Big Data-Based Methods for Power Distribution System Operation Optimization and Energy Efficiency Control
By using big data-based methods to collect and label multi-source data streams, calculate the comprehensive efficiency distance ratio, generate quadruplet trigger signals, and construct a weighted multi-objective optimization model, the problem of difficulty in coordinating the optimization of safety and energy efficiency objectives in power distribution systems is solved, and the coordinated optimization and adaptive control of safety and energy efficiency are realized.
Patent Information
- Application Number
- CN202611478306.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-22
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies, it is difficult to coordinate and optimize the safety and energy efficiency goals of power distribution systems, resulting in suboptimal control oscillations and strategies or delayed responses, making it impossible to adaptively adjust according to the real-time status of the system.
By adopting a big data-based approach, multi-source data streams are collected and labeled to extract safety margin and energy efficiency level features, calculate the comprehensive efficiency distance ratio, generate a four-tuple structured trigger signal, construct a weighted multi-objective optimization model, carry out collaborative control, and adjust the acquisition strategy of the sensing terminal.
It achieves synergistic optimization of safety and energy efficiency goals, avoids control oscillations and suboptimal strategies, improves the accuracy and adaptability of control effects, and saves data acquisition resources.
Smart Images

Figure CN123121518A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power distribution system operation control technology, and more specifically to a method for power distribution system operation optimization and energy efficiency control based on big data. Background Technology
[0002] As the hub connecting the transmission network and end users, the operation optimization and energy efficiency control of the power distribution system are key links to ensure power supply reliability and improve energy utilization efficiency. With the large-scale access of distributed photovoltaic, energy storage systems and flexible loads in the power distribution network, the uncertainty and complexity of the system operation status have increased significantly, and the coupling contradiction between safety constraints and energy efficiency targets has become increasingly prominent. Based on big data technology, the refined operation management of the power distribution system has become an important research direction in the field of smart grids.
[0003] In existing technologies, power distribution system operation optimization and energy efficiency control are usually treated as two relatively independent problems: operation optimization focuses on safety indicators such as voltage and reactive power regulation and topology reconfiguration, with the core goal of eliminating over-limits and improving power quality; while energy efficiency control focuses on economic indicators such as reducing network losses, renewable energy consumption and demand-side management, with the goal of maximizing energy utilization efficiency.
[0004] While this type of decoupling approach is convenient for engineering implementation, it ignores the strong coupling relationship between the two at the physical level. Any adjustment action in the operation optimization will change the system's energy efficiency level, and the implementation of energy efficiency control must also be completed through the reallocation of the operating state. When safety and energy efficiency goals conflict, multi-objective optimization methods with fixed weights or manually preset rules are difficult to adaptively adjust the emphasis according to the real-time situation of the system, which can easily lead to control oscillations, suboptimal strategies, or delayed responses.
[0005] Therefore, there is an urgent need for a power distribution system operation optimization and energy efficiency control method that can incorporate safety margin and energy efficiency level into a unified quantitative evaluation framework and adaptively switch control strategies according to the real-time coupling situation of the system, so as to achieve the optimal synergy between safety and energy efficiency objectives, while endowing the system with the ability to continuously learn and evolve from historical control experience. Summary of the Invention
[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a method for power distribution system operation optimization and energy efficiency control based on big data, which can effectively solve the problem that safety and energy efficiency objectives are difficult to optimize in a coordinated manner in the background technology.
[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for power distribution system operation optimization and energy efficiency control based on big data, including: S1. Based on the current operating scenario and energy efficiency status of the target power distribution system, collect and synchronously label multi-source data streams with tags that characterize the evolution trend of its safety-energy efficiency coupling situation.
[0008] S2. Extract safety margin features and energy efficiency level features from the tagged multi-source data stream, substitute them into the comprehensive efficiency distance ratio for quantitative evaluation, determine the current safety-energy efficiency coupled situation mode of the target power distribution system, and generate a four-tuple structured trigger signal carrying the target weight vector, constraint level and dominant target identifier.
[0009] S3. Analyze the structured trigger signal of the quadruple, extract the target weight vector, constraint level and dominant target identifier, and construct a weighted multi-objective optimization model.
[0010] The system generates a set of coordinated control instructions for adjustable resources within the target power distribution system and executes differentiated control. Simultaneously, it adjusts the acquisition strategy of the sensing terminal based on the situation mode identifier code in the trigger signal.
[0011] S4. Obtain the actual operating data of the target power distribution system after control, recalculate the comprehensive efficiency distance ratio and compare it with that before control, and quantitatively evaluate the control effect.
[0012] Preferably, the process of collecting and synchronously labeling the corresponding multi-source data streams includes: The sensing terminals deployed at each key node of the target power distribution system collect basic electrical quantities such as feeder voltage, current, transformer load rate and key node line loss rate in low frequency mode under normal conditions. Anomaly probes are deployed in the sensing terminals to calculate the deviation of each key parameter from its historical steady-state baseline.
[0013] When the deviation of any key parameter exceeds the adaptive threshold, the acquisition mode is switched to high-frequency reconnaissance mode. Based on the preset information blind spot model, the system actively requests related data from the data platform to obtain meteorological environment data of the target power distribution system, bus voltage data of the upper substation, power flow data of adjacent feeders, and operating condition data of distributed photovoltaic and energy storage in the same area.
[0014] The multi-source data streams that are aggregated to the data platform are used to calculate the data performance contribution factor for each data stream through a synchronous annotation engine, and then divided by the time delay from data collection to data storage for timeliness correction.
[0015] Based on the comparison between the calculation results and the preset performance contribution threshold, each data stream is labeled as one of three categories: high value - performance sensitive, routine - energy efficiency concern, or low priority - safety redundancy, and then packaged and output in a unified time-aligned format.
[0016] Preferably, the process of extracting the safety margin characteristics and energy efficiency level characteristics of the target power distribution system includes: From the tagged multi-source data stream output by S1, obtain the current effective voltage value, the lower limit of the voltage, the current active power transmitted, and the rated capacity of the connected distribution transformer for each key node where the sensing terminal is deployed. Calculate the degree to which the voltage of each node deviates from its lower limit and weight it according to the ratio of the load of that node to the transformer capacity. Take the average value for all key nodes to obtain the load-weighted voltage stability margin as a safety margin characteristic quantity.
[0017] Simultaneously, the current comprehensive line loss rate of the target power distribution system is obtained, as well as the historical best line loss rate that is within the same range as the current load rate and is selected from the historical statistics database. The line loss deviation efficiency index is obtained by dividing the historical best line loss rate by the current comprehensive line loss rate, which serves as a characteristic quantity of energy efficiency level.
[0018] Preferably, the process of quantitatively assessing the precise status coordinates of the current system in the safety-energy efficiency four-quadrant includes: The extracted load-weighted voltage stability margin is used as the safety dimension input, and the line loss deviation efficiency index is used as the energy efficiency dimension input to calculate the comprehensive efficiency distance ratio of the target power distribution system.
[0019] The calculation logic of the comprehensive efficiency distance ratio is as follows: the ratio of the safety margin characteristic quantity to the energy efficiency level characteristic quantity is used as the main term, and then multiplied by a logarithmic correction term with the distance between the safety margin characteristic quantity and the preset safety margin critical threshold as the independent variable.
[0020] The calculated comprehensive efficiency distance ratio is compared with the preset first threshold, second threshold and third threshold respectively. When the comprehensive efficiency distance ratio is greater than or equal to the third situation judgment threshold, the target power distribution system is determined to be in a safe, relaxed and efficient state.
[0021] When the overall efficiency distance ratio is between the second and third situation determination thresholds, the target power distribution system is determined to be in a safe, relaxed, and inefficient state.
[0022] When the comprehensive efficiency distance ratio is between the first situation determination threshold and the second situation determination threshold, the target power distribution system is determined to be in a safe, tense and efficient state.
[0023] When the overall efficiency distance ratio is lower than the first situation judgment threshold, the target power distribution system is judged to be in a state of safety tension and inefficiency.
[0024] Preferably, the process of evaluating the corresponding coupling situation mode based on the calculation results and generating a trigger signal includes: Based on the comparison results of the comprehensive efficiency distance ratio with the first threshold, the second threshold and the third threshold, the current situation mode of the target power distribution system is determined. The situation mode includes four modes: safety tension and inefficiency, safety tension and high efficiency, safety relaxation and inefficiency, and safety relaxation and high efficiency.
[0025] Based on the determined situation pattern, a four-tuple structured trigger signal is constructed. The trigger signal includes four fields: situation pattern identifier code, target priority weight allocation vector, dominant objective function type identifier, and security constraint level identifier.
[0026] The completed trigger signal is sent to the collaborative control solver in the form of a structured message for subsequent steps.
[0027] Preferably, the process of receiving and parsing the quadruple structured trigger signal includes: The collaborative control solver receives the four-tuple structured trigger signal sent by S2 and determines one of the four current situation modes from the situation mode identifier field of the trigger signal.
[0028] Extract safety weights and energy efficiency weights from the target priority weight allocation vector field. Directly assign the safety weight values to the weighting coefficients of the safety targets in the optimization objective function, and directly assign the energy efficiency weight values to the weighting coefficients of the energy efficiency targets.
[0029] When the situational pattern identifier code corresponds to a safe, relaxed, and efficient mode, the control cost weight is extracted from the weight allocation vector and assigned to the weighting coefficient of the control cost target. The sum of the three weighting coefficients is always one.
[0030] Extract the constraint type from the safety constraint level identifier field. When the identifier is a hard constraint, set the safety constraint as a rigid boundary that must be strictly satisfied in the optimization search. When the identifier is a soft constraint, convert the safety constraint into a flexible boundary that can be touched at a penalty cost.
[0031] The target type is extracted from the dominant objective function type identifier field, and the solver is directly indexed to the preset solver strategy template that matches the current situation mode. The solver strategy template includes a safety limit elimination template, a safety margin expansion template, an energy efficiency maximization template, and a control cost minimization template.
[0032] Preferably, the process of generating a set of coordinated control instructions for adjustable resources of the target power distribution system includes: The normalized weight vector, composed of the weighting coefficients of the safety target, the energy efficiency target, and the control cost target, is combined with the safety cost function, the energy efficiency cost function, and the control cost function to form a weighted multi-objective optimization model.
[0033] The safety cost function measures the safety level by the weighted sum of squares of the voltage amplitude and the deviation of the rated voltage at each node under the current control variable.
[0034] The energy efficiency cost function measures energy efficiency by the ratio of total active power loss in the target distribution system to total active power flowing into the outlet.
[0035] The control cost function measures the adjustment cost by the weighted sum of the differences between the current adjustment target value of each controllable resource and the actual execution value of the previous control cycle.
[0036] The constraints are dynamically set based on the resolved constraint level.
[0037] Based on the indexed solution strategy template, a matching solution algorithm is selected. Within the feasible region that satisfies the constraints, the optimal solution of the control variables that minimizes the objective function is searched. The optimal solution is decomposed into the set value of the number of capacitor switching groups, the set value of the energy storage charging and discharging power, the target value of the tap position of the on-load tap changer, and the interruptible load reduction amount. These are then packaged into a set of coordinated control instructions.
[0038] Preferably, the process of adjusting the acquisition strategy of the sensing terminal according to the situational pattern identifier code in the trigger signal includes: After the coordinated control instruction set is executed, the situation mode identification code in the trigger signal is synchronously transmitted to the sensing terminal, and the acquisition strategy is dynamically adjusted according to the four modes indicated by the situation mode identification code: When the situation mode identifier code corresponds to a security-stressed and inefficient mode or a security-stressed and efficient mode, it is determined that the target power distribution system is in the recovery observation period after control. The sensing terminal maintains a high-frequency reconnaissance mode and reduces the adaptive threshold of the abnormal probe.
[0039] When the situational pattern identifier code corresponds to a safe and inefficient mode, it is determined that the target power distribution system is in the active energy efficiency tapping stage, the sensing terminal maintains the high-frequency reconnaissance mode and the probe threshold returns to the normal value.
[0040] When the situation mode identifier code corresponds to a safe, relaxed, and efficient mode, it is determined that the target power distribution system is close to the ideal operating state. The sensing terminal switches from high-frequency reconnaissance mode back to low-frequency economic mode, and the abnormal probe threshold is restored to the default value.
[0041] Preferably, the process of quantifying and evaluating the control effect includes: After obtaining the actual operating data of the target power distribution system after control, the comprehensive efficiency distance ratio after control is recalculated and compared with the comprehensive efficiency distance ratio before control. The effective displacement of the closed loop is obtained by dividing the difference between the comprehensive efficiency distance ratios before and after control by the absolute value of the comprehensive efficiency distance ratio before control.
[0042] By combining the preset closed-loop effective displacement threshold, the control effect is evaluated to determine whether it meets the expected requirements.
[0043] The technical solution provided by this invention has the following advantages compared with the known prior art: 1. By integrating the comprehensive efficiency distance ratio index, the safety margin characteristic quantity and energy efficiency level characteristic quantity, which originally had different dimensions and could not be directly compared, are merged into a single continuous value. This locks the precise status coordinates of the target power distribution system in the "safety-energy efficiency" four quadrant. Starting from the physical coupling relationship, this solution uses a single index to answer the two questions of "how safe is the power distribution system" and "how efficient is the power distribution system", avoiding the arbitrariness of human weighting when scoring multiple indicators.
[0044] 2. The use of quadruplet trigger signals enables adaptive switching of the control strategy, avoiding control oscillations and suboptimal strategies. This scheme automatically generates quadruplet trigger signals containing target weight vectors, constraint levels, and dominant target identifiers based on the range in which the comprehensive efficiency distance ratio falls. When safety is critical, safety weights are high and constraints are hard constraints; when safety is relaxed, energy efficiency weights are high and constraints can be relaxed to soft constraints; when both safety and energy efficiency are optimal, the scheme switches to the minimum control cost mode. This allows the solver to directly call the matching solution parameters and constraint settings without re-judgment or manual intervention when facing different operating scenarios, which helps to fundamentally eliminate the possibility of conflict between safety and energy efficiency objectives at the control execution layer.
[0045] 3. It not only judges whether the control is effective, but also quantifies and verifies the comprehensive efficiency distance ratio displacement before and after each control. It precipitates the effective samples into structured knowledge, which is used to regularly and automatically update the situation judgment threshold and the weight configuration of safety and energy efficiency in each mode. Samples with poor control effect will also be marked and attributed. Once the deviation continuously points to the same control channel, it can also trigger an early warning and manual review, so that the assessment accuracy and control strategy will not degrade due to equipment aging or changes in operating characteristics. On the contrary, it becomes more accurate with the increase of operating time.
[0046] 4. By embedding perceptual intelligence into the data acquisition stage, low-frequency acquisition under normal conditions saves resources. Once the probe detects an anomaly, the information blind spot model immediately drives the related data request, selectively pulling data to fill cognitive gaps. Then, each data point is labeled with a value level through a data effectiveness contribution factor, making the data entering the subsequent situation assessment both concise and high-value. This avoids the problem of uploading massive amounts of data and seriously wasting computing resources in traditional solutions, providing a feasible data foundation for the online real-time control of complex power distribution networks. Attached Figure Description
[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0048] Figure 1 This is a schematic diagram of the implementation steps of the present invention. Detailed Implementation
[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0050] The present invention will be further described below with reference to embodiments.
[0051] Please see Figure 1 The power distribution system operation optimization and energy efficiency control method based on big data includes at least the following: S1. Based on the current operating scenario and energy efficiency status of the target power distribution system, the system actively collects and synchronously labels the corresponding multi-source data streams using pre-installed sensing terminals and data platforms, providing input for the comprehensive situation assessment in S2.
[0052] In one specific embodiment, the process of collecting and synchronously labeling the corresponding multi-source data streams is as follows: the sensing terminals deployed at each key node of the target power distribution system (such as the end of the feeder, the tie switch, the low-voltage side of the distribution transformer, and the grid connection point of the distributed power source) collect basic electrical quantities such as feeder voltage, current, transformer load rate, and key node line loss rate in low-frequency mode under normal conditions (all monitored electrical quantities are within the limit range), and continuously connect the collected data to the data platform.
[0053] Meanwhile, a lightweight anomaly probe is deployed within the sensing terminal. This probe continuously calculates the deviation of each key parameter from its historical steady-state baseline using a sliding time window.
[0054] The historical steady-state baseline is calculated by the data platform based on the same type of time period data of similar historical operating days of the target power distribution system, and is updated on a rolling basis at set time intervals (such as daily).
[0055] When the anomaly probe detects that the deviation of any key parameter exceeds the adaptive threshold, the acquisition mode will be switched from low-frequency mode to high-frequency detection mode.
[0056] After entering the high-frequency reconnaissance mode, based on the preset information blind spot model, a related data request is initiated to the data platform, thereby simultaneously acquiring four types of related data within that time period: meteorological environment data of the target power distribution system coverage area (including temperature, solar irradiance and wind speed, from the meteorological service interface), real-time bus voltage data of the upper-level substation (from the dispatch automation system), power flow data of adjacent feeders that are connected to the target power distribution system (from the distribution automation system), and real-time output data of distributed photovoltaics and state of charge data of energy storage systems within the same distribution area (from the distributed resource management platform).
[0057] After the multi-source data streams are aggregated into the data platform, a performance label is attached to each data stream through a synchronous labeling engine, and the data performance contribution factor of each data stream is calculated.
[0058] Based on the calculation results, the data stream is labeled as one of three categories: "high value - efficiency sensitive", "routine - energy efficiency concern", or "low priority - safety redundancy".
[0059] Through the data platform, the 80th percentile of the data performance contribution factor distribution of all data streams in the target power distribution system within a set historical time period (such as the past month) is taken as the initial value and updated once at the beginning of each month.
[0060] If the data performance contribution factor of a current data stream is greater than or equal to the initial value, then the data stream is marked as "high value - performance sensitive"; if the data performance contribution factor of a current data stream is greater than or equal to half of the initial value, but less than the initial value, then the data stream is marked as "normal - energy efficiency concern"; if the data performance contribution factor of a current data stream is less than half of the initial value, then the data stream is marked as "low priority - safety redundancy".
[0061] After labeling is completed, all labeled multi-source data streams are packaged in a uniform time-aligned format as standardized output.
[0062] Data performance contribution factor for each data stream The calculation formula is: , For each data stream, assign a number index. The value of is a positive integer. This represents the instantaneous percentage deviation of the current voltage from its rated value. This represents the instantaneous percentage deviation of the current line loss rate from its historical steady-state baseline. and These are the preset safety sensitivity coefficient and energy efficiency sensitivity coefficient, respectively. The data latency is obtained by subtracting the data entry timestamp automatically recorded by the data platform when the data is received from the data collection timestamp of the sensing terminal. Let 10 be the infinitesimal positive number we are setting. -6 .
[0063] It should be noted that the specific process by which the probe continuously calculates the deviation of each key parameter from its historical steady-state baseline using a sliding time window is as follows: Taking an industrial feeder with a high proportion of photovoltaics as an example, the probe uses the voltage data at the end of the feeder during the load ramp-up period from 9:00 to 10:00 AM on weekdays of the past 30 similar operating days to calculate the average value and fluctuation range as the historical steady-state baseline for the same period today. At 9:30 AM on a certain day, when the photovoltaic output of the target power distribution system drops sharply due to sudden cloud cover, the probe extracts the newly collected voltage time-series data from the past 15 minutes and finds that its average value has deviated from the baseline value by more than 5%, and this deviation has lasted for 3 sliding window cycles. Therefore, it is determined that the voltage deviation has exceeded the adaptive threshold, triggering a switching of the acquisition mode.
[0064] The adaptive threshold is dynamically determined by the load period type of the target power distribution system. Taking an industrial feeder with a high proportion of photovoltaics in this scheme as an example, the data platform divides the feeder into the "load ramp-up high risk period" from 9 am to 10 am on weekdays based on historical operating data. At this time, the corresponding adaptive threshold is reduced to ±5% of the baseline value, while the adaptive threshold is relaxed to ±12% for the "load low and stable period" from 3 am to 4 am.
[0065] It is obtained by taking the percentage difference between the effective voltage value collected in real time by the sensing terminal at the end of the feeder of the target power distribution system and the rated voltage value of the feeder. The current line loss rate is calculated in real time by the data platform based on the measured power values at both ends of the feeder. Then, the historical steady-state baseline value corresponding to this period is retrieved, and the difference between the two is divided by the steady-state baseline value and the percentage is obtained. The historical steady-state baseline value comes from the statistical average of the line loss rate of the same type of period in the past 30 similar operating days.
[0066] The safety sensitivity coefficient is obtained by dividing the number of voltage over-limit events that actually occurred on the feeder of the target power distribution system in the past year by the total number of operating days in the year and normalizing it to the range of [0, 1]. The data is directly extracted from the historical alarm record database of the power distribution automation system on a monthly basis. The energy efficiency sensitivity coefficient is obtained by dividing the number of months on the feeder that failed to meet the line loss assessment target in the previous year by 12 months and normalizing it. Both are updated by the data platform at the beginning of each year and remain unchanged throughout the year.
[0067] The preset information blind zone model is a dynamically generated anomaly-blind zone mapping matrix. When the lightweight probe detects, for example, a deviation of the "feeder end voltage" from the baseline, the model automatically determines that: at present, the voltage amplitude data alone cannot simultaneously infer the specific contributions of the three dimensions that caused this deviation: upstream power supply fluctuations, intermittent power output of new energy sources on this line, and downstream impact load mutations. These three factors constitute the information blind zone at the current moment. At this time, the model automatically generates accurate data request instructions to fill the blind zone based on the preset causal relationship table. That is, it requests the bus voltage data of the upper substation (excluding the influence of the power supply side) upwards, requests the second-level power output data of the distributed photovoltaic connected to the feeder horizontally (excluding new energy disturbances), and requests the real-time power data of the large end users downwards (to confirm load mutations).
[0068] Taking an industrial feeder with a high proportion of photovoltaics as an example, when the feeder experiences voltage flicker when clouds frequently pass by at midday, the probe detects voltage fluctuations exceeding the limit. The information blind spot model determines that it cannot distinguish whether the cloud disturbance affects the photovoltaic output or whether the equipment in the factory is activated due to an impact. It then triggers a high-frequency reconnaissance mode, simultaneously pulling the photovoltaic inverter output curve and the factory's incoming power curve for that period. Once the two complementary data are aligned, the blind spot is immediately eliminated. Only then can sufficient information be provided to the synchronous labeling engine to accurately calculate the data performance contribution factor and label it with high-value, performance-sensitive tags.
[0069] By embedding perceptual intelligence into the data acquisition stage, low-frequency data acquisition under normal conditions saves resources. Once the probe detects an anomaly, the information blind spot model immediately drives the related data request, selectively pulling data to fill cognitive gaps. Then, each data point is labeled with a value level through a data efficiency contribution factor, making the data entering the subsequent situation assessment both concise and high-value. This avoids the problem of uploading massive amounts of data and seriously wasting computing resources in traditional solutions, providing a feasible data foundation for the online real-time control of complex power distribution networks.
[0070] S2. Extract the safety margin and energy efficiency level characteristics of the target power distribution system from the tagged multi-source data stream output by S1. Substitute the extracted characteristics into the comprehensive efficiency distance ratio calculation formula to quantitatively evaluate the accurate status coordinates of the current system in the safety-energy efficiency four quadrants. Based on the calculation results, evaluate the corresponding coupled status mode and generate the differentiated collaborative control strategy trigger signal required by S3.
[0071] In one specific embodiment, the process of extracting the safety margin characteristics and energy efficiency level characteristics of the target power distribution system is as follows: The total number of key nodes deployed by the sensing terminals is obtained from S1, and the index of each key node is recorded as follows: , , This represents the total number of critical nodes. The value is a positive integer, and the current effective voltage value corresponding to each key node is obtained. Lower limit of voltage Current active power transmitted Rated capacity of the distribution transformer This leads to the calculation of the security margin feature quantity corresponding to the labeled multi-source data stream. : .
[0072] Obtain the current comprehensive line loss rate corresponding to the target power distribution system. and the historical best line loss rate within the same load rate range Then, the energy efficiency level feature quantity corresponding to the labeled multi-source data stream is calculated. : .
[0073] when The closer the value is to zero or becomes negative, the tighter the safety margin of the target power distribution system, and the more likely it is to experience a risk of voltage falling below the lower limit under heavy load conditions; when The closer the value is to 1, the closer the current line loss is to the historical best level, indicating excellent energy efficiency. When the value is below 1, it indicates that there is significant room for the current line loss to worsen, and the need for energy efficiency control is urgent.
[0074] It should be noted that the current effective voltage value of each key node is directly taken from the real-time collection of the sensing terminals deployed at each node in S1, and the tagged voltage data stream output after processing by the synchronous labeling engine; the lower limit of the voltage is calculated by multiplying the minimum allowable voltage percentage specified in the power supply voltage deviation standard of the target power distribution system area by the rated voltage value corresponding to the voltage level of the node; the active power currently transmitted is directly taken from the tagged power data stream of the node in S1; the rated capacity of the distribution transformer connected to the node comes from the equipment ledger database of the target power distribution system and is directly called according to the corresponding transformer nameplate parameters.
[0075] The current comprehensive line loss rate is calculated by the data platform based on the tagged power data streams of the target distribution system feeder at the beginning and end, output by S1, and then divided by the active power at the beginning. The historical best line loss rate within the same load rate range is obtained by selecting the average of the top 5% of samples with the lowest line loss rate in the past 12 months from the historical statistical database of the target distribution system line loss management system. The current load rate is determined in real time by the ratio of the active power at the beginning of the feeder to the rated capacity of the feeder.
[0076] In one specific embodiment, the process of quantitatively assessing the precise status coordinates of the current system in the safety-energy efficiency four-quadrant framework is as follows: based on the safety margin feature quantity and energy efficiency level feature quantity corresponding to the labeled multi-source data stream, combined with the safety margin critical threshold... and smoothing adjustment constant The comprehensive efficiency distance ratio of the target power distribution system in terms of both safety and energy efficiency dimensions was calculated. : .
[0077] Combining the preset first threshold, second threshold, and third threshold, when When the second threshold is greater than or equal to the third threshold, the target power distribution system is determined to be in a safe, relaxed, and efficient state; when the second threshold is less than or equal to the third threshold, the target power distribution system is determined to be in a safe, relaxed, and efficient state. When the threshold is less than the third threshold, the target power distribution system is determined to be in a state of relaxed safety and inefficiency; when the first threshold is less than or equal to the third threshold, the target power distribution system is determined to be in a state of relaxed safety and inefficiency. When the threshold is less than the second threshold, the target power distribution system is determined to be in a safe, tense, and efficient state; when... When the value is less than the first threshold, the target power distribution system is determined to be in a state of safety tension and inefficiency.
[0078] It should be noted that the determination of the critical threshold for safety margin is based on the maximum allowable voltage drop in the power supply voltage deviation standard for the target distribution system area: when the voltage at the end of the feeder just drops to the lower limit of the standard, the average value of the load-weighted voltage stability margin of each key node under this extreme condition is taken. This average value, after being normalized, is the critical threshold for safety margin, representing a unified warning line for the system to enter the safety tension zone. The smoothing adjustment constant is determined by extracting all past safety margin characteristic values falling within the critical neighborhood of 0.03 to 0.07 from the historical operating data of the target distribution system, and calculating the corresponding values of these sampling points. The minimum value, taken as one-tenth of the minimum value and rounded up, is the smoothing adjustment constant.
[0079] The first, second, and third thresholds were determined by substituting the historical operating data of the target power distribution system over the past 12 months into the comprehensive efficiency distance ratio calculation formula for each day and time period before the system was first put into operation. After obtaining the distribution of comprehensive efficiency distance values at all sampling times throughout the year, the three natural quartiles of this distribution were used to determine the thresholds.
[0080] By integrating the comprehensive efficiency distance ratio index, the safety margin characteristic quantity and energy efficiency level characteristic quantity, which were originally of different dimensions and could not be directly compared, are merged into a single continuous value. This locks in the precise status coordinates of the target power distribution system in the "safety-energy efficiency" four quadrant. Starting from the physical coupling relationship, this solution uses a single index to answer the two questions of "how safe is the power distribution system" and "how efficient is the power distribution system", avoiding the arbitrariness of human weighting when scoring multiple indicators.
[0081] In a specific embodiment, the process of evaluating the corresponding coupled situation mode based on the calculation results is as follows: Based on the comprehensive efficiency distance ratio and judgment results of the target power distribution system corresponding to the safety and energy efficiency dimensions, a four-tuple structured trigger signal corresponding to the current situation mode is constructed, and its signal format is as follows: ,in This represents a situational awareness code, with values from 1 to 4, corresponding to four modes: "security tense and inefficient," "security tense and efficient," "security relaxed and inefficient," and "security relaxed and efficient." This is represented as a target priority weight allocation vector under the current mode, including safety weight and energy efficiency weight. This represents the dominant objective function type recommended for S3 optimization, indicated by a text enumeration value. This represents the level of safety constraints during S3 optimization, indicated by enumerated values for hard or soft constraints.
[0082] After the trigger signal is constructed, it is sent to the S3 collaborative control solver in the form of a structured message through the bus interface of the data middle platform.
[0083] It should be noted that the values of each field in the trigger signal are automatically generated based on the situation pattern, and the mapping rule is as follows: when When =1, Choose [0.9, 0.1] (0.9 corresponds to the safety weight, and 0.1 corresponds to the energy efficiency weight). Take "Safety Limit Exceedance Elimination" Take "hard constraints"; when When =2, Take [0.6, 0.4], Take "expanding safety margin" Take "hard constraints"; when When =3, Take [0.1, 0.9], Take "maximizing energy efficiency potential" Take "soft constraints"; when When =4, Take [0.2, 0.2], Choose "minimize control costs" The "soft constraint" is adopted, and the remaining 0.6 weights are allocated to the control cost objective to reflect the core principle of "laissez-faire" in this model.
[0084] The values of safety weight and energy efficiency weight are determined by the situation identification and strategy triggering engine before the target power distribution system is first put into operation. Based on the pairwise comparison and scoring of the relative importance of safety and energy efficiency under four situation modes by operation experts, a judgment matrix is constructed and its eigenvector is solved to obtain the initial benchmark weights of safety weight and energy efficiency weight. Then, this initial weight is substituted into the historical data of the past 12 months, and closed-loop control simulation is performed for each situation mode. The dual optimization objectives are to minimize the annual comprehensive network loss (referring to the total active power loss of the target power distribution system in a complete statistical period due to resistance heating, iron core loss, etc. In this scheme, the calculation scope of this indicator covers the sum of all line losses and transformer losses from the feeder outlet metering point of the upper substation to the low-voltage side outlet metering point of each distribution transformer) and maximize the voltage qualification rate. The final weight values are determined after reverse iterative calibration by a genetic algorithm (this process is a parameter optimization process with the dual optimization objectives of minimizing comprehensive network loss and maximizing voltage qualification rate, which will not be elaborated here).
[0085] The genetic algorithm is used to perform closed-loop simulation reverse calibration on the initial weights obtained by expert scoring under four situation modes. With the dual objectives of minimizing comprehensive network loss and maximizing voltage qualification rate, iteratively searches for the final weight value that optimizes the annual operation of the power distribution system for this specific objective.
[0086] The use of quadruplet trigger signals enables adaptive switching of control strategies, avoiding control oscillations and suboptimal strategies. This scheme automatically generates quadruplet trigger signals containing target weight vectors, constraint levels, and dominant target identifiers based on the range in which the comprehensive efficiency distance ratio falls. When safety is critical, safety weights are high and constraints are hard constraints; when safety is relaxed, energy efficiency weights are high and constraints can be relaxed to soft constraints; when both safety and energy efficiency are optimal, the scheme switches to the minimum control cost mode. This allows the solver to directly call matching solution parameters and constraint settings when facing different operating scenarios without re-judgment or manual intervention, which helps to fundamentally eliminate the possibility of conflict between safety and energy efficiency objectives at the control execution layer.
[0087] S3 receives the four-tuple structured trigger signal output by S2, extracts the target weight vector, constraint level, and dominant target identifier, assigns the extracted weight vector and constraint level to the collaborative optimization solver, constructs a weighted multi-objective optimization model matching the current situation pattern, and solves it online. This generates a collaborative control instruction set for the adjustable resources of the target power distribution system. Based on the generated instruction set, differentiated control is implemented on the adjustable resources of the target power distribution system. Furthermore, the acquisition strategy of the sensing terminal is adjusted according to the situation pattern identifier code in the trigger signal, forming a situation-driven adaptive closed loop. In one specific embodiment, the process of receiving the four-tuple structured trigger signal output by S2 and extracting the target weight vector, constraint level, and dominant target identifier from it is as follows: from the trigger signal... The safety weight and energy efficiency weight are extracted and mapped to the weight coefficients of the objective function in the collaborative optimization model. The mapping rule is: the weight coefficient of the safety objective is taken as the value of the safety weight, and the weight coefficient of the energy efficiency objective is taken as the value of the energy efficiency weight.
[0088] when When it is 4, In addition to safety weights and energy efficiency weights, the objective function also includes a control cost weight. The value of the control cost weight is 1 minus the safety weights and energy efficiency weights. The weighting coefficient of the control cost objective in the optimization objective function is taken as the value of the control cost weight. When the value is not 4, the weighting coefficient of the control cost objective in the optimization objective function is 0, and the sum of the weighting coefficients of the safety objective, the energy efficiency objective, and the control cost objective in the optimization objective function is always 1.
[0089] From the trigger signal The constraint level identifier is extracted. When the identifier is a hard constraint, the solver sets the safety constraint as a rigid boundary that must be satisfied, and the optimization algorithm does not allow the constraint to be exceeded when searching for a feasible solution. When the identifier is a soft constraint, the solver converts the safety constraint into a flexible boundary with a penalty term, allowing it to be touched at a penalty cost during the optimization process.
[0090] From the trigger signal Extract the dominant target type identifier and directly index one of the four preset solution strategy templates within the solver. The four templates are: a safe limit elimination template, and a corresponding template. A value of 1 activates the over-limit correction function for voltage and load rate upon startup; safety margin expansion module, corresponding to... The value is 2. Upon startup, the incremental reactive power compensation capacity and the finely adjustable energy storage charging and discharging power are calculated first; the energy efficiency optimization template corresponds to... The value is 3. At startup, the upper limit constraint on the energy efficiency target is lifted, allowing the solver to perform global optimization of indicators such as network loss and peak-valley difference within the safe boundary; the control cost template is minimized, corresponding to... The value is 4. At startup, the maximum number of iterations of the solver (the maximum number of search loops allowed by the solver in a single optimization calculation) and the convergence accuracy (the threshold for determining the solution to be completed when the relative change of the objective function value between two consecutive search results is less than this value) are both reduced to the daily maintenance level.
[0091] It should be noted that the solver is a dedicated computing engine within S3 used to perform numerical solutions for the co-optimization model.
[0092] The over-limit correction function is The emergency correction mechanism automatically activated by the solver under state 1 refers to the control behavior in which, when the voltage of any node in the target power distribution system or the load rate of any device exceeds the safety limit, the solver prioritizes the scheduling of reactive power compensation devices, energy storage systems and interruptible loads to forcibly pull the out-of-limit parameters back to within the limit with minimal action.
[0093] An optimization algorithm is a numerical computation method used within the solver to search for the set of control variables that maximizes the objective function within the feasible region that satisfies the constraints. In this scheme, the optimization algorithm automatically switches based on the solution strategy template and boundary condition type issued by S3: when the constraint level is hard constraint, the algorithm uses classic optimization algorithms that can strictly handle inequality constraints, such as the interior point method or sequential quadratic programming; when the constraint level is soft constraint, the algorithm switches to the augmented Lagrange method or the exterior point penalty function method.
[0094] Rigid boundaries refer to constraints set as absolute, insurmountable limits during the optimization process. Any candidate solution found by the algorithm that violates these constraints is deemed an illegal solution and discarded. In this scheme, rigid boundaries specifically refer to safety constraints such as the lower and upper limits of voltage safety in the target power distribution system, and the upper limit of load rate for transformers and lines. For example, if the voltage at the end of an industrial feeder drops to 0.92 pu due to a sudden decrease in photovoltaic output, while the allowable lower limit of voltage specified by the power supply standard for the area where the feeder is located is 0.93 pu, and the current situation is determined to be... If the constraint level is 1 (safety is tight and inefficient), the constraint level issued by S2 is a hard constraint. At this time, when the solver searches for control strategies, it will take "the voltage of all nodes must not be lower than 0.93pu" as a constraint. If the solution further increases the downstream energy storage discharge power of the feeder, causing the voltage to drop below 0.92pu, it will be discarded.
[0095] A flexible boundary with a penalty term refers to mathematically transforming an originally inviolable safety constraint by adding a penalty term to the objective function. This allows the constraint to be temporarily violated in extreme cases, but each violation incurs a cost in the objective function proportional to the degree of violation. In this scheme, the flexible boundary only... Enabled when the constraint level is 3 (safe but inefficient); for example, when the industrial feeder has sufficient safety margin and the voltage generally remains above 0.98 pu during off-peak hours, but the line loss rate is significantly higher due to the light load on the transformer, S2 issues a constraint level of "soft constraint". In this case, the solver embeds a voltage deviation penalty term into the objective function, allowing the voltage at a certain node to temporarily drop from 0.98 pu to 0.94 pu during the optimization process (still within the acceptable range, but touching the soft boundary), in exchange for an active increase in the transformer load rate and a significant decrease in line loss.
[0096] An industrial feeder with a high proportion of photovoltaic power was used at midday. Taking scenario 2 as an example, the solver first scans the capacitor banks that are installed but currently in hot standby and not in operation within the transformer area to which the feeder belongs, and calculates the total amount of reactive power compensation that can be added after grouping them into operation (this is a grouping and summing of the rated capacity of the installed capacitor banks, which is a conventional electrical calculation); at the same time, it retrieves the current state of charge and real-time charging and discharging power of the energy storage system in the same transformer area, and evaluates the active power discharge power amplitude that the energy storage system can finely adjust upward without significantly increasing line loss or sacrificing the current high energy efficiency operation level; after the two are calculated together, a combined compensation scheme is formed with the minimum reactive power increase and the minimum active power fine adjustment, which exactly raises the voltage at the end of the feeder from the near lower limit value to the target value within the safety buffer.
[0097] In one specific embodiment, the process of generating a set of coordinated control instructions for adjustable resources of a target power distribution system is as follows: [Objective function] The specific form is as follows: ,in , and These represent the weighting coefficients of the safety objective, the energy efficiency objective, and the control cost objective in the optimization objective function, respectively. , and These are respectively represented as the safety cost function, energy efficiency cost function, and control cost function.
[0098] Get the current control variable Voltage amplitude at each node under action and the rated voltage of each node. Then, combined with the importance weights of each node Then construct the security cost function. : , For each node, use its corresponding index number. , The total number corresponding to each node. The value can be a positive integer.
[0099] Get the current control variable Total active power loss of the target power distribution system under the action The total active power flowing into the target distribution system from the feeder gate of the upstream substation. Therefore, an energy efficiency cost function is constructed. : .
[0100] Obtain the adjustment target values of each controllable resource in the current solution. The actual execution value in the previous control cycle and adjustment cost coefficient ,in This serves as the index number for each controllable resource. , The total number of controllable resources. The value of is a positive integer, which is then used to construct the control cost function. : , Controllable resources The normalized baseline value (read from the target power distribution system equipment ledger database).
[0101] The constraint settings are dynamically switched based on the level identifier. When a constraint is a hard constraint, the safety constraint is set to rigid: ,in and Represented as the first The upper and lower limits of the allowable voltage for each node are determined based on the standard for voltage deviation of the target power distribution system in the area.
[0102] The adjustment range of each controllable resource is constrained by its own physical limits: the number of capacitor switching groups is a non-negative integer and does not exceed the total number of installed groups; the energy storage charging and discharging power does not exceed its rated charging and discharging power and the state of charge is not lower than the minimum allowable value; the tap position of the on-load tap changing transformer does not exceed its maximum adjustable tap position range; and the interruptible load reduction does not exceed the maximum interruptible capacity contracted by the user.
[0103] when When it is a soft constraint, the safety constraint is converted into a penalty term and incorporated into the objective function, allowing it to be added to the objective function as a penalty factor during the optimization process.
[0104] Based on the injected solution strategy template index, the solver selects the matching solution algorithm template to enable the fast solver for mixed integer linear programming. The template for safety margin expansion and energy efficiency optimization adopts the interior point method solver, while the template for minimizing control cost increases the convergence tolerance and reduces the number of iterations.
[0105] After the solution converges, the solver will The optimal solution is decomposed into specific adjustment instructions for each controllable resource, including the set value of the number of capacitor switching groups, the set value of the energy storage charging and discharging power, the target value of the tap position of the on-load tap changer, and the interruptible load reduction amount. These instructions are aligned with a unified timestamp and packaged into a set of coordinated control instructions.
[0106] It should be noted that the key node index number These are the node numbers traversed when calculating the load-weighted voltage stability margin. The node set is limited to the key nodes in S1 where sensing terminals are deployed. These key nodes were selected because they can simultaneously characterize the coupling features of safety and energy efficiency; each node has a corresponding index number. These are the node numbers traversed by the optimization model in S3 when applying voltage safety constraints. The node set covers all nodes in the target power distribution system that need to meet the upper and lower voltage limits, with a range greater than... Wider.
[0107] control variables It is a vector set of the operating settings of each controllable resource in the target power distribution system that the solver actively adjusts during the optimization process. Specifically, it includes the number of capacitor switching groups, the charging and discharging power of the energy storage system, the tap position of the on-load tap changer, and the interruptible load reduction.
[0108] The voltage amplitude is calculated in real time by the power flow calculation module embedded in the solver, based on the current network topology of the target power distribution system, the load power of each node, and the controllable resource adjustment amount collected by S1, using the forward-backward substitution method; the rated voltage of each node is directly taken from the nominal voltage level value of that node recorded in the equipment ledger database; the importance weight of each node is calculated by the load statistics module of the data platform based on the monthly statistical average proportion of the load carried by that node to the total load of the target power distribution system, and is automatically updated once at the beginning of each month.
[0109] Total active power loss is calculated by the power flow calculation module embedded within the solver, based on the topology and load data collected by S1 and the current control variables. The total active power flowing into the feeder gate of the upstream substation is obtained in real time by calculating the sum of the square of the current and the product of the resistance of each branch. The total active power is directly taken from the active power data stream collected by S1 at the feeder gate of the target distribution system at the same time.
[0110] The specific process of converting safety constraints into penalty terms is as follows: a penalty term is directly added to the objective function. The penalty term includes a penalty factor with a value of a maximum positive constant on the order of 10⁶ and a voltage over-limit penalty function (defined as the sum of squares of the deviations of the voltage of each node from the allowable upper and lower limits). Taking an industrial feeder with a high proportion of photovoltaics as an example, when it is in the "safety-relaxed and inefficient" state of mode 3, after the solver relaxes the voltage constraint to a soft constraint, in a scenario where a large photovoltaic power generation causes a voltage rise one afternoon, in order to pursue the energy efficiency target of reducing the feeder network loss from 3.2% to 2.8%, the optimization algorithm increases the energy storage charging power by 15%. Although this instruction causes the voltage of the end node to briefly touch the upper limit of 1.07 pu and produce a small over-limit of about 0.0002 pu, the penalty cost paid for this is 10. 6 ×(0.0002)2 is only 4×10-2 The magnitude of the error is far below the benefit of improved energy efficiency, so the solver determines the solution is acceptable and outputs the result for execution. The solution algorithm refers to the numerical optimization calculation method embedded in the collaborative control solver, which automatically switches the selected method for online solving of weighted multi-objective optimization models based on the solution strategy template index and constraint level issued by S3. It covers mixed integer linear programming solvers, interior point solvers, and lightweight solution methods that reduce convergence accuracy and iteration count.
[0111] In one specific embodiment, the process of adjusting the acquisition strategy of the sensing terminal according to the situational pattern identifier code in the trigger signal is as follows: based on the collaborative control instruction set, execute the corresponding instruction.
[0112] After the instruction is executed, the trigger signal is activated. The identifier is synchronously transmitted to the sensing terminal for dynamic adjustment of the data collection strategy. The adjustment rules are as follows: when 1 (safety is tense and inefficient) or When the value is 2 (safety stress and efficiency), it indicates that the target power distribution system is experiencing a safety stress situation and is in the recovery observation period after control. At this time, the sensing terminal maintains a high-frequency detection mode and reduces the adaptive threshold of the abnormal probe to 80% of the normal threshold.
[0113] when When the value is 3 (safe, relaxed, and inefficient), it indicates that the target power distribution system is in the active energy efficiency tapping stage. At this time, the sensing terminal maintains the high-frequency reconnaissance mode, but the probe threshold returns to the normal value.
[0114] when When the value is 4 (safe, relaxed, and efficient), it indicates that the target power distribution system is close to the ideal operating point. At this time, the sensing terminal switches from high-frequency reconnaissance mode back to low-frequency economic mode, and the abnormal probe threshold is restored to the default value.
[0115] S4. Using the actual operating data fed back by the target power distribution system after S3, recalculate the comprehensive efficiency distance ratio after control and compare it with the difference of the comprehensive efficiency distance ratio before control to quantitatively evaluate whether the control has made the target power distribution system status reach the expected level.
[0116] In one specific embodiment, the specific process of quantitatively evaluating whether the control has enabled the target power distribution system to achieve the expected status is as follows: after S3 is completed, the sensing terminal of the target power distribution system continuously transmits back the operating data after control according to the acquisition strategy adjusted in the third sub-step of step three.
[0117] Obtain the comprehensive efficiency distance ratio of the target power distribution system in S2, considering both safety and energy efficiency dimensions. And the corresponding overall performance distance ratio after control Calculate the closed-loop effective displacement of the target power distribution system after operation optimization and energy efficiency control. : .
[0118] Combined with the preset closed-loop effective displacement threshold , When the control is deemed effective, the complete situational pattern, control strategy (i.e., target weight vector and constraint level), comprehensive effectiveness distance ratio before and after control, and closed-loop effective displacement are packaged into a structured sample, marked as a valid sample, and stored in the database.
[0119] Conversely, if the control effect is not as expected, the sample will be packaged but marked as a sample to be reviewed.
[0120] It should be noted that after the samples to be reviewed are stored in the database, the data platform automatically retrieves them in a preset offline analysis window (such as the period of lowest system load every Sunday morning). By retrospectively examining the pre-control situation characteristics, executed strategy parameters, and post-execution feedback data corresponding to the sample, it compares each one to see if the displacement failure was caused by non-model factors such as communication timeouts, inadequate equipment response, or sudden external disturbances. The attribution conclusion is then marked in the review field of the sample. If 30 consecutive sets of samples to be reviewed all point to the execution deviation of the same control channel under the same situation mode, a preventive maintenance warning for that channel is triggered, and the control strategy corresponding to that situation mode is marked as requiring manual review.
[0121] A computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement any step of a method for optimizing the operation and controlling the energy efficiency of a power distribution system based on big data.
[0122] It not only judges whether the control is effective, but also quantifies and verifies the comprehensive efficiency distance ratio displacement before and after each control. It precipitates the effective samples into structured knowledge, which is used to regularly and automatically update the situation judgment threshold and the weight configuration of safety and energy efficiency in each mode. Samples with poor control effect will also be marked and attributed. Once the deviation continuously points to the same control channel, it can also trigger an early warning and manual review, so that the assessment accuracy and control strategy will not degrade due to equipment aging or changes in operating characteristics. On the contrary, it becomes more accurate with the increase of operating time.
[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.
Claims
1. A power distribution system operation optimization and energy efficiency control method based on big data, characterized in that, include: S1. Based on the current operating scenario and energy efficiency status of the target power distribution system, collect and simultaneously label multi-source data streams with tags that characterize the evolution trend of its safety-energy efficiency coupling situation. S2. Extract safety margin features and energy efficiency level features from the tagged multi-source data stream, substitute them into the comprehensive efficiency distance ratio for quantitative evaluation, determine the current safety-energy efficiency coupled situation mode of the target power distribution system, and generate a four-tuple structured trigger signal carrying the target weight vector, constraint level and dominant target identifier. S3. Analyze the four-tuple structured trigger signal, extract the target weight vector, constraint level and dominant target identifier, and construct a weighted multi-objective optimization model; The system generates a set of coordinated control instructions for adjustable resources within the target power distribution system and executes differentiated control. Simultaneously, it adjusts the acquisition strategy of the sensing terminal based on the situation mode identifier code in the trigger signal. S4. Obtain the actual operating data of the target power distribution system after control, recalculate the comprehensive efficiency distance ratio and compare it with that before control, and quantitatively evaluate the control effect.
2. The big data based power distribution system operation optimization and energy efficiency control method of claim 1, wherein, The process of collecting and synchronously labeling the corresponding multi-source data streams includes: The sensing terminals deployed at each key node of the target power distribution system collect basic electrical quantities such as feeder voltage, current, transformer load rate and key node line loss rate in low frequency mode under normal conditions. Anomaly probes are deployed in the sensing terminals to calculate the deviation of each key parameter from its historical steady-state baseline. When the deviation of any key parameter exceeds the adaptive threshold, the acquisition mode will be switched to high-frequency reconnaissance mode. Based on the preset information blind spot model, the system will actively request related data from the data platform to obtain meteorological environment data of the target power distribution system, bus voltage data of the upper substation, power flow data of adjacent feeders, and operating status data of distributed photovoltaic and energy storage in the same area. The multi-source data streams that are aggregated to the data platform are used to calculate the data performance contribution factor for each data stream through a synchronous annotation engine, and then divided by the latency from data collection to data entry for timeliness correction. Based on the comparison between the calculation results and the preset performance contribution threshold, each data stream is labeled as one of three categories: high value - performance sensitive, routine - energy efficiency concern, or low priority - safety redundancy, and then packaged and output in a unified time-aligned format.
3. The big data based power distribution system operation optimization and energy efficiency control method, as claimed in claim 1 wherein, The process of extracting the safety margin characteristics and energy efficiency level characteristics of the target power distribution system includes: From the tagged multi-source data stream output by S1, obtain the current effective voltage value, the lower limit of voltage, the current active power transmitted, and the rated capacity of the connected distribution transformer for each key node with deployed sensing terminals. Calculate the degree to which the voltage of each node deviates from its lower limit and weight it according to the ratio of the load of that node to the transformer capacity. Take the average value for all key nodes to obtain the load-weighted voltage stability margin as a safety margin characteristic quantity. Simultaneously, the current comprehensive line loss rate of the target power distribution system is obtained, as well as the historical best line loss rate that is within the same range as the current load rate and is selected from the historical statistics database. The line loss deviation efficiency index is obtained by dividing the historical best line loss rate by the current comprehensive line loss rate, which serves as a characteristic quantity of energy efficiency level.
4. The big data based power distribution system operation optimization and energy efficiency control method of claim 3, wherein, The process of quantitatively assessing the current system's precise status coordinates in the safety-energy efficiency four-quadrant includes: The extracted load-weighted voltage stability margin is used as the safety dimension input, and the line loss deviation efficiency index is used as the energy efficiency dimension input to calculate the comprehensive efficiency distance ratio of the target power distribution system. The calculation logic of the comprehensive efficiency distance ratio is as follows: the ratio of the safety margin characteristic quantity to the energy efficiency level characteristic quantity is used as the main term, and then multiplied by a logarithmic correction term with the distance between the safety margin characteristic quantity and the preset safety margin critical threshold as the independent variable. The calculated comprehensive efficiency distance ratio is compared with the preset first threshold, second threshold and third threshold respectively. When the comprehensive efficiency distance ratio is greater than or equal to the third situation judgment threshold, the target power distribution system is determined to be in a safe, relaxed and efficient state. When the comprehensive efficiency distance ratio is between the second situation determination threshold and the third situation determination threshold, the target power distribution system is determined to be in a safe, relaxed and inefficient state. When the comprehensive efficiency distance ratio is between the first situation determination threshold and the second situation determination threshold, the target power distribution system is determined to be in a safe, tense and efficient state. When the overall efficiency distance ratio is lower than the first situation judgment threshold, the target power distribution system is judged to be in a state of safety tension and inefficiency.
5. The big data based power distribution system operation optimization and energy efficiency control method as claimed in claim 1 wherein, The process of evaluating the corresponding coupling situation mode based on the calculation results and generating trigger signals includes: Based on the comparison results of the comprehensive efficiency distance ratio with the first threshold, the second threshold and the third threshold, the current situation mode of the target power distribution system is determined. The situation mode includes four modes: safety tension and inefficiency, safety tension and high efficiency, safety relaxation and inefficiency, and safety relaxation and high efficiency. Based on the determined situation pattern, a four-tuple structured trigger signal is constructed. The trigger signal includes four fields: situation pattern identifier code, target priority weight allocation vector, dominant objective function type identifier, and security constraint level identifier. The completed trigger signal is sent to the collaborative control solver in the form of a structured message for subsequent steps.
6. The big data based power distribution system operation optimization and energy efficiency control method as claimed in claim 1 wherein, The process of receiving and parsing the four-tuple structured trigger signal includes: The collaborative control solver receives the four-tuple structured trigger signal sent by S2 and determines one of the four current situation modes from the situation mode identifier field of the trigger signal. Extract safety weights and energy efficiency weights from the target priority weight allocation vector field. Directly assign the values of safety weights to the weighting coefficients of safety targets in the optimization objective function, and directly assign the values of energy efficiency weights to the weighting coefficients of energy efficiency targets. When the situational pattern identifier code corresponds to a safe, relaxed, and efficient mode, the control cost weight is extracted from the weight allocation vector, and the control cost weight is assigned to the weighting coefficient of the control cost target. The sum of the three weighting coefficients is always one. Extract the constraint type from the safety constraint level identifier field. When the identifier is a hard constraint, set the safety constraint as a rigid boundary that must be strictly satisfied in the optimization search. When the identifier is a soft constraint, convert the safety constraint into a flexible boundary that can be touched at a penalty cost. The target type is extracted from the dominant objective function type identifier field, and the solver is directly indexed to the preset solver strategy template that matches the current situation mode. The solver strategy template includes a safety limit elimination template, a safety margin expansion template, an energy efficiency maximization template, and a control cost minimization template.
7. The big data based power distribution system operation optimization and energy efficiency control method of claim 6, wherein, The process of generating a set of coordinated control instructions for adjustable resources in a target power distribution system includes: The normalized weight vector, composed of the weighting coefficients of safety objectives, energy efficiency objectives, and control cost objectives, is combined with the safety cost function, energy efficiency cost function, and control cost function to form a weighted multi-objective optimization model. Among them, the safety cost function measures the safety level by the weighted sum of squares of the voltage amplitude and the deviation of the rated voltage at each node under the current control variable. The energy efficiency cost function measures the energy efficiency level by the ratio of the total active power loss of the target distribution system to the total active power flowing into the outlet. The control cost function measures the adjustment cost by the weighted sum of the differences between the current adjustment target value of each controllable resource and the actual execution value of the previous control cycle. The constraints are dynamically set based on the resolved constraint level; Based on the indexed solution strategy template, a matching solution algorithm is selected. Within the feasible region that satisfies the constraints, the optimal solution of the control variables that minimizes the objective function is searched. The optimal solution is decomposed into the set value of the number of capacitor switching groups, the set value of the energy storage charging and discharging power, the target value of the tap position of the on-load tap changer, and the interruptible load reduction amount. These are then packaged into a set of coordinated control instructions.
8. The method for power distribution system operation optimization and energy efficiency control based on big data according to claim 1, characterized in that, The process of adjusting the acquisition strategy of the sensing terminal based on the situational pattern identifier code in the trigger signal includes: After the coordinated control instruction set is executed, the situation mode identification code in the trigger signal is synchronously transmitted to the sensing terminal, and the acquisition strategy is dynamically adjusted according to the four modes indicated by the situation mode identification code: When the situation mode identifier code corresponds to a security tension and inefficiency mode or a security tension and high efficiency mode, it is determined that the target power distribution system is in the recovery observation period after control. The sensing terminal maintains a high-frequency reconnaissance mode and reduces the adaptive threshold of the abnormal probe. When the situational pattern identifier code corresponds to a safe, relaxed, and inefficient mode, it is determined that the target power distribution system is in the active energy efficiency tapping stage, the sensing terminal maintains a high-frequency reconnaissance mode and the probe threshold is restored to the normal value. When the situation mode identifier code corresponds to a safe, relaxed, and efficient mode, it is determined that the target power distribution system is close to the ideal operating state. The sensing terminal switches from high-frequency reconnaissance mode back to low-frequency economic mode, and the abnormal probe threshold is restored to the default value.
9. The method for power distribution system operation optimization and energy efficiency control based on big data according to claim 1, characterized in that, The process of quantifying and evaluating the effectiveness of control includes: After obtaining the actual operating data of the target power distribution system after control, recalculate the comprehensive efficiency distance ratio after control and compare it with the comprehensive efficiency distance ratio before control. The effective displacement of the closed loop is obtained by dividing the difference between the comprehensive efficiency distance ratios before and after control by the absolute value of the comprehensive efficiency distance ratio before control. By combining the preset closed-loop effective displacement threshold, the control effect is evaluated to determine whether it meets the expected requirements.
10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 9.
