Energy storage system power distribution method based on data analysis

Abstract

The invention discloses an energy storage system power distribution method based on data analysis, and relates to the field of energy storage system power distribution, and the method comprises the following steps: obtaining the real-time power request value of each load, the charge state of an energy storage battery and the historical behavior data of each load in a current scheduling period; generating and maintaining each load behavior score according to the historical behavior data; identifying a dominant factor and a correlation factor which influence the behavior score, and establishing a correlation strength relation model between behavior factors; an allocation weight is set based on a power request value and a behavior score, and a load with a high behavior score obtains a higher priority; outputting an actual power supply power value and adding smooth start control to the low-score load; after allocation is executed, actual operation data are obtained, and behavior scores are dynamically adjusted for allocation of the next period; the method can adapt to load behavior changes, and the intelligent scheduling level and the long-term operation efficiency of the system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of energy storage system power distribution, in particular to an energy storage system power distribution method based on data analysis. BACKGROUND

[0002] With the large-scale access of renewable energy and the increasing diversification of load types, the role of energy storage systems in power systems is increasingly prominent, and gradually becomes a key device for regulating power grid operation, improving system flexibility and stability. Especially in the context of widespread deployment of distributed energy, how to efficiently allocate power to energy storage resources to ensure that various loads obtain power as needed, reasonably and stably has become an important research direction for the operation optimization of energy storage systems.

[0003] In practical applications, one of the core problems faced by energy storage systems is power distribution. Due to the real-time and uncertainty of the power demand of various loads, the operating efficiency, response capability and stability of different loads are quite different. The traditional allocation scheme has three main defects: first, it only allocates according to the current power request value, ignoring the reference value of the historical behavior data of the load; second, it lacks a multi-dimensional evaluation system for the operating characteristics of the load, and cannot accurately identify the key influencing factors; third, it uses a static allocation strategy, which is difficult to adapt to the dynamic changes of load behavior. These problems lead to phenomena such as unreasonable power distribution, low energy efficiency, and intensified voltage fluctuations in the system.

[0004] Although the existing improved scheme introduces a priority mechanism, it still has obvious deficiencies: first, the priority setting usually only considers a single indicator, and fails to comprehensively evaluate key parameters such as the power request deviation rate, energy efficiency value and voltage disturbance response time of the load; second, it lacks modeling and analysis of the correlation between load behavior factors, and cannot identify the interaction between dominant factors and associated factors; finally, the dynamic adjustment mechanism is imperfect, and it is difficult to respond to changes in load behavior in a timely manner.

[0005] In view of the above problems, the existing technology needs to be improved. SUMMARY

[0006] In view of the deficiencies of the prior art, the present application provides an energy storage system power distribution method based on data analysis.

[0007] In order to achieve the above purpose, the technical scheme of the present application is as follows:

[0008] In the first aspect, the present application discloses an energy storage system power distribution method based on data analysis, comprising the following steps:

[0009] obtaining the real-time power request value of each load, the state of charge of the energy storage battery and the historical behavior data of each load in the current scheduling period;

[0010] According to the historical behavior data, a behavior score is generated and maintained for each load;

[0011] Behavior factors affecting the behavior score are identified, the behavior factors including dominant factors and associated factors, and a relationship model describing the association strength between the behavior factors is established;

[0012] Based on the power request value of each load and the behavior score thereof, a distribution weight is set, and the higher the behavior score, the higher the priority of the load;

[0013] Actual power supply values of the loads are output according to the distribution priority, and a smooth start control is added to the load whose behavior score is lower than a set threshold value;

[0014] After the power distribution is performed, actual running data of each load is obtained, the behavior score is dynamically adjusted according to the behavior factors and the association strength in the relationship model, and the power distribution in the next scheduling period is performed.

[0015] In a second aspect, the application discloses a power distribution system of an energy storage system based on data analysis, which comprises:

[0016] A data acquisition module is configured to acquire real-time power request values of loads in a current scheduling period, a state of charge of an energy storage battery and historical behavior data of the loads;

[0017] A behavior extraction module is configured to generate and maintain a behavior score for each load according to the historical behavior data;

[0018] A behavior analysis module is configured to identify behavior factors affecting the behavior score, the behavior factors including dominant factors and associated factors, and to establish a relationship model describing the association strength between the behavior factors;

[0019] A power distribution module is configured to set a distribution weight based on the power request value of each load and the behavior score thereof, and the higher the behavior score, the higher the priority of the load; actual power supply values of the loads are output according to the distribution priority, and a smooth start control is added to the load whose behavior score is lower than a set threshold value;

[0020] A distribution adjustment module is configured to, after the power distribution is performed, obtain actual running data of each load, dynamically adjust the behavior score according to the behavior factors and the association strength in the relationship model, and perform the power distribution in the next scheduling period.

[0021] Compared with the prior art, the application has the following beneficial effects:

[0022] 1. By introducing a behavior scoring mechanism, the running reliability is evaluated based on historical behavior data such as power request deviation rate, energy efficiency, voltage disturbance response time, etc. Compared with the traditional allocation strategy which only depends on real-time request power, this method can dynamically identify efficient and stable loads and prioritize power supply, thereby improving overall energy efficiency and system stability;

[0023] 2. By constructing a "correlation strength model" between behavior factors, the dominant factor and its associated factors are identified, and the behavior score is dynamically adjusted based on the coupling strength. When the dominant or associated factors are abnormal, the system can quickly reduce the corresponding load score and limit its power acquisition, effectively reducing the risk of fault propagation and enhancing the system's ability to predict and prevent complex load behavior;

[0024] 3. A three-stage smooth start control mechanism is used to limit the impact of sudden start of loads with low behavior scores on the system, and to suppress voltage fluctuations and current disturbances. This gradual power supply method ensures safe operation of the system, taking into account power supply fairness and network stability;

[0025] 4. The load behavior score is dynamically adjusted through operation data feedback, and the updated score is applied to power redistribution in the next scheduling period, ensuring that the allocation rules can continuously evolve, adapt to changes in load behavior, and improve the intelligent scheduling level and long-term operation efficiency of the system. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0027] Figure 1 It is a whole block diagram of the method of the embodiment of the present application.

[0028] Figure 2 It is a method flowchart of the embodiment of the present application.

[0029] Figure 3 It is a whole block diagram of the system of the second embodiment of the present application. DETAILED DESCRIPTION

[0030] The technical solutions of the present application will be described in detail below in conjunction with the embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0031] Summary of the application: In the prior art, the power distribution of energy storage systems often uses static strategies or simple priority rules, which are difficult to adapt to complex and variable load demands. Traditional methods ignore the influence of load historical behavior on system operation, for example, the power distribution of an energy storage system in an industrial park is unbalanced during peak load period, and high-energy-consuming low-efficiency equipment obtains too much power, while precise instruments with fast response speed are unreasonably power-limited due to lack of historical data support. Such scenarios expose the defects of traditional distribution mechanism that cannot dynamically identify the reliability differences of loads, resulting in low system operation efficiency and stability risk.

[0032] To solve the above problems, researchers found that the historical behavior data of the load contains its operation characteristic law; through the analysis of a certain power grid case, it was found that the load with frequent power request deviation often accompanies the phenomenon of voltage disturbance response delay; further research found that there is potential correlation between different behavior indicators, for example, there is statistical correlation between startup delay time and running interruption frequency. Based on this, a behavior scoring system is proposed to quantify the reliability of the load, and the core influencing factors are determined through coupling analysis, so as to establish a dynamic weight distribution mechanism.

[0033] Embodiment one:

[0034] As shown in Figures 1-2 A power distribution method for energy storage system based on data analysis, comprising the following steps: obtaining the real-time power request value of each load in the current scheduling period, the state of charge of the energy storage battery and the historical behavior data of each load; generating and maintaining a behavior score for each load according to the historical behavior data; identifying behavior factors that affect the behavior score, the behavior factors including dominant factors and associated factors; at the same time, a relationship model describing the correlation strength between behavior factors is established; based on the power request value and the behavior score of each load, set the distribution weight; the higher the behavior score, the higher the priority of the load; according to the distribution priority, output the actual power supply value of each load, and add a smooth startup control to the load whose behavior score is lower than the set threshold; after executing the power distribution, obtain the actual running data of each load, dynamically adjust the behavior score according to the behavior factors and the correlation strength in the relationship model, and use it for power distribution in the next scheduling period.

[0035] The historical behavior data can be a set of quantifiable indicators generated in historical operation, and can be implemented by using parameters such as power request deviation rate and voltage disturbance response time, which can reflect the power demand accuracy and grid adaptability of the load. The behavior score can be a quantitative indicator for comprehensive evaluation of the historical behavior of the load, and can be implemented by using weighted summation or machine learning algorithm, and is used to represent the operation reliability of the load. The dominant factor can be a core parameter that has the greatest influence on the behavior score, and can be determined by summing the maximum value of the row and column of the correlation strength matrix, for example, the voltage disturbance response time can be identified as the dominant factor. The smooth start control can be a gradual power supply strategy implemented for low-reliability loads, and can be implemented by using a phased power ramping mechanism, and is used to reduce the impact of sudden load on the power grid.

[0036] Specifically, when the energy storage system receives a load power request, the historical operation record of the load is first called, and the current behavior score is calculated by a preset scoring model. For example, the behavior score of a certain motor load is rated as medium because the historical power request deviation rate is high. The system identifies the voltage disturbance response time of the load as the dominant factor, and there is a strong correlation between the start delay time. In the power distribution stage, the full demand of the high-score load is preferentially met, and the remaining power is distributed to the medium-score load in a linear proportion. For loads with a score below the threshold, a delayed start and power ramping strategy is used when supplying power. After distribution is completed, the system monitors the actual operation data of each load, and when an abnormality is found in the dominant factor, the corresponding load score is automatically adjusted and the correlation model is updated.

[0037] The existing scheme usually only distributes based on the current request power, for example, by using a load type preset priority. The present scheme solves the problem that a static strategy cannot adapt to changes in load behavior by constructing a dynamic behavior scoring system and incorporating historical performance into real-time decision-making. The factor correlation analysis introduced can predict the impact of potential abnormal loads on the system in advance, for example, when a load correlation factor is abnormal, even if the dominant factor is not yet abnormal, the scoring adjustment mechanism can be triggered, enhancing the system's risk prediction ability.

[0038] Through the above technical solutions, the present application realizes dynamic power distribution based on load historical behavior, effectively distinguishing the power supply priority of different loads. Through correlation analysis of behavior factors, potential abnormal loads can be identified and preventive control can be implemented in advance, reducing the risk of system operation. The introduction of smooth start control ensures the basic power supply demand of low-score loads while avoiding the impact of power mutation on the stability of the power grid, improving the overall scheduling efficiency of the energy storage system.

[0039] The historical power request deviation rate is obtained by calculating the difference ratio of the historical request power and the actual consumed power of the load, the energy consumption efficiency value is obtained by calculating the energy consumption corresponding to the unit output power of the load, and the voltage disturbance response time is obtained by the time required for the load to recover to a stable working state after voltage fluctuation.

[0040] The historical power request deviation rate refers to the difference ratio between the historical request power and the actual consumed power of the load, which can be specifically realized by accumulating and summing the difference between the request power and the actual consumed power at each time in the historical period, and then dividing by the percentage value of the total request power. This index is used to quantify the accuracy of the load power demand, and provides a data basis for evaluating the reliability of the load behavior. The energy consumption efficiency value refers to the energy consumed by the load to complete unit functional output, which can be specifically calculated by dividing the total energy consumption of the actual operation of the load by the effective output power. This index reflects the energy utilization efficiency of the load, and provides an energy efficiency evaluation basis for optimizing power distribution. The voltage disturbance response time refers to the time length for the load to recover to a stable working state after the voltage of the power grid fluctuates, which can be specifically measured by monitoring the time difference from the voltage mutation moment to the recovery of the load output current or power to the allowable fluctuation range. This index is used to evaluate the anti-interference ability of the load, and provides a basis for judging the running stability of the load.

[0041] Specifically, the calculation of the historical power request deviation rate can be based on historical data records, for example, the difference between the request power and the actual consumed power of the load per hour in the past 24 hours is counted, and the total deviation amount is accumulated and proportionally calculated with the total request power. When the deviation rate is high, it indicates that the load has inaccurate power demand prediction or abnormal execution. The calculation of the energy consumption efficiency value can be combined with the specific device type, for example, for motor type loads, the conversion efficiency of input electric energy and mechanical power output can be monitored to evaluate; for lighting type loads, the ratio of luminous flux output to power consumption can be measured. The measurement of the voltage disturbance response time can be simulated in an experimental environment by simulating voltage sag or surge conditions, and the time length required for the load to recover from voltage anomaly to normal operating parameters is recorded. The shorter the time, the stronger the stability of the load.

[0042] The traditional power distribution method usually only focuses on the current request power of the load, and lacks in-depth analysis of historical behavior data. The existing technology does not establish a quantitative evaluation system for power request deviation, energy efficiency performance and voltage response capability, which leads to the inability to accurately identify high energy consumption and low stability loads. By introducing three quantitative indexes of historical power request deviation rate, energy consumption efficiency value and voltage disturbance response time, the running characteristics of the load can be more comprehensively described, and data support is provided for dynamically adjusting the distribution strategy.

[0043] Through the technical solution, the application can effectively identify load devices with large power demand fluctuations, low energy utilization efficiency, and poor anti-interference ability, providing multi-dimensional evaluation basis for power distribution decision-making. By quantitatively analyzing historical request deviations, power distribution errors caused by inaccurate load demand prediction can be reduced; by energy efficiency evaluation, the power supply demand of high-efficiency loads can be preferentially guaranteed; by voltage response time monitoring, loads with poor stability can be found and limited in time. These technical means jointly improve the accuracy and reliability of power distribution of energy storage systems.

[0044] The application further proposes that the historical behavior data includes historical power request deviation rate, energy consumption efficiency value, voltage disturbance response time, start-up delay time, and running interruption frequency.

[0045] The historical power request deviation rate refers to the difference between the historical request power and the actual consumed power of the load, which can be calculated by calculating the absolute value of the difference between the power request value and the actual consumed value in the historical period, and the percentage of the request value, to evaluate the reliability of the load power demand. The energy consumption efficiency value refers to the energy consumption corresponding to the unit output power of the load, which can be calculated by calculating the ratio of the effective output power and the consumed energy of the load in a unit time, reflecting the energy utilization efficiency of the load. The voltage disturbance response time refers to the time required for the load to recover to a stable working state after the voltage fluctuation of the power grid, which can be measured by detecting the time interval from the voltage mutation time to the load output parameter returning to the allowable fluctuation range, to evaluate the anti-interference ability of the load. The start-up delay time refers to the interval between receiving the start-up instruction and the actual start-up of the load, which can be calculated by recording the time difference between the instruction sending time and the load power rising to the threshold level, reflecting the response speed of the load. The running interruption frequency refers to the number of unplanned shutdowns of the load in a unit time, which can be obtained by calculating the ratio of the number of abnormal stoppages of the load in a preset time window to the total running time, to evaluate the stability of the load running.

[0046] Specifically, in constructing the load behavior score, the historical power request deviation rate data is collected to identify whether the load has the behavior of falsely reporting or underestimating the power demand, and the energy consumption efficiency value can be used to judge the rationality of the load's energy utilization. The voltage disturbance response time and the startup delay time jointly reflect the dynamic response characteristics of the load, and the running interruption frequency is directly related to the long-term reliability of the load. These data are collected and analyzed in multiple dimensions to form the basis for comprehensive evaluation of the load behavior. For example, when the historical power request deviation rate of a certain load is consistently high, its behavior score will be negatively affected; if the load also has a high running interruption frequency, it will further reduce its score level. In the data collection process, the voltage response parameters can be obtained in real time through the sensors and communication modules deployed on the load side, and the power request and actual consumption data in each time period can be recorded through the historical database.

[0047] The traditional method usually only focuses on the current power demand or a single historical indicator of the load, such as only monitoring the power deviation or energy consumption efficiency. However, the present scheme integrates the power request deviation rate, energy consumption efficiency value, voltage response time, startup delay, and running interruption frequency to construct a multi-dimensional evaluation system. The existing technology does not consider the correlation between voltage disturbance response time and startup delay time, such as the risk of system instability caused by loads that start quickly but have poor voltage disturbance resistance. In addition, existing schemes lack continuous tracking of running interruption frequency, making it difficult to discover the potential threat of intermittent faults of loads to the overall stability of the system.

[0048] Through the above technical scheme, the present application can accurately identify load devices with large power demand prediction deviation, low energy utilization efficiency, weak dynamic response capability, or poor running reliability. For example, in the scenario of frequent voltage fluctuations in the power grid, by analyzing the voltage disturbance response time data, loads with insufficient anti-interference capability can be screened out, so that their priority can be reasonably reduced in power allocation. At the same time, by cross-analyzing the startup delay time and the running interruption frequency, high-priority power can be effectively allocated to loads with slow response or hidden fault risks, thereby improving the reliability and safety of the power supply decision of the energy storage system.

[0049] The present application further proposes a calculation method for the behavior score: a weight coefficient is assigned to the power request deviation rate, energy consumption efficiency value, voltage disturbance response time, startup delay time, and running interruption frequency of each load, and the initial behavior score is calculated by weighted summation.

[0050] The power request deviation rate refers to the difference between the historical request power and the actual consumption power of the load, and can be calculated by calculating the percentage of (historical request power-actual consumption power) / historical request power. The energy efficiency value refers to the energy consumption corresponding to the unit output power of the load, and can be calculated by dividing the total energy consumption by the effective output power. The voltage disturbance response time refers to the time required for the load to recover to a stable working state after voltage fluctuation, and can be calculated by recording the time difference between the voltage fluctuation time and the load output power stabilization time. The start-up delay time refers to the waiting time from receiving the start-up instruction to the actual start-up of the load, and can be calculated by measuring the difference between the instruction sending time and the actual start-up time of the load. The running interruption frequency refers to the number of unplanned shutdowns of the load in a unit time, and can be calculated by counting the number of abnormal shutdown events in the historical running period.

[0051] Specifically, in the power distribution process of the energy storage system, the power request deviation rate, energy efficiency value, voltage disturbance response time, start-up delay time and running interruption frequency are normalized and assigned different weight coefficients. For example, the weight coefficient of the power request deviation rate can be set to 0.3, the energy efficiency value to 0.25, the voltage disturbance response time to 0.2, the start-up delay time to 0.15, and the running interruption frequency to 0.1. The initial behavior score of each load is obtained by multiplying the values of the indicators by their corresponding weights and then adding them up. The score result directly reflects the comprehensive performance of the load. This calculation method can avoid the one-sidedness of single indicator evaluation, and at the same time, through the difference of the weight coefficients, the priority of the influence degree of different indicators on the system operation is reflected.

[0052] The traditional method usually only scores based on a single indicator of power request deviation or energy efficiency, while the present scheme can more comprehensively reflect the historical behavior characteristics of the load through multi-dimensional indicator weighted calculation. The influence of voltage disturbance response time and start-up delay time on system stability is not considered in the prior art, but the present scheme introduces these dynamic response indicators, effectively improving the matching degree between the scoring model and the actual running state.

[0053] Through the above technical scheme, the present application can establish a more accurate load behavior evaluation system, so that the energy storage system can consider the energy efficiency, response speed and running stability of the load in power distribution. Through comprehensive evaluation of multiple indicators, false judgments caused by abnormal single indicators are avoided, and reliable quantitative basis is provided for subsequent priority division in power distribution.

[0054] The application further proposes a construction method of a relationship model, comprising: extracting behavior factors from historical behavior data, the behavior factors including voltage disturbance response time, start-up delay time, power request deviation rate, and energy consumption efficiency value; statistically determining the probability of simultaneous abnormality of each pair of behavior factors, and calculating coupling strength values between the behavior factors; the coupling strength values are determined in the following manner: if the abnormality occurrence rate of a behavior factor B is higher than the average abnormality occurrence rate of the system when a behavior factor A is abnormal, it is determined that the behavior factor A and the behavior factor B have a positive coupling; the coupling strength value is determined according to the increase amplitude ratio of the abnormality occurrence rate of the behavior factor B when the behavior factor A is abnormal; all the behavior factors are arranged as the row and column dimensions of a matrix, and the coupling strength values are stored in the matrix elements; the dominant factor is the behavior factor with the highest total coupling strength value, which is determined by calculating and sorting the sum of the coupling strength values of each row; and the associated factor is the remaining behavior factor that has a positive coupling with the dominant factor.

[0055] The coupling strength value is a quantitative index of the abnormal state correlation degree between two behavior factors, which can be realized by calculating the ratio of the conditional probability to the reference probability, for example, when the abnormality of a behavior factor A leads to an increase in the abnormality occurrence rate of a behavior factor B, the ratio is used to measure the correlation strength. The dominant factor is the behavior factor that has the greatest impact on the load operation stability, which can be realized by summing and sorting the rows of the coupling strength matrix, for example, the factor with the highest total coupling strength is taken as the key control object. The associated factor is the other factor that has a positive coupling relationship with the dominant factor, which can be realized by screening the column factors corresponding to the non-zero elements in the row corresponding to the dominant factor in the coupling strength matrix, for example, when the dominant factor is the voltage disturbance response time, if it has a positive coupling with the power request deviation rate, the latter is listed as the associated factor.

[0056] Specifically, by constructing the coupling strength matrix between the behavior factors, the mutual influence relationship between different operating characteristics can be quantified. First, key indicators such as voltage disturbance response time and start-up delay time are extracted from historical operating data as analysis dimensions, when the voltage disturbance response time of a certain load is abnormal, the system will statistically determine the abnormality occurrence rate of other behavior factors in the same time window. If the abnormality occurrence rate of the power request deviation rate is significantly higher than the historical average at this time, a positive coupling relationship between the two is generated, and the specific coupling strength value is calculated by the abnormality occurrence rate increase amplitude ratio. Subsequently, all the behavior factors are arranged as the row and column of the matrix, the coupling strength values between the factors are filled into the corresponding matrix elements, and a complete correlation relationship model is formed. Finally, by accumulating the coupling strength values of each row, the dominant factor with the greatest impact on the overall system is identified, and the factors that have a positive correlation with the dominant factor are listed as the associated factors that need to be monitored in coordination.

[0057] The prior art generally analyzes the influence of a single behavior factor on the system independently, and lacks modeling of the correlation between multiple factors. The scheme can reveal the synergistic abnormal pattern hidden between multiple factors by constructing the coupling strength matrix, for example, when the start-up delay time is abnormal, it may trigger the synchronous abnormality of the power request deviation rate through the coupling effect. The identification of this correlation enables the system to more accurately locate the problem source, thereby comprehensively considering the influence of multiple factors when dynamically adjusting the behavior score.

[0058] Through the above technical scheme, the core driving factor of load operation abnormality and its associated influence chain can be effectively identified, avoiding the score deviation caused by isolated analysis of a single factor. For example, when the dominant factor is abnormal, the system not only reduces its own score, but also synchronously adjusts the score weight of the associated factor according to the coupling strength value, thereby more comprehensively reflecting the change trend of the load operation state and providing more reliable data support for subsequent power allocation.

[0059] The application further proposes a setting rule for distribution weight: according to the preset behavior score interval, the load is divided into high score interval load, medium score interval load and low score interval load; the load with behavior score in the high score interval obtains full power allocation; the load with behavior score in the low score interval is limited to power allocation and triggers smooth start control; the load with behavior score in the medium score interval is allocated the remaining available power in a linear proportion to the behavior score; the remaining available power is the maximum output power of the energy storage system minus the sum of the full allocation power of the high score interval load.

[0060] Among them, the behavior score interval refers to a numerical range preset for dividing the priority of the load, which can be realized by using fixed threshold or dynamically adjusted interval boundary value, and the load classification management is realized by interval division. Full power allocation refers to a power supply mode that completely meets the load request power, which can be realized by directly outputting the request power value to ensure the stable operation of high-priority loads. The remaining available power refers to the remaining power supply capacity of the system after deducting the allocation of high-score loads, which can be obtained by calculating the difference between the maximum output power of the energy storage system and the sum of the high-score load power in real time, providing a dynamic benchmark for medium-priority load allocation. Linear proportional allocation refers to calculating the allocation proportion according to the relative position of the score value in the interval, which can be realized by multiplying the remaining power by the load score mapped to the 0-1 interval, so that the allocation result is positively correlated with the load performance.

[0061] Specifically, in the power allocation process, first, the behavior score of each load is compared with the preset interval threshold, for example, loads with scores higher than 80 are classified into the high interval, 60-80 into the medium interval, and lower than 60 into the low interval. The power request of the high-interval load is fully approved, ensuring the stable operation of the key load. The remaining available power is obtained by deducting the total amount of allocated high-interval power from the total output capacity of the system. The medium-interval load is allocated the remaining power in proportion to its score within the interval, for example, a load with a score of 70 can obtain 50% of the remaining power. The low-interval load only obtains basic power supply guarantee, while starting the delayed power supply and power ramp control to avoid the impact of sudden load on the system.

[0062] The traditional power allocation method usually adopts fixed priority or average allocation strategy, which cannot be dynamically adjusted according to the historical performance of the load. The existing scheme only considers the current request parameters when classifying loads, lacks interval management of behavior scores and secondary allocation mechanism of remaining power. The present scheme realizes priority hierarchical management by establishing score interval division rules, and optimizes the utilization rate of system resources by combining dynamic calculation of remaining power and linear allocation algorithm.

[0063] Through the above technical solutions, the present application can dynamically adjust the power supply priority according to the historical behavior of the load, reasonably allocate the remaining power resources under the premise of ensuring the stable operation of high-score loads. By setting the score interval threshold, the excessive consumption of system capacity by inefficient loads is avoided, and the smooth start mechanism is used to reduce the instantaneous impact of low-score loads on the power grid. The allocation rule effectively balances the contradiction between power supply reliability and system efficiency, and improves the adaptability of the energy storage system in complex load environment.

[0064] The present application further proposes a smooth start control in three stages: after receiving the load start request, the power supply is delayed for a fixed time before starting; after the power supply starts, the output power is controlled to increase from zero to the target value in a linear or exponential manner; and after reaching the target power, the maximum output power value is continuously limited.

[0065] The delay fixed time length refers to the time interval from the load sending a start request to starting power supply, and can be implemented by setting a fixed delay period by a timer, for example, set to 500 milliseconds to 2 seconds, for preventing instantaneous impact caused by simultaneous start of multiple low-score loads. The power increases linearly or exponentially, which means that the output power increases linearly or curvilinearly over time, and the output of the inverter can be controlled by a slope function or an exponential function, for example, the power is linearly increased from 0 to 80% of the rated value within the first 3 seconds, to avoid voltage fluctuation caused by sudden change of power. The maximum output power value is continuously limited, which means that a dynamic power upper limit is applied to the load in the target power stage, and the power limiter can be used to monitor and limit the part exceeding the threshold in real time, for example, the maximum allowed power is set to 90% of the rated power, to suppress the risk of overload in abnormal working conditions.

[0066] Specifically, when the low-score load triggers a start request, it first enters a delay stage, during which the energy storage system does not respond to its power demand, effectively isolating the instantaneous disturbance of potential abnormal loads to the power grid. After the delay ends, it enters a power ramp-up stage, in which the slope of the power output curve is controlled by a linear or exponential function, so that the current change rate is constrained within a safe range, avoiding protection actions caused by sudden change of current. When the power reaches the preset target value, the power limiting module monitors the actual power consumption of the load in real time, and if the power value exceeds the dynamic upper limit, the output is automatically reduced to the allowed range to prevent power-off caused by overload.

[0067] The traditional smooth start scheme usually only uses single-stage control, such as only delaying power supply or only limiting start current, which cannot balance the instantaneous impact suppression and long-term running stability. The present scheme uses a multi-stage control mechanism to eliminate the competition effect of simultaneous start of multiple devices in the initial start stage, reduce the current change rate in the power ramp-up stage, and continuously constrain the power upper limit in the stable running stage, forming whole-cycle protection.

[0068] Through the above technical solutions, the present application can effectively reduce the impact of low-score load start on the energy storage system, reduce the damage of voltage fluctuation and current mutation to power electronic devices, and avoid abnormal loads occupying too much system capacity for a long time through dynamic power limitation, thereby improving the power supply stability of the energy storage system and the service life of the equipment.

[0069] The present application further proposes a dynamic adjustment process of the behavior score, including: if the dominant factor is abnormal, the behavior score is adjusted by a first amplitude; if the associated factor associated with the dominant factor is abnormal, the behavior score is adjusted by a second amplitude according to the coupling strength value in the association matrix, and the second amplitude increases with the increase of the association strength; if the same abnormality lasts for multiple cycles, the adjustment amplitude is increased cycle by cycle; and for the load that performs normally, the behavior score is adjusted by a fixed rate.

[0070] The dominant factor refers to the behavior factor with the highest coupling strength value in the relationship model, which can be determined by statistically analyzing the abnormal coupling strength of each behavior factor in historical data and sorting, and is used to reflect the key factor with the greatest impact on the system. The correlation factor refers to other behavior factors that have a positive coupling relationship with the dominant factor, which can be identified by calculating the abnormal occurrence rate increase ratio between behavior factors, and is used to describe the mutual influence relationship between different behavior factors. The first amplitude reduction refers to a preset proportion of score deduction according to the dominant factor abnormality, which can be implemented by setting a fixed adjustment coefficient or a dynamic coefficient based on the severity of the abnormality, and is used to quickly respond to abnormal changes in key indicators. The second amplitude reduction refers to determining the adjustment amplitude according to the coupling strength value of the correlation factor and the dominant factor, which can be implemented by mapping the coupling strength value to an adjustment coefficient and multiplying it by the basic reduction amount, and is used to quantify the influence degree of the correlation factor abnormality on the system. The gradually increasing reduction amplitude in each cycle refers to gradually increasing the score penalty for the sustained abnormal load in the subsequent cycle, which can be implemented by using a linear or exponential increasing adjustment strategy, and is used to prevent long-term abnormal load from causing cumulative impact on the system. The fixed rate increase refers to restoring the score of the normally running load according to the preset period, which can be implemented by using a constant increment or a proportion adjustment based on the difference between the current score and the upper limit value, and is used to maintain the dynamic balance of the load score.

[0071] Specifically, after power allocation, the abnormal state of the dominant factor and the correlation factor is identified by monitoring the actual running data of each load. When the dominant factor is abnormal, the load behavior score is directly reduced by the first amplitude, for example, by 10%. If the correlation factor is also abnormal, the second amplitude reduction amount is calculated according to the coupling strength value recorded in the relationship model, for example, when the coupling strength value is 0.8, a 5% reduction amount is added. For the same abnormal load, the reduction amplitude increases by 2% for each additional cycle. For the load without abnormality, its score is restored to 1% of the original value every scheduling cycle. Through the above mechanism, the score adjustment not only reflects the direct impact of the key indicators, but also quantifies the synergistic effect of the correlation factors, while avoiding excessive interference of short-term fluctuations on the score, thereby more accurately guiding the power allocation decision in the next cycle.

[0072] The traditional method usually only makes static score adjustment according to a single abnormal event, without considering the correlation between behavior factors and the abnormal persistence. The present scheme can more comprehensively evaluate the impact of load abnormality on the system by introducing the coupling strength model and the dynamic adjustment mechanism, avoid score distortion caused by isolated events or short-term fluctuations, and strengthen the inhibition effect on long-term abnormal state.

[0073] By the technical solution, the application realizes fine and dynamic management of load behavior scoring. When a dominant factor is abnormal, the scoring is adjusted preferentially, ensuring that key indicators are responded quickly. The coupling strength of associated factors is quantified to superimpose influences, improving the rationality of scoring adjustment. The punishment is gradually increased for continuous abnormality, preventing inefficient occupation of system resources. The scoring of normally running loads is restored periodically, maintaining the fairness of allocation strategies. Thus, the power allocation of the energy storage system can more accurately reflect the actual running state of the load, optimize system resource utilization, and improve power supply stability.

[0074] Embodiment two:

[0075] As shown in Figure 3 A data analysis-based energy storage system power allocation system includes:

[0076] A data acquisition module is configured to acquire real-time power request values of loads, state of charge of energy storage batteries, and historical behavior data of the loads in a current scheduling period.

[0077] A behavior extraction module is configured to generate and maintain a behavior score for each load based on the historical behavior data.

[0078] A behavior analysis module is configured to identify behavior factors affecting the behavior score, including a dominant factor and an associated factor, and to establish a relationship model describing the association strength between the behavior factors.

[0079] A power allocation module is configured to set allocation weights based on the power request values and behavior scores of the loads, with higher behavior scores indicating higher allocation priorities. The actual power supply values of the loads are output based on the allocation priorities, and a smooth start control is added to loads with behavior scores below a set threshold.

[0080] An allocation adjustment module is configured to acquire actual running data of the loads after power allocation, dynamically adjust the behavior scores based on the behavior factors and the association strength in the relationship model, and use the behavior scores for power allocation in the next scheduling period.

[0081] The above content is merely an example and description of the structure of the application. Those skilled in the art can make various modifications, supplements, or substitutions using similar methods without deviating from the structure of the application or exceeding the scope defined by the claims, and such modifications, supplements, or substitutions shall fall within the protection scope of the application.

[0082] In the description of the specification, reference to "one embodiment", "an example", "a specific example" or the like means that a particular feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. The appearances of the phrases "in one embodiment", "an example", "a specific example" or the like in various places in the specification are not necessarily referring to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0083] The preferred embodiments of the application disclosed above are only to help explain the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments. Obviously, many modifications and variations can be made in light of the contents of the specification. The specification selects and specifically describes these embodiments in order to better explain the principles and practical application of the application, so that those skilled in the art can well understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A power allocation method for an energy storage system based on data analysis, characterized in that, Includes the following steps: Obtain the real-time power request value of each load, the state of charge of the energy storage battery, and the historical behavior data of each load within the current scheduling cycle; Based on the historical behavior data, a behavior score is generated and maintained for each load. Identify the behavioral factors that influence the behavioral score, including dominant factors and related factors; and establish a relationship model describing the strength of the association between the behavioral factors. Based on the current power request values ​​and behavior scores of each load, allocation weights are set; The higher the behavior score, the higher the load allocation priority; The actual power supply value of each load is output according to the allocation priority, and smooth start control is added to the loads whose behavior score is lower than the set threshold. After power allocation is performed, the actual operating data of each load is obtained. Based on its behavior factors and the correlation strength in the relationship model, the behavior score is dynamically adjusted and used for power allocation in the next scheduling cycle.

2. The power allocation method for an energy storage system based on data analysis according to claim 1, characterized in that: The historical power request deviation rate is obtained by calculating the ratio of the difference between the historical power requested by the load and the actual power consumed. The energy efficiency value is obtained by calculating the energy consumption corresponding to the unit output power of the load. The voltage disturbance response time is obtained by the time required for the load to recover to a stable operating state after a voltage fluctuation.

3. The power allocation method for an energy storage system based on data analysis according to claim 2, characterized in that: The historical behavior data includes: historical power request deviation rate, energy efficiency value, voltage disturbance response time, startup delay time, and operation interruption frequency.

4. The power allocation method for an energy storage system based on data analysis according to claim 3, characterized in that: The behavioral score is calculated as follows: For each load, a weighting coefficient is assigned to the power request deviation rate, energy efficiency value, voltage disturbance response time, startup delay time and operation interruption frequency, and the initial behavior score is calculated by weighted summation.

5. The power allocation method for an energy storage system based on data analysis according to claim 4, characterized in that: The relational model is constructed as follows: Behavioral factors are extracted from historical behavioral data, including: voltage disturbance response time, startup delay time, power request deviation rate, and energy efficiency value. Calculate the probability of both behavioral factors exhibiting abnormalities simultaneously, and then calculate the coupling strength between the behavioral factors. The coupling strength value is determined as follows: if the abnormal occurrence rate of behavioral factor B when behavioral factor A is abnormal is higher than the average abnormal occurrence rate of the system, then behavioral factor A and behavioral factor B are considered to be positively coupled; the coupling strength value is determined according to the increase ratio of the abnormal occurrence rate of behavioral factor B when behavioral factor A is abnormal. Using all behavioral factors as the matrix row and column dimensions, the matrix element values ​​store the step coupling strength values; The dominant factor is the behavioral factor with the highest total coupling strength, which is determined by calculating the sum of the coupling strength values ​​of each row and sorting them. The correlation factors are the remaining behavioral factors that are positively coupled with the dominant factor.

6. The power allocation method for an energy storage system based on data analysis according to claim 5, characterized in that: The rules for setting the allocation weights are as follows: The load is divided into high-score range load, medium-score range load and low-score range load according to the preset behavior scoring range. Loads with high behavioral scores receive full power allocation; Loads with low behavioral scores are subject to power allocation restrictions and trigger smooth start-up control. For loads with behavioral scores in the middle range, the remaining available power is allocated in a linear proportion to the behavioral score. The remaining available power = maximum output power of the energy storage system - total power allocated to the load in the high-scoring range.

7. The power allocation method for an energy storage system based on data analysis according to claim 6, characterized in that: The smooth start control is divided into three stages: After receiving a load start request, power supply will begin after a fixed delay. After power supply begins, the control output power increases from zero to the target value in a linear or exponential manner; Once the target power is reached, the maximum output power value is continuously limited.

8. The power allocation method for an energy storage system based on data analysis according to claim 7, characterized in that: The dynamic adjustment process of the behavioral score includes: If the dominant factor is abnormal, the behavioral score will be lowered by the first magnitude. If the associated factors that are associated with the dominant factor are abnormal, the behavioral score is adjusted down by a second magnitude based on the coupling strength value in the association matrix; the second magnitude adjustment value increases as the association strength increases. If the same anomaly persists for multiple cycles, the downward adjustment will be increased cycle by cycle. For loads that perform normally, the behavior score is increased at a fixed rate.

9. A power distribution system for an energy storage system based on data analysis, characterized in that: Using a data analysis-based power allocation method for an energy storage system as described in any one of claims 1 to 8, comprising: The data acquisition module is used to acquire the real-time power request values ​​of each load, the state of charge of the energy storage battery, and the historical behavior data of each load within the current scheduling cycle. The behavior extraction module is used to generate and maintain a behavior score for each load based on the historical behavior data. The behavior analysis module is used to identify behavioral factors that influence the behavior score, including dominant factors and related factors; and to establish a relationship model describing the strength of the correlation between the behavioral factors. The power allocation module is used to set allocation weights based on the current power request values ​​and behavior scores of each load; the higher the behavior score, the higher the allocation priority of the load; the actual power supply value of each load is output according to the allocation priority, and smooth start control is added to the loads whose behavior scores are lower than the set threshold. The allocation adjustment module is used to obtain the actual operating data of each load after performing power allocation, dynamically adjust the behavior score according to its behavior factors and the correlation strength in the relationship model, and use it for power allocation in the next scheduling cycle.

Images