 600 kW, and 600 kW to 800 kW. The corresponding photovoltaic output and energy storage output values within the corresponding time periods are matched, and the values are aggregated and then the difference between the load consumption and the sum of the two is compared. This difference is defined as the energy deviation value and compared with the actual working state of the device, and then compared with the standard state of the device. The standard state is usually set by combining the rated working parameters publicly announced by the manufacturer and the normal range recorded during daily inspections. For example, the power of a certain device should be maintained in the range of 200 kW to 250 kW under the rated current and rated voltage conditions. When it is detected that the current device power exceeds 250 kW or is lower than 0 kW, it can be classified as abnormal. The moment corresponding to the difference is recorded in the abnormal list. If the energy deviation in several consecutive time periods is much higher than the allocation value of the device under normal conditions, key investigations are required. After statistically analyzing the deviations of each device, a three-way comparison of photovoltaic power generation, energy storage output, and load consumption is carried out to find the interval with the most significant deviation and record its duration. Finally, a detailed review of the work records of all devices is carried out. For example, for the interval where the temperature is between 40°C and 60°C, it is determined that the device is operating at a relatively high temperature, and it is compared with the rated working temperature, such as 45°C, to determine whether it is within the acceptable range. After summarizing all deviations, temperature, power, and load information, the microgrid stability assessment results are generated and sorted out.

[0149] Based on the microgrid stability assessment results, the operation response delay of the device in each time period is extracted and compared with the energy compensation data required for the load side demand change. When calculating the delay, first count the time difference between the sudden change in load demand and the completion of the energy storage output or photovoltaic power generation adjustment, and compare this time difference with the previously established delay reference value. This delay reference value is obtained by statistically analyzing multiple records in the actual work log of the device. For example, select similar load sudden increase scenarios from the past few days and observe the average time required for the energy storage device to complete the power increase. If the average time is 30 seconds, then set 30 seconds as the reference value. Check whether the current time period exceeds this reference value. If it exceeds, it is regarded as a slow response. At the same time, perform integral calculation on the power change curve to obtain the energy compensation amount. This energy compensation amount can be achieved on-site by the energy storage device adding output to the load side. After statistically analyzing the response status of all devices in the same time period, it is judged whether the demand for instantaneous peak load is met. When the overall comparison is completed, it is integrated into a record including whether the response delay exceeds the standard and whether the energy compensation amount is satisfied, and it is superimposed and verified with the comprehensive curves of photovoltaic power generation and energy storage devices. Finally, all inspection results are summarized and combined with the current energy flow consistency to form the state assessment content of the overall operation of the microgrid at the current moment, and the operation results are obtained.

[0150] The steps to obtain the adjusted system state are as follows:

[0151] According to the operation results, combined with the output power of photovoltaic power generation equipment, the remaining capacity and response rate of energy storage equipment, the dynamic balance between energy supply and load demand is analyzed, the energy storage charging and discharging plan and photovoltaic power output adjustment are executed, and dynamic adjustment control instructions are generated;

[0152] Based on the dynamic adjustment control instructions, the operating status of each device after adjustment is analyzed, the matching of the energy output of photovoltaic power generation and energy storage equipment with the load demand is verified, the energy flow is redistributed, and the adjusted system status is generated.

[0153] Specifically, according to the operation results, combined with the output power of photovoltaic power generation equipment, the remaining capacity and response rate of energy storage equipment, first summarize the photovoltaic power generation of the day in the range of 0kW to 300kW time period by time, and check whether there is a collection record greater than 300kW. If found, the record is marked as overlimit and checked whether it is a collection error or equipment abnormality. Then read the remaining capacity data of the energy storage device, compare the 0kWh to 500kWh interval to identify whether the device is in a low capacity or sufficient capacity state, and record the response rate of the energy storage device at the same time. The response rate is the average value obtained by several rapid discharge tests on the energy storage device. For example, the energy storage device is repeatedly increased from a discharge rate of 20kW to a discharge rate of 200kW, and the required time is recorded respectively. The average is taken to obtain the response rate benchmark. When the device response speed is significantly different from the current load demand growth rate, it is necessary to allocate a longer transition period to the energy storage device. Check the peak load demand and power consumption cycle in different time periods, and match them with the power output capacity of photovoltaic equipment and energy storage equipment. If it is found that the demand in a certain period exceeds the pre-set safety threshold, the safety threshold is established in combination with the historical maximum load data and margin assessment. For example, the highest load in the past month is 450kW and 50kW margin is added to set it to 500kW. In this period, it is necessary to link multiple energy storage devices and photovoltaic joint output. After checking the remaining capacity and response rate of each energy storage device, the charging and discharging plans are sorted from large to small according to the remaining capacity, and the photovoltaic power output is compared with the energy storage power distribution. If the photovoltaic power itself is lower than the demand gap, the discharge power is further increased at the energy storage end, but it needs to be kept below the power upper limit and the output distribution is adjusted in combination with the response rate. After completion, the specific power issuance instructions of each device are recorded and the time index is spliced for execution within this scheduling cycle, and finally a dynamic adjustment control instruction is formed.

[0154] Based on the dynamic adjustment control instruction, the power output value of each device after execution is checked against the target power set in the previous instruction. When it is found that the output value of a certain device deviates too much from the target value, it is reconfirmed whether it is affected by capacity limitation or temperature. The temperature is set to 0℃ to 45℃ as the normal working range. If it exceeds 45℃, it is considered as high temperature operation and needs to be derated. At the same time, check whether the consumption distribution on the load side has instantaneous peak or segment drop and record the peak duration. Compare the peak duration with the pre-calculated energy storage discharge duration. If the energy storage discharge can support the duration, continue to maintain the original instruction, otherwise add additional points. Allocate devices with more sufficient capacity. When multiple devices are discharging together, evaluate the power stability recorded every few minutes. If a large power difference is found between the devices, a secondary correction allocation is performed. Finally, the photovoltaic power generation data is retrieved in parallel. When the photovoltaic power reaches a certain level, the discharge of the energy storage device can be reduced accordingly. This process is iterated several times together with the response rate and capacity margin to balance the energy flow. After confirming that all devices are within an acceptable range and the load demand is supplied with relatively uniform energy, a new power allocation instruction is output for each device and the current scheduling completion time is recorded, finally generating the adjusted system state.

Claims

1. A microgrid energy management and control system based on energy storage and distributed photovoltaic, characterized in that, The system includes: A load prediction and scheduling module that collects real-time power grid data, analyzes the current load and photovoltaic power generation, performs time series analysis, and generates real-time load prediction data; based on the real-time load prediction data, adjusts the charging and discharging plan of the energy storage device to generate an energy storage scheduling strategy. A causal inference fault diagnosis module that uses the real-time load prediction data to construct a causal chain, identifies key influencing factors in photovoltaic operation, and generates a fault warning signal; performs feedforward correction on the warning signal, adjusts the energy distribution strategy, and generates a corrected energy flow direction. A real-time energy optimization module that uses the corrected energy flow direction to execute a model predictive control algorithm to optimize and regulate the energy flow in the microgrid and generate an optimized energy configuration. A system stability monitoring module that monitors the operating state of the entire microgrid based on the optimized energy configuration, performs stability analysis, and generates an operating result; dynamically adjusts the deviation in the operating result to obtain an adjusted system state.

2. The microgrid energy management and control system based on energy storage and distributed photovoltaic according to claim 1, characterized in that, The steps for obtaining the real-time load prediction data are as follows: Collect real-time power grid data, including photovoltaic power generation, load current, and voltage, to obtain a real-time operating state data set of the power grid. According to the real-time operating state data set of the power grid, calculate the real-time load correlation evaluation value, and the calculation formula is: Among them, CR is the real-time load correlation evaluation value, is the photovoltaic power generation, I i is the real-time load current, U i is the real-time voltage, P i is the predicted load current, rm is the number of sampling points, and max(P) represents the maximum value of the load current; According to the real-time load correlation evaluation value, combined with a time series model, deduce the load demand trend in the future time period to generate real-time load prediction data.

3. The microgrid energy management and control system based on energy storage and distributed photovoltaic according to claim 1, characterized in that, The steps for obtaining the energy storage scheduling strategy are as follows: Based on the real-time load prediction data, obtain the charging and discharging state of the current energy storage device to obtain an energy storage device state parameter set. According to the energy storage device state parameter set, calculate the energy storage adjustment adaptability of the energy storage device, and the calculation formula is: Among them, ED is the energy storage regulation adaptability, is the charge and discharge power of each energy storage device, is the real-time load prediction data, is the current load capacity of the energy storage device, T j is the remaining capacity of the energy storage device, N is the total number of energy storage devices, and k is the index value representing different energy storage devices; According to the energy storage adjustment adaptability, combined with the maximum charging and discharging rate of the energy storage device, optimize the charging and discharging plan of the energy storage device to generate an energy storage scheduling strategy.

4. The microgrid energy management and control system based on energy storage and distributed photovoltaic according to claim 1, characterized in that, The steps for obtaining the fault warning signal are as follows: Based on the real-time load prediction data, use the Pearson correlation coefficient to evaluate the correlation between data, combine expert knowledge to determine the causal relationship between data, and construct a causal chain. Based on the causal chain, calculate the comprehensive influence index, and the calculation formula is: where ZI is the comprehensive influence index, is the efficiency of the photovoltaic component, M i is the maintenance index of the photovoltaic component, A i is the absolute value of the difference between the photovoltaic output and the load demand, and n is the number of factors; Based on the comprehensive influence index, if the influence index exceeds the preset threshold, it is identified as a fault source and a fault warning signal is generated.

5. The microgrid energy management and control system based on energy storage and distributed photovoltaic according to claim 1, characterized in that, The steps for obtaining the corrected energy flow direction are as follows: Based on the fault warning signal, calculate the difference between the photovoltaic output deviation and the load demand to generate an energy distribution abnormal parameter set. According to the energy distribution abnormal parameter set, analyze the remaining capacity, charging and discharging power, and conversion efficiency of the energy storage device, evaluate the ability of the energy storage system to respond to real-time load demand, and combine the difference between the photovoltaic output and the load demand to generate an energy distribution adjustment strategy. According to the energy distribution adjustment strategy, execute the adjustment of the charging and discharging plan of the energy storage device, modify the energy output ratio of photovoltaic power generation and the energy storage system, and generate a corrected energy flow direction.

6. The microgrid energy management and control system based on energy storage and distributed photovoltaics according to claim 1, characterized in that, The steps for obtaining the optimized energy configuration are as follows: Based on the corrected energy flow direction, extract parameters such as photovoltaic power generation, charging and discharging state of the energy storage device, and real-time load demand to generate an energy flow optimization parameter set. According to the energy flow optimization parameter set, calculate the dynamic regulation coefficient of energy flow. The calculation formula is as follows: Among them, D is the energy flow state adjustment coefficient, Q i is the real-time photovoltaic power generation, E i is the current output power of the energy storage device, L i is the real-time load demand, C i is the remaining capacity of the energy storage device, R i is the operating temperature of the photovoltaic device, F i is the battery aging factor of the energy storage device, and m is the number of sampling points; Based on the dynamic regulation coefficient of energy flow, combined with the real-time load demand and the maximum charge and discharge capacity of the energy storage device, allocate the proportion of photovoltaic output and energy storage charge and discharge to generate the optimized energy configuration.

7. The microgrid energy management and control system based on energy storage and distributed photovoltaic according to claim 1, characterized in that The steps for obtaining the operation result are as follows: Based on the optimized energy configuration, call the parameters of photovoltaic power generation, the charge and discharge status of the energy storage device, and the real-time load demand to generate a set of microgrid operation status parameters; According to the set of microgrid operation status parameters, analyze the energy balance of each device under different load levels, extract the energy deviation values of photovoltaic power generation, energy storage output, and load consumption, and compare them with the real-time working status and standard status of the device to generate the microgrid stability assessment result; Based on the microgrid stability assessment result, calculate the energy flow consistency in the current operation state, and compare and analyze the operation response delay of each device and the energy compensation for load changes to generate the operation result.

8. The microgrid energy management and control system based on energy storage and distributed photovoltaic according to claim 1, characterized in that, The steps for obtaining the adjusted system state are as follows: According to the operation result, combined with the output power of the photovoltaic power generation device, the remaining capacity and response rate of the energy storage device, analyze the dynamic balance of energy supply and load demand, execute the energy storage charge and discharge plan and adjust the photovoltaic power output to generate the dynamic adjustment control instruction; Based on the dynamic adjustment control instruction, analyze the operation status of each device after adjustment, verify the matching of the energy output of the photovoltaic power generation and energy storage devices with the load demand, and execute the redistribution of the energy flow direction to generate the adjusted system state.

Images