Health diagnosis and energy consumption optimization evaluation method for flow battery based on multi-evidence fusion

Abstract

The application discloses a method for flow battery health diagnosis and energy consumption optimization evaluation based on multi-evidence fusion, relates to the technical field of battery system diagnosis and energy consumption optimization, and comprises the following steps: obtaining historical operation data of a flow battery system, performing data cleaning, steady state identification and working condition binning processing on the historical operation data to obtain processed data; constructing a first preset index based on the processed data, and calculating a system health score based on the first preset index; and dividing data samples into a low-flow group and a high-flow group within a preset working condition interval based on the processed data; by taking historical operation data as a data basis, distinguishing data samples with different historical performances, evaluating the health state of the system, calculating a theoretical reference flow boundary and a flow margin, determining an optimization interval and an optimization level, and realizing multi-level comprehensive judgment of the flow battery system, the flow battery system can be conveniently distinguished in terms of flow optimization feasibility, an optimization range and evidence reliability corresponding to different working conditions.

Description

Technical Field

[0001] This invention relates to the field of battery system diagnostics and energy consumption optimization technology, and in particular to a method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion. Background Technology

[0002] Flow batteries are one of the mainstream technologies in the large-scale energy storage industry, widely used in scenarios such as grid peak shaving and renewable energy consumption. A typical flow battery energy storage system includes a stack, storage tank, circulation pump, piping, sensors, BMS, PCS, and control system. During system operation, auxiliary equipment such as the circulation pump generates significant energy consumption. Adjusting the electrolyte flow rate can reduce pump power consumption and achieve overall system energy saving. Therefore, optimizing flow battery operation has become a key research direction in the industry.

[0003] Currently, the industry mainly uses PID control, model predictive control, or fixed flow lookup table strategies to regulate electrolyte flow. These solutions directly regulate the flow of field equipment, and the operating logic assumes that the system has the conditions for reduced flow operation. The flow parameters are adjusted in real time through control algorithms.

[0004] However, existing technologies for flow regulation lack a comprehensive pre-assessment of system health status, historical operating data, response mechanisms, and safety margins. For systems that have been operating at high flow rates for a long time, historical operating data may lack low-flow sample references; relying solely on historical samples may underestimate the potential optimization space; relying solely on theoretical boundaries may overestimate the executable space. Therefore, existing technologies lack a comprehensive judgment mechanism before flow optimization, and cannot determine whether the system is optimizable, the optimization range, and the corresponding level of evidence for optimization. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion. This method solves the technical problem that existing technologies lack a comprehensive judgment mechanism before flow optimization, making it impossible to determine whether the system can be optimized, the optimization range, and the corresponding evidence optimization level.

[0006] The technical means employed in this invention are as follows: In a first aspect, the present invention provides a method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion, including: Historical operating data of the flow battery system is acquired, and the historical operating data is cleaned, steady-state identified, and processed by operating condition to obtain the processed data. A first preset index is constructed based on the processed data, and a system health score is calculated based on the first preset index. Based on the processed data, the data samples are divided into a low flow group and a high flow group within a preset operating range. Based on a second preset indicator, it is determined whether the low-flow group has experienced performance degradation relative to the high-flow group; if the low-flow group has not experienced performance degradation relative to the high-flow group, the low-flow group is determined to be a group supported by historical evidence; if the low-flow group has experienced performance degradation relative to the high-flow group, the low-flow group is determined to be a group with historical risks. Calculate the theoretical reference flow boundary based on the electrochemical reactant demand, and calculate the flow margin of the actual flow relative to the theoretical reference flow boundary. By integrating the historical evidence support group, the historical risk group, the flow margin, and the theoretical reference flow boundary, a fusion classification result is obtained. Based on the fusion classification result, shadow operation suggestions and auxiliary energy consumption reduction estimation results are obtained.

[0007] Furthermore, the historical operating data includes timestamps, total subsystem power, total subsystem voltage, SOC, open-circuit voltage, positive electrolyte flow rate, negative electrolyte flow rate, positive electrolyte temperature, negative electrolyte temperature, and at least one of multiple single-pile voltages. The data cleaning includes removing at least one of the following: static condition data, abnormal SOC data, abnormal open circuit voltage, abnormal positive electrolyte flow rate, abnormal negative electrolyte flow rate, abnormal polarization loss, and abnormal inter-pile voltage. The steady-state identification includes: arranging data according to the time sequence of the system and subsystems; when adjacent data have the same charging and discharging mode, and the sampling time interval, current change, flow rate change, temperature change and voltage jump all meet the preset threshold, the operating interval corresponding to the adjacent data is marked as the steady-state operating data interval. The operating condition partitioning includes: discretizing the data samples according to preset current range, SOC range and temperature range, wherein the width of the current range is 20-30A, the width of the SOC range is 5-8%, and the width of the temperature range is 1-3℃.

[0008] Furthermore, the formula for calculating the system health score is as follows: H = aS1 + bS2 + cS3 + dS4 + eS5 + fS6 Wherein, H is the system health score, S1 is the inter-pile consistency score, S2 is the polarization loss score, S3 is the end-of-SOC risk score, S4 is the temperature status score, S5 is the positive and negative electrode electrolyte flow balance score, S6 is the operational stability score, and a, b, c, d, e, and f are the weighting coefficients of the corresponding scores.

[0009] Furthermore, the inter-pile consistency score is determined based on the quantile of the inter-pile voltage range, the polarization loss score is determined based on the polarization loss quantile, the SOC end-of-life risk score is determined based on the inter-pile voltage range and polarization loss at the end of charging or discharging, the temperature status score is determined based on the electrolyte temperature quantile, the positive and negative electrode electrolyte flow balance score is determined based on the ratio of the positive and negative electrode electrolyte flow difference to the average flow, and the operational stability score is determined based on the voltage jump between adjacent piles.

[0010] Further, the step of determining whether the low-flow group has experienced performance degradation relative to the high-flow group based on a second preset indicator; if the low-flow group has not experienced performance degradation relative to the high-flow group, the low-flow group is determined to be a group supported by historical evidence; if the low-flow group has experienced performance degradation relative to the high-flow group, the low-flow group is determined to be a group with historical risks, includes: Based on the polarization loss increment, adverse voltage change, inter-pile pressure difference increment, temperature increment, mean current difference, mean SOC difference, and mean temperature difference, it is determined whether the low flow rate group has experienced performance degradation relative to the high flow rate group. If the polarization loss increment, the adverse voltage change, the inter-pile pressure difference increment, the temperature increment, the mean current difference, the mean SOC difference, and the mean temperature difference do not exceed the corresponding preset thresholds, then the low flow rate group has not experienced performance degradation relative to the high flow rate group, and the low flow rate group is determined to be a group supported by historical evidence. If at least one of the polarization loss increment, the adverse voltage change, the inter-pile pressure difference increment, the temperature increment, the average current difference, the average SOC difference, and the average temperature difference exceeds the corresponding preset threshold, then the low flow rate group experiences performance degradation relative to the high flow rate group, and the low flow rate group is determined to be a historical risk group.

[0011] Furthermore, the formula for calculating the theoretical reference flow boundary is as follows: Q need =N cell ×|I| / (n×F×C×ΔSOC)×3600, Among them, Q need As the theoretical reference flow boundary, N cell denoted as the number of cells in a single fuel cell stack, I as the stack current, n as the number of reacting electrons, F as the Faraday constant, C as the effective vanadium ion concentration, and ΔSOC as the allowable SOC change per pass.

[0012] Furthermore, the fusion of the historical evidence support group, the historical risk group, the flow margin, and the theoretical reference flow boundary to obtain the fused classification result includes: By integrating the historical evidence support group, the historical risk group, the flow margin, and the theoretical reference flow boundary, a fusion classification result is obtained, which includes the historical evidence support group, the margin inference candidate group, the observation group, and the historical risk group.

[0013] Furthermore, the margin inference candidate group consists of data samples that meet preset conditions and have insufficient historical evidence. The preset operating conditions are as follows: under the non-default blocking SOC end-stage operating conditions, the inter-pile pressure difference, polarization loss, temperature and positive and negative electrode electrolyte flow imbalance all meet the health margin threshold, and the ratio of actual flow to theoretical reference safe flow is not lower than the preset theoretical margin threshold.

[0014] Furthermore, based on the fusion classification results, the auxiliary energy consumption reduction estimation results are obtained, including: calculating the pump power consumption proxy quantity, calculating the difference in pump power consumption proxy quantity before and after the flow reduction for the historical evidence support group and the combination of the historical evidence support group and the margin inference candidate group, and obtaining the auxiliary energy consumption reduction estimation results.

[0015] Furthermore, the shadow operation recommendations include: a reduction ratio recommendation, a traffic scaling factor recommendation, a confidence level, and monitoring metrics; The monitoring indicators include at least one of the following: minimum single-unit voltage, maximum single-unit voltage, inter-unit voltage difference, polarization loss, temperature, and alarm signal.

[0016] Compared with the prior art, the present invention has the following advantages: This invention cleanses, identifies steady-state conditions, and categorizes operating conditions of historical operating data of a flow battery system, using this historical data as the foundation. It divides the system into high and low flow rate samples based on multiple indicators, distinguishing between groups with superior and inferior performance to differentiate data samples with different historical performance. A system health score is calculated using a first preset indicator to assess the system's health status. Furthermore, by calculating theoretical reference flow rate boundaries and flow margins, the optimization range and optimization level are determined, enabling a multi-level comprehensive assessment of the flow battery system. This facilitates the differentiation of the feasibility, scope, and reliability of flow rate optimization under different operating conditions.

[0017] Based on the above reasons, this invention can be widely applied in fields such as battery system diagnosis and energy consumption optimization. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the sub-indicators and weighted calculation of the health score of the flow battery system in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating shadow operation suggestions and energy consumption estimation in an embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0023] Please see Figure 1 , Figure 1 This is a flowchart illustrating the method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion in an embodiment of the present invention.

[0024] This invention provides a method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion, comprising the following steps: Step 101: Obtain historical operating data of the flow battery system. Perform data cleaning, steady-state identification, and operating condition binning on the historical operating data to obtain processed data. By performing data cleaning, steady-state identification, and operating condition binning on the historical operating data of the flow battery system, abnormal noise data can be filtered out, different stable operating ranges of the flow battery system can be distinguished, and multiple sets of operating condition data samples can be obtained, which facilitates accurate analysis of the operating characteristics of the flow battery system under various operating conditions.

[0025] In some embodiments, historical operating data includes timestamps, total subsystem power, total subsystem voltage, SOC (State of Charge), open-circuit voltage, positive electrolyte flow rate, negative electrolyte flow rate, positive electrolyte temperature, negative electrolyte temperature, and at least one of multiple single-pile voltages.

[0026] Specifically, historical operating data of the flow battery system can be obtained from data sources such as BMS (Battery Management System), PCS (Power Conversion System), circulating pump equipment, and temperature sensors.

[0027] For each subsystem, the subsystem current can be calculated based on the subsystem's total power and total voltage. The stack current is then calculated by combining the number of parallel branches; the average electrolyte flow rate, minimum electrolyte flow rate, and electrolyte flow rate imbalance ratio are calculated based on the positive and negative electrolyte flow rates; and the average stack voltage, maximum single stack voltage, minimum single stack voltage, and voltage distribution range between stacks are calculated based on multiple single stack voltages.

[0028] In charging mode, polarization loss can be expressed as the difference between the average stack voltage and the stack voltage calculated from the open-circuit voltage; in discharging mode, polarization loss can be expressed as the difference between the stack voltage calculated from the open-circuit voltage and the average stack voltage. By obtaining these characteristic parameters, the electrical operating state, fluid transport state, and thermal operating state of the flow battery system can be observed.

[0029] Data cleaning includes removing at least one of the following: static condition data, abnormal SOC data, abnormal open-circuit voltage, abnormal positive electrolyte flow rate, abnormal negative electrolyte flow rate, abnormal polarization loss, and abnormal inter-stall voltage. Removing static condition data can filter out invalid records of the flow battery system without power interaction, and removing various abnormal parameter data can reduce their interference with the operating condition analysis of the flow battery system, so that the historical operating data for steady-state identification and operating condition classification are consistent with the actual charging and discharging operating state of the flow battery system.

[0030] Steady-state identification includes: arranging data according to the time sequence of the system and subsystems; when adjacent data have the same charging and discharging mode, and the sampling time interval, current change, flow rate change, temperature change and voltage jump all meet the preset threshold, the operating interval corresponding to the adjacent data is marked as the steady-state operating data interval.

[0031] Specifically, for the historical operating data of the flow battery system after data cleaning, the data is first sorted chronologically according to the corresponding flow battery system and each subsystem. This chronological sorting allows for an orderly review of the sequential operating states of the historical operating data of the flow battery system. Next, adjacent sets of data are traversed to compare their charging and discharging modes. Then, the sampling time interval, current change, flow rate change, temperature change, and voltage fluctuation between the two sets of data are checked to determine if the values ​​of these parameters meet preset thresholds. If adjacent data meet the requirements of consistent charging and discharging modes and all parameter values ​​meet the preset thresholds, the operating interval corresponding to this segment of consecutive adjacent data is marked as a steady-state operating data interval. If any parameter value does not meet the corresponding preset threshold, the interval is divided at the location of the adjacent data.

[0032] By performing joint verification based on multi-dimensional operating parameters, it is possible to identify continuous operating segments with relatively smooth fluctuations within the flow battery system, effectively filtering out start-stop, switching, and transient disturbance data. The marked steady-state operating data intervals can separate operating data with less fluctuation interference, making it easier to accurately distinguish the operating states of the flow battery system at different operating stages.

[0033] The operating condition partitioning includes discretizing the data samples according to preset current range, SOC range and temperature range. The current range width is 20-30A, the SOC range width is 5-8%, and the temperature range width is 1-3℃.

[0034] For example, the current range width can be 25A, the SOC range width can be 5%, and the temperature range width can be 2℃. By dividing the data into operating condition zones, zones can be formed for the same system, subsystem, charge / discharge mode, and with similar current, SOC, and temperature. Discretizing the steady-state operating data samples according to the current range, SOC range, and temperature range allows data with similar operating characteristics of the flow battery system to be grouped into the same operating condition zone, reducing the fluctuation range of current, SOC, and temperature within a single zone, and facilitating comparison of the differences in the operating performance of the flow battery system under similar operating parameter conditions.

[0035] Please see Figure 2 , Figure 2 This is a schematic diagram illustrating the sub-indicators and weighted calculation of the health score of the flow battery system in an embodiment of the present invention.

[0036] Step 102: Construct a first preset index based on the processed data, and calculate the system health score based on the first preset index. Constructing a multi-dimensional first preset index based on the processed data, and then weighting and summing the scores of each item within the first preset index, can reduce the interference of fluctuations in single-dimensional parameters on the overall health assessment. The first preset index may include inter-pile consistency, polarization loss, end-of-state charge (SOC) risk, temperature status, positive and negative electrode electrolyte flow balance, and operational stability.

[0037] In some embodiments, the formula for calculating the system health score is: H = aS1 + bS2 + cS3 + dS4 + eS5 + fS6 Wherein, H is the system health score, S1 is the inter-pile consistency score, S2 is the polarization loss score, S3 is the end-of-SOC risk score, S4 is the temperature status score, S5 is the positive and negative electrode electrolyte flow balance score, S6 is the operational stability score, and a, b, c, d, e, and f are the weighting coefficients of the corresponding scores.

[0038] Specifically, the values ​​of a, b, c, d, e, and f can be 0.25, 0.2, 0.2, 0.15, 0.1, and 0.1, respectively, with the sum of each weighting coefficient being 1. Each sub-item score can be calculated based on a low-value-first scoring function: when the index value is less than the good threshold, the sub-item score is set to 100; when the index value is greater than the warning threshold, the sub-item score is set to 0; when the index value is within the range of the good and warning thresholds, a linear mapping operation is performed on the index value to output the corresponding sub-item score. This sub-item scoring system does not only refer to the single parameter of the stack average voltage, but also includes six dimensions: inter-stack consistency, polarization loss, SOC end-of-life risk, temperature state, positive and negative electrode electrolyte flow balance, and operational stability. It is adaptable to the assessment of the operational health status of multi-stack structures and multi-fluid loop structures in flow battery systems.

[0039] In some embodiments, the inter-stall consistency score is determined based on the quantile of the inter-stall voltage range, the polarization loss score is determined based on the polarization loss quantile, the SOC end-of-life risk score is determined based on the inter-stall voltage range and polarization loss at the end of charging or discharging, the temperature status score is determined based on the electrolyte temperature quantile, the positive and negative electrode electrolyte flow balance score is determined based on the ratio of the positive and negative electrode electrolyte flow difference to the average flow, and the operational stability score is determined based on the voltage jump between adjacent stacks.

[0040] Please refer to Table 1, which is the health assessment table for flow battery systems. Table 1 shows six health dimensions applicable to the health assessment of flow battery systems: inter-stack consistency, polarization loss, end-of-charge risk, temperature state, flow balance, and operational stability. Each health dimension is matched with corresponding example indicators and their engineering meaning is clearly defined. Among them, the 95th percentile value (P95) of the multi-stack voltage range (stack_v_range) is used as the example indicator for the inter-stack consistency dimension. This indicator can reflect the degree of voltage dispersion among multiple stacks within the flow battery system and the potential risks of weaker stacks. The 95th percentile value (P95) of the polarization loss (v_loss) is used as the example indicator for the polarization loss dimension. This indicator can reflect the electrolyte reactant supply status. The following indicators are considered: energy loss due to stack internal resistance; risk at the end of charge state using inter-stack pressure difference and polarization loss as examples, which can identify operating conditions where current reduction is not advisable at the end of charge and discharge; temperature state using the 95th percentile of electrolyte temperature (P95) as an example, which reflects the thermal safety margin of the flow battery system during operation; flow balance using the ratio of the difference in positive and negative electrode flow rates to the average flow rate, which reflects the operational risks caused by imbalance in electrolyte supply; and operational stability using the 99th percentile of voltage jump (P99) as an example, which reflects the stability of the steady-state operation of the flow battery system and abnormal disturbances during operation.

[0041] Table 1 Health Assessment Form for Flow Battery Systems

[0042] Step 103: Based on the processed data, divide the data samples into low-flow-rate groups and high-flow-rate groups within a preset operating condition range. Within each steady-state operating condition range, calculate the corresponding quantiles based on the flow parameters of the flow battery system, and group the data samples within the range based on the flow quantiles; for example, data samples with flow parameter values ​​less than or equal to the 25th quantile are classified as low-flow-rate groups, and data samples with flow parameter values ​​greater than or equal to the 75th quantile are classified as high-flow-rate groups; the data sample division operation is only performed when the total number of samples in the range, the number of samples in each group, and the flow difference between the high-flow-rate group and the low-flow-rate group all reach preset minimum limits.

[0043] By classifying the processed data samples within the steady-state operating range into low-flow and high-flow groups according to flow quantiles, and setting minimum judgment thresholds for sample size and flow difference, it is possible to filter out operating ranges with insufficient sample size and small differences between high and low flow rates. This reduces the interference of data with small flow differences on the comparison of flow-related operating characteristics of the flow battery system, and improves the discriminability of the comparison results of system operating status under different flow conditions.

[0044] Step 104: Based on the second preset index, determine whether the low flow group has experienced performance degradation relative to the high flow group; if the low flow group has not experienced performance degradation relative to the high flow group, determine the low flow group as the historical evidence support group; if the low flow group has experienced performance degradation relative to the high flow group, determine the low flow group as the historical risk group.

[0045] By comparing the operational performance of data samples from the low-flow group and the high-flow group using the second preset index, the impact of low-flow conditions on the operational performance of the flow battery system can be analyzed. By dividing the system into a historical evidence support group and a historical risk group, two different operating states of the flow battery system under low-flow conditions can be distinguished, intuitively reflecting the performance fluctuation characteristics corresponding to low-flow operating conditions.

[0046] In some embodiments, a second preset indicator is used to determine whether the low-flow group has experienced performance degradation relative to the high-flow group; if the low-flow group has not experienced performance degradation relative to the high-flow group, the low-flow group is determined to be a group supported by historical evidence; if the low-flow group has experienced performance degradation relative to the high-flow group, the low-flow group is determined to be a group with historical risks, including the following steps: Step 1041: Based on the increment of polarization loss, adverse voltage change, increment of inter-pile pressure difference, temperature increase, mean difference of current, mean difference of SOC, and mean difference of temperature, determine whether the low-flow-rate group has experienced performance degradation compared to the high-flow-rate group. The second preset index includes the increment of polarization loss, adverse voltage change, increment of inter-pile pressure difference, temperature increase, mean difference of current, mean difference of SOC, and mean difference of temperature. By comparing the differences of the corresponding parameter indices for the low-flow-rate group and the high-flow-rate group, the operational differences between the two groups of samples are quantified from multiple dimensions such as polarization, voltage consistency, thermal state, and operating conditions, which facilitates the identification of the causes of performance degradation of the flow battery system under low-flow-rate conditions.

[0047] Step 1042: If the increment of polarization loss, adverse voltage change, inter-pile pressure difference, temperature increase, mean difference of current, mean difference of SOC, and mean difference of temperature all do not exceed the corresponding preset thresholds, then the low-flow group has not experienced performance degradation relative to the high-flow group, and the low-flow group is determined to be the group supported by historical evidence. By comparing the matching of operating parameters with the corresponding preset thresholds, the judgment rule that all indicators do not exceed the limits can identify samples within the low-flow group where multi-dimensional operating parameters do not show obvious abnormal fluctuations; by labeling the corresponding low-flow group samples, it can be shown that the corresponding low-flow operating conditions have not negatively affected the operating performance of the flow battery system.

[0048] Step 1043: If at least one of the following exceeds a preset threshold: polarization loss increment, adverse voltage change, inter-pile pressure difference increment, temperature increment, mean current difference, mean SOC difference, and mean temperature difference, then the low-flow group experiences performance degradation relative to the high-flow group, and the low-flow group is determined to be a historical risk group. By using the rule of triggering a judgment based on a single indicator exceeding the limit, samples with significant deviations in any type of operating parameter within the low-flow group can be identified. By marking the corresponding low-flow group samples, the corresponding data samples that cause performance degradation of the flow battery system under low-flow conditions can be identified.

[0049] Step 105: Calculate the theoretical reference flow boundary based on the electrochemical reactant demand, and calculate the flow margin of the actual flow rate relative to the theoretical reference flow boundary. By solving for the theoretical reference flow boundary through the electrochemical reactant demand and calculating the flow margin corresponding to the actual flow rate, the surplus of electrolyte supply in the flow battery system relative to the electrochemical reaction consumption demand is quantified, and the matching state of electrolyte supply and electrochemical reaction consumption under low flow conditions can be intuitively distinguished.

[0050] In some embodiments, the formula for calculating the theoretical reference flow boundary is: Q need =N cell ×|I| / (n×F×C×ΔSOC)×3600, Among them, Q need As the theoretical reference flow boundary, N cell denoted as the number of cells in a single fuel cell stack, I as the stack current, n as the number of reacting electrons, F as the Faraday constant, C as the effective vanadium ion concentration, and ΔSOC as the allowable SOC change per pass.

[0051] Flow margin is the ratio of actual flow to the theoretical reference flow boundary.

[0052] It is worth noting that when the number of data samples corresponding to historical low flow rates is insufficient, this invention will not directly determine that there is no space for flow optimization adjustment in the corresponding operating condition. Instead, it introduces a theoretical reference flow boundary for judgment. This theoretical reference flow boundary serves only as a reference constraint in the process of identifying potential flow optimization space, rather than as the minimum safe flow rate required for on-site engineering operation or the lower limit threshold of flow rate within the closed-loop control logic. Only when the inter-reactor pressure difference, polarization loss, temperature state, positive and negative electrode electrolyte flow balance state, and theoretical flow margin all meet the corresponding judgment criteria will the operating condition sample with insufficient historical evidence samples be identified as a candidate group for flow margin extrapolation.

[0053] Please see Figure 3 , Figure 3 This is a schematic diagram illustrating shadow operation suggestions and energy consumption estimation in an embodiment of the present invention.

[0054] Step 106: Integrate historical evidence support group, historical risk group, flow margin, and theoretical reference flow boundary to obtain fusion classification results. Based on the fusion classification results, obtain shadow operation suggestions and auxiliary energy consumption reduction estimation results. By integrating multi-dimensional judgment information from historical evidence support group, historical risk group, flow margin, and theoretical reference flow boundary, a fusion classification result is generated. This result combines the historical actual operating performance of the flow battery system with the theoretical supply constraints of electrochemical reactions, outputting shadow operation suggestions and auxiliary energy consumption reduction estimation results that include operational safety boundaries and energy consumption optimization potential.

[0055] In some embodiments, the historical evidence support group, the historical risk group, the flow margin, and the theoretical reference flow boundary are fused to obtain a fused classification result, including: fusing the historical evidence support group, the historical risk group, the flow margin, and the theoretical reference flow boundary to obtain a fused classification result including the historical evidence support group, the margin inference candidate group, the observation group, and the historical risk group.

[0056] By integrating four types of information—historical evidence support group, historical risk group, flow margin, and theoretical reference flow boundary—four sample intervals are divided into historical evidence support group, margin inference candidate group, observation group, and historical risk group, thus distinguishing the operating safety level and optimization potential interval corresponding to different low flow conditions of flow battery systems.

[0057] In some embodiments, the margin inference candidate group consists of data samples that meet preset conditions but lack sufficient historical evidence. The preset conditions are: under a non-default shutdown SOC (State of Charge) end-stage operating condition, the inter-pile pressure difference, polarization loss, temperature, and positive / negative electrode electrolyte flow imbalance all meet the health margin threshold, and the ratio of actual flow rate to theoretical reference safe flow rate is not lower than the preset theoretical margin threshold. By limiting the dual judgment thresholds of multi-dimensional health margin and theoretical flow margin, the margin inference candidate group is screened to identify flow battery system operating conditions with ample safety margins across multiple operating indicators when historical low flow rate samples are insufficient, reducing misjudgments caused by judging solely based on the theoretical flow rate dimension.

[0058] The theoretical reference safe flow rate calculation formula is as follows: , in, For theoretical reference safe flow rate, For the safety factor, Q need As a theoretical reference flow boundary, This is the flow correction factor at the end of the SOC period.

[0059] Please refer to Table 2, which is a schematic table of fusion classification results. The table shows the four categories of operating condition samples, their corresponding meanings, and the corresponding recommended actions. Among them, Category A is the historical evidence support group, which represents that the historical data samples of low flow under the same steady-state operating condition have not shown significant performance degradation, and corresponding shadow flow reduction operation suggestions can be generated; Category B is the margin inference candidate group, which represents that the number of historical low flow data samples that can be collected is relatively small, but the system health margin and theoretical flow margin corresponding to the operating condition are at a relatively good level, and a cautious shadow operation approach is adopted, only outputting reference suggestions without executing closed-loop flow control; Category C is the observation group, which represents that the corresponding operating condition has insufficient data samples, and the system health margin or theoretical flow margin is not up to standard, and corresponding no flow reduction related suggestions are given, and the operating status of the operating condition is continuously monitored or supplementary historical data of the corresponding operating condition is collected; Category D is the historical risk group, which represents that polarization, voltage, inter-reactor pressure difference, or temperature-related operating risks have appeared in the low flow historical samples, and corresponding flow reduction operations are prohibited and operating risk warning information is output externally.

[0060] Table 2. Schematic diagram of fusion classification results

[0061] In some embodiments, the shadow operation recommendations include: a current reduction ratio recommendation, a flow scaling factor recommendation, a confidence level, and monitoring indicators; the monitoring indicators include at least one of the following: minimum single-pile voltage, maximum single-pile voltage, inter-pile voltage difference, polarization loss, temperature, and alarm signals. By outputting current reduction ratio recommendations, flow scaling factor recommendations, and confidence levels, along with multiple types of stack operation monitoring indicators, it is convenient for operators to comprehensively evaluate the suitability and operational risks of the flow battery system's current reduction scheme by combining the confidence level with multiple stack operation indicators.

[0062] For Class A historical evidence-supported operating conditions and Class B margin inference candidate operating conditions, this invention outputs the suggested flow reduction ratio, suggested flow scaling factor, confidence level, and detection index description for the matching operating conditions. The shadow operation process only completes the calculation and solution of the recommended flow value and does not directly issue closed-loop control commands to the field equipment. The field end can maintain the original flow control strategy and continuously monitor various indicators such as the minimum single stack voltage, the maximum single stack voltage, the inter-stack voltage difference, polarization loss, temperature, and alarm signals.

[0063] In some embodiments, the auxiliary energy consumption reduction estimation result is obtained based on the fusion classification result, including: calculating the pump power consumption proxy quantity, calculating the difference in pump power consumption proxy quantity before and after the flow reduction for the historical evidence support group and the combination of the historical evidence support group and the margin inference candidate group, and obtaining the auxiliary energy consumption reduction estimation result.

[0064] The auxiliary energy consumption reduction estimation adopts a hierarchical estimation method, using the cubic flow rate as the pump power consumption proxy. For the single operating scenario of the historical evidence support group and the combined operating scenario of the historical evidence support group and the margin inference candidate group, the pump power consumption proxy before and after the flow rate reduction operation is solved respectively. The reduction ratio of pump power consumption proxy in the two scenarios is calculated. Then, based on the proportion coefficient of pump power consumption proxy in the whole system, the auxiliary energy consumption reduction potential value in the corresponding scenario is calculated to obtain the energy consumption reduction estimation result.

[0065] This invention cleanses, identifies steady-state conditions, and categorizes operating conditions of historical operating data of a flow battery system, using this historical data as the foundation. It divides the system into high and low flow rate samples based on multiple indicators, distinguishing between groups with superior and inferior performance to differentiate data samples with different historical performance. A system health score is calculated using a first preset indicator to assess the system's health status. Furthermore, by calculating theoretical reference flow rate boundaries and flow margins, the optimization range and optimization level are determined, enabling a multi-level comprehensive assessment of the flow battery system. This facilitates the differentiation of the feasibility, scope, and reliability of flow rate optimization under different operating conditions.

[0066] 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 or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion, characterized in that, include: Historical operating data of the flow battery system is acquired, and the historical operating data is cleaned, steady-state identified, and processed by operating condition to obtain the processed data. A first preset index is constructed based on the processed data, a system health score is calculated based on the first preset index, and the health status of the flow battery system is evaluated based on the system health score. Based on the processed data, the data samples are divided into a low flow group and a high flow group within a preset operating range. Based on the second preset index, it is determined whether the low flow group has experienced performance degradation relative to the high flow group; If the low-flow group does not experience performance degradation relative to the high-flow group, the low-flow group is determined to be a group supported by historical evidence. If the low-flow group experiences performance degradation relative to the high-flow group, the low-flow group is determined to be a historical risk group. Calculate the theoretical reference flow boundary based on the electrochemical reactant demand, and calculate the flow margin of the actual flow relative to the theoretical reference flow boundary. By fusing the historical evidence support group, the historical risk group, the flow margin, and the theoretical reference flow boundary, a fused classification result is obtained, which includes the historical evidence support group, the margin inference candidate group, the observation group, and the historical risk group. The margin inference candidate group is a data sample that meets the preset conditions and has insufficient historical evidence. The preset conditions are: under the non-default blocking SOC end-stage conditions, the inter-pile pressure difference, polarization loss, temperature and positive and negative electrode electrolyte flow imbalance all meet the health margin threshold, and the ratio of actual flow to theoretical reference safe flow is not lower than the preset theoretical margin threshold. Based on the fusion classification results, shadow operation suggestions and auxiliary energy consumption reduction estimates are obtained.

2. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 1, characterized in that, The historical operating data includes timestamps, total subsystem power, total subsystem voltage, SOC, open-circuit voltage, positive electrolyte flow rate, negative electrolyte flow rate, positive electrolyte temperature, negative electrolyte temperature, and at least one of multiple single-pile voltages. The data cleaning includes removing at least one of the following: static condition data, abnormal SOC data, abnormal open circuit voltage, abnormal positive electrolyte flow rate, abnormal negative electrolyte flow rate, abnormal polarization loss, and abnormal inter-pile voltage. The steady-state identification includes: arranging data according to the time sequence of the system and subsystems; when adjacent data have the same charging and discharging mode, and the sampling time interval, current change, flow rate change, temperature change and voltage jump all meet the preset threshold, the operating interval corresponding to the adjacent data is marked as the steady-state operating data interval. The operating condition partitioning includes: discretizing the data samples according to preset current range, SOC range and temperature range, wherein the width of the current range is 20-30A, the width of the SOC range is 5-8%, and the width of the temperature range is 1-3℃.

3. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 1, characterized in that, The formula for calculating the system health score is as follows: H = aS1 + bS2 + cS3 + dS4 + eS5 + fS6 Wherein, H is the system health score, S1 is the inter-pile consistency score, S2 is the polarization loss score, S3 is the end-of-SOC risk score, S4 is the temperature status score, S5 is the positive and negative electrode electrolyte flow balance score, S6 is the operational stability score, and a, b, c, d, e, and f are the weighting coefficients of the corresponding scores.

4. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 3, characterized in that, The inter-reactor consistency score is determined based on the quantile of the inter-reactor voltage range; the polarization loss score is determined based on the polarization loss quantile; the SOC end-of-life risk score is determined based on the inter-reactor voltage range and polarization loss at the end of charging or discharging; the temperature status score is determined based on the electrolyte temperature quantile; the positive and negative electrode electrolyte flow balance score is determined based on the ratio of the positive and negative electrode electrolyte flow difference to the average flow; and the operational stability score is determined based on the voltage jump between adjacent reactors.

5. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 1, characterized in that, The second preset index is used to determine whether the low-flow group has experienced performance degradation relative to the high-flow group; If the low-flow group does not experience performance degradation relative to the high-flow group, the low-flow group is determined to be a group supported by historical evidence. If the low-flow group experiences performance degradation relative to the high-flow group, the low-flow group is determined to be a historical risk group, including: Based on the polarization loss increment, adverse voltage change, inter-pile pressure difference increment, temperature increment, mean current difference, mean SOC difference, and mean temperature difference, it is determined whether the low flow rate group has experienced performance degradation relative to the high flow rate group. If the polarization loss increment, the adverse voltage change, the inter-pile pressure difference increment, the temperature increment, the mean current difference, the mean SOC difference, and the mean temperature difference do not exceed the corresponding preset thresholds, then the low flow rate group has not experienced performance degradation relative to the high flow rate group, and the low flow rate group is determined to be a group supported by historical evidence. If at least one of the polarization loss increment, the adverse voltage change, the inter-pile pressure difference increment, the temperature increment, the average current difference, the average SOC difference, and the average temperature difference exceeds the corresponding preset threshold, then the low flow rate group experiences performance degradation relative to the high flow rate group, and the low flow rate group is determined to be a historical risk group.

6. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 1, characterized in that, The formula for calculating the theoretical reference flow boundary is as follows: Q need =N cell ×|I| / (n×F×C×ΔSOC)×3600, Among them, Q need As the theoretical reference flow boundary, N cell denoted as the number of cells in a single fuel cell stack, I as the stack current, n as the number of reacting electrons, F as the Faraday constant, C as the effective vanadium ion concentration, and ΔSOC as the allowable SOC change per pass.

7. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 1, characterized in that, The auxiliary energy consumption reduction estimation result is obtained based on the fusion classification result, including: calculating the pump power consumption proxy quantity, calculating the difference in pump power consumption proxy quantity before and after the flow reduction for the historical evidence support group and the combination of the historical evidence support group and the margin inference candidate group, and obtaining the auxiliary energy consumption reduction estimation result.

8. The method for health diagnosis and energy consumption optimization assessment of flow batteries based on multi-evidence fusion according to claim 1, characterized in that, The shadow operation recommendations include: recommendations for flow reduction ratio, flow scaling factor, confidence level, and monitoring indicators; The monitoring indicators include at least one of the following: minimum single-unit voltage, maximum single-unit voltage, inter-unit voltage difference, polarization loss, temperature, and alarm signal.

Images