Battery abnormal state intelligent alarm method and system

Abstract

The invention relates to the technical field of abnormal alarm, in particular to an intelligent alarm method and system for the abnormal state of a battery, and the method comprises the following steps: obtaining a current period log, calculating a residual vector, analyzing the change trend of a residual track included angle, extracting an abnormal tendency period, and judging whether temperature rise abnormity exists or not through the combination of a load grade and temperature rise deviation. And judging whether microcirculation characteristics are met or not, correcting an alarm level, and triggering multi-source state abnormity alarm. According to the method, the evolution trend of the battery state can be dynamically tracked and the formation process of the abnormal tendency can be accurately captured by constructing the state residual feature vector and combining with multi-cycle data analysis, and the false alarm rate is reduced and the real-time monitoring capability of the battery abnormal state is improved through refined load and temperature rise response judgment; by detecting the trend change, the temperature response offset and the microcirculation phenomenon in the periodic change of the battery, the alarm response can be adjusted in real time under different load conditions, and the accuracy and the reliability of an alarm mechanism are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of abnormal alarm, and in particular to a battery abnormal state intelligent alarm method and system. BACKGROUND

[0002] The technical field of abnormal alarm is related to the monitoring and alarm mechanism of battery management systems, and is widely used in the process of battery charging, use and storage to ensure the safety and reliability of battery state. This technical field mainly covers battery state monitoring, fault diagnosis, alarm system design and intelligent processing, etc. core matters, including real-time detection of battery voltage, current, temperature and other parameters, identification of abnormal state, generation of alarm signal, intelligent prediction and response to abnormal behavior of battery, etc. With the continuous development of battery technology, especially in the fields of electric vehicles, renewable energy storage, etc. application, the abnormal alarm technology has become one of the key technologies to ensure the safety of the battery.

[0003] Among them, the traditional battery abnormal state intelligent alarm method refers to the real-time monitoring of the working state of the battery, the collection of battery voltage, current, temperature and other parameters, the identification of the abnormal state of the battery by using rule judgment, model prediction and other methods, and the issuance of alarm signal. The traditional method usually monitors the battery state by setting threshold value, if the monitoring parameter exceeds the preset threshold value, the alarm is triggered. This method has the problems of slow response speed, high false alarm rate, etc. in actual application, and cannot accurately adapt to the change of battery state in different environments. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to provide a battery abnormal state intelligent alarm method and system.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme, a battery abnormal state intelligent alarm method, comprising the following steps: S1: obtaining the battery operation log at the end of the current charge and discharge cycle, extracting the maximum terminal voltage, minimum terminal voltage, shell temperature peak value, shell temperature valley value and working current peak value, calculating the cycle deviation value item by item, constructing the state residual feature vector of the current cycle, and generating the cycle state residual set; S2: based on the cycle state residual set, calculating the angle evolution trend of any adjacent three cycle vectors in the Euclidean space trajectory, judging whether there is a continuous growth section, locating the cycle index corresponding to the end point of the section, identifying as an abnormal tendency evolution point, and obtaining the residual trajectory evolution cycle index; S3: calling the residual trajectory evolution period identifier, matching the corresponding load interval of the load level according to the average working current and the battery monomer power output value, detecting whether the current period parameter deviates from the temperature rise rate limit interval and the steady state maintenance time interval range at the same time to obtain a load level temperature rise response deviation flag; S4: according to the residual trajectory evolution period identifier, extracting the state of charge change amplitude and single period duration in the abnormal tendency evolution point period, if the microcirculation judgment standard is met and the disturbance is not out of limit, executing alarm level downward correction to obtain an alarm level decreasing label.

[0006] The application improves that the period state residual set includes a state offset vector sequence, a period index mapping structure and a parameter deviation classification label, the residual trajectory evolution period identifier includes a period number annotation group, a trajectory evolution trend index and a residual growth inflection point index, the load level temperature rise response deviation flag includes a load level identification result, a temperature rise interval deviation identifier and a thermal response instability instruction, and the alarm level decreasing label includes a level down trigger label, a disturbance tolerance judgment identifier and a microcirculation identification sequence.

[0007] The application improves that the period state residual set acquisition step is specifically: S111: acquiring the battery operation log at the end of the current charging and discharging period, performing data structure splitting on the recorded terminal voltage measurement sequence and the shell temperature sampling sequence in the period, respectively screening the maximum and minimum values in the voltage sequence and the peak and valley values in the temperature sequence, monitoring the maximum amplitude of the current sampling sequence in the period, acquiring the numerical limit characteristics of the monitoring parameters in the period, and obtaining a period boundary state index group; S112: based on each parameter value in the period boundary state index group, calling the same parameter item in multiple continuous charging and discharging periods in the previous stage, constructing a multi-period boundary state matrix of the same structure, and using the period index as a reference, respectively calculating the difference value of each state index in the current period and the corresponding index in the history period, constructing a difference vector arranged in the parameter dimension, and acquiring the period state offset information; S113: according to the arrangement order of the period state offset information in the parameter dimension, indexing and encoding each parameter offset value, and aggregating the offset vectors to form a continuous sequence structure, and simultaneously integrating the mapping relationship of each period index and the offset vector in time sequence, generating a period state residual set.

[0008] The application improves that the residual trajectory evolution period identifier acquisition step is specifically: S211: Based on the five-dimensional vector data frame corresponding to each period in the set of period state residuals, a sequence trajectory composed of consecutive period vectors is constructed in time sequence, a vector triplet formed by every three consecutive periods is retrieved, the included angle structure of the triplet in Euclidean space is called, and an included angle evolution sequence is constructed according to the vector dot product and modulus product to generate residual trajectory included angle sequence information; S212: The included angle sequence data in the residual trajectory included angle sequence information is called, an index increasing trend judgment is performed on each sequence, a subsequence section with an included angle value amplitude state is extracted according to the period index position, and a section path with a continuous growth attribute is screened to obtain an included angle trend growth index set; S213: According to the period position number in the included angle trend growth index set, the evolution sequence of the period index is combined, the tail period position is aggregated, and an abnormal state evolution point index is identified, and a mapping structure of residual evolution nodes and period numbers is established to obtain a residual trajectory evolution period identifier.

[0009] The application improves that the obtaining step of the load level temperature rise response offset flag is specifically: S311: Based on the residual trajectory evolution period identifier, the average working current and the battery monomer power output value in the corresponding period are obtained, the working current and the power output value in the current period are mapped to the corresponding load level interval based on the relationship between the current and the power according to the load level standard, and the load level classification result corresponding to the current period is obtained; S312: According to the load level classification result, the temperature rise rate limit interval and the steady state maintenance time interval corresponding to the load level are called, the temperature rise rate and the temperature stabilization time in the current period are retrieved, and the temperature rise rate and the temperature stabilization time are compared with the rate and time interval under the load level to judge whether the temperature response of the current period deviates from the upper and lower boundaries of the set interval, if the temperature rise response data deviates from the set interval and the deviation of the temperature response exceeds the threshold value, the period is marked as an abnormal period, and a load level temperature rise response offset flag is generated.

[0010] The application improves that the obtaining step of the alarm level decreasing label is specifically: S411: According to the residual trajectory evolution period identifier, the state of charge change amplitude and the single period duration in the abnormal tendency evolution point period are extracted, the state of charge change amplitude and the period duration are compared, and it is judged whether the microcirculation judgment threshold is satisfied, if the threshold condition is satisfied, the microcirculation judgment process is entered, and a microcirculation threshold judgment result is generated; S412: According to the microcirculation threshold judgment result, the microcirculation cumulative number and the terminal voltage fluctuation amplitude interval in the current monitoring window are called, the disturbance degree evaluation value of the current period is calculated, it is judged whether the disturbance degree evaluation value of the period is within the allowable disturbance range, and the disturbance range matching state is obtained.

[0011] S413: Based on the disturbance range matching state, if the microcirculation determination standard is met and the disturbance degree evaluation value does not exceed the allowable disturbance range, the alarm level is down-regulated, the alarm level of the current period is corrected, and an alarm level decreasing tag is generated.

[0012] The formula for calculating the disturbance degree evaluation value of the current period is specifically improved as follows: ; Among them, represents the maximum value of the terminal voltage of the current period, represents the average value of the terminal voltage of the current period, represents the maximum value of the current of the current period, represents the average value of the current of the current period, is the duration of the current period, is the internal resistance constant of the battery, is the thermal effect coefficient of the battery, represents the disturbance degree evaluation value.

[0013] The method further comprises the following steps: S5: According to the load level temperature rise response offset flag and the alarm level decreasing tag, it is judged whether the temperature rise response offset exists at the same time and the down-regulated alarm level meets the alarm trigger threshold. If the conditions are met, the period indicated by the residual trajectory evolution period identifier is taken as the trigger point, the corresponding period is defined as an abnormal alarm period, the battery abnormal alarm is implemented, and a multi-source state joint alarm result is obtained. The multi-source state joint alarm result comprises an abnormal period warning number, an alarm state level label, and a trigger basis index mapping.

[0014] The acquisition step of the multi-source state joint alarm result is specifically improved as follows: S511: Based on the load level temperature rise response offset flag and the alarm level decreasing tag, the temperature rise response data and the alarm level information of the current period are obtained, it is judged whether the temperature rise response offset exists at the same time and the down-regulated alarm level meets the alarm trigger threshold, and if the conditions are met, the corresponding period is taken as the trigger point of the abnormal alarm period, and an abnormal alarm period trigger identifier is generated. S512: According to the abnormal alarm period trigger identifier, the battery abnormal alarm is implemented, the battery abnormal state information is sent to the management personnel, the alarm time and the feedback processing result are recorded, and a multi-source state joint alarm result is obtained.

[0015] A battery abnormal state intelligent alarm system is used to realize the above-mentioned battery abnormal state intelligent alarm method, and the system comprises: a cycle residual construction module, which obtains a battery operation log at an end time of a current charge and discharge cycle, extracts a maximum terminal voltage, a minimum terminal voltage, a shell temperature peak value, a shell temperature valley value and a working current peak value, calculates a cycle deviation value item by item, constructs a state residual feature vector of the current cycle, and generates a cycle state residual set; a residual evolution identification module, which calculates an included angle evolution trend of any adjacent three cycle vectors in a Euclidean space trajectory based on the cycle state residual set, judges whether a continuously increasing section exists, locates a cycle index corresponding to a section end point, identifies the cycle index as an abnormal tendency evolution point, and obtains a residual trajectory evolution cycle identification; a load level division module, which calls the residual trajectory evolution cycle identification, matches a corresponding load interval according to a working current mean value and a battery monomer power output value, detects whether current cycle parameters deviate from a temperature rise rate limit interval and a steady state maintenance time interval range at the same time, and obtains a load level temperature rise response deviation flag; an alarm level correction module, which extracts a state of charge change amplitude and a single cycle duration in an abnormal tendency evolution point cycle according to the residual trajectory evolution cycle identification, executes alarm level downward correction if a microcirculation judgment standard is met and a disturbance is not out of limit, and obtains an alarm level decreasing tag; an alarm implementation processing module, which judges whether a temperature rise response deviation exists at the same time according to the load level temperature rise response deviation flag and the alarm level decreasing tag, and whether the alarm level after being adjusted downward meets an alarm triggering threshold, takes a cycle indicated by the residual trajectory evolution cycle identification as a trigger point if the conditions are met, defines a corresponding cycle as an abnormal alarm cycle, implements battery abnormal alarm, and obtains a multi-source state joint alarm result.

[0016] Compared with the prior art, the application has the following advantages and positive effects: In the application, the state residual feature vector is constructed, and multi-cycle data analysis is combined, so that the evolution trend of the battery state can be dynamically tracked, the formation process of the abnormal tendency can be accurately captured, the false alarm rate is reduced through refined load and temperature rise response judgment, the real-time monitoring capability of the abnormal state of the battery is improved, the trend change in the periodic change of the battery, the temperature response deviation and the microcirculation phenomenon are detected, the alarm response can be adjusted in real time under different load conditions, the alarm is more targeted and timely, the problem that the prior art cannot adapt to environmental changes and the alarm precision is insufficient is solved, and the accuracy and reliability of the alarm mechanism are ensured. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The method flowchart of the application; Figure 2 The flowchart of the application for obtaining a cycle state residual set; Figure 3A flow chart for obtaining the residual trajectory evolution period identifier of the application; Figure 4 A flow chart for obtaining the load level temperature rise response offset flag of the application; Figure 5 A flow chart for obtaining the alarm level decreasing tag of the application; Figure 6 A flow chart for obtaining the multi-source state joint alarm result of the application. DETAILED DESCRIPTION

[0018] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0019] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0020] Please refer to Figure 1 The present application provides a technical solution, an intelligent alarm method for battery abnormal state, comprising the following steps: S1: Obtain the battery operation log at the end of the current charge and discharge period, extract the maximum terminal voltage, minimum terminal voltage, shell temperature peak value, shell temperature valley value and working current peak value, call the same parameters of the previous stage for a plurality of continuous periods, calculate the period deviation value by item, construct the state residual feature vector of the current period, and generate the period state residual set; S2: Based on the period state residual set, according to the time sequence arrangement composed of the five-dimensional vector of each period, calculate the evolution trend of the included angle in the Euclidean space trajectory of any adjacent three period vectors, extract the included angle change amplitude in the trend, judge whether there is a continuous growth section, and locate the period index corresponding to the end point of the section, mark as an abnormal tendency evolution point, and obtain the residual trajectory evolution period identifier; S3: calling residual trajectory evolution period identification, according to the average working current in the corresponding period and the battery monomer power output value, based on the load level classification standard, dividing into the corresponding load interval of the load level, extracting the temperature rise rate limit interval and the steady state maintenance time interval under the load level, detecting whether the current period temperature response data deviates from the range of the two intervals at the same time, obtaining the load level temperature rise response deviation flag; S4: according to the residual trajectory evolution period identification, extracting the state of charge change amplitude and single period duration in the abnormal tendency evolution point period, judging whether the microcirculation judgment threshold is met at the same time, if met, entering the microcirculation judgment process, calling the microcirculation cumulative number and terminal voltage fluctuation amplitude interval in the current monitoring window, judging whether the period is in the allowable disturbance range, if the microcirculation judgment standard is met and the disturbance is not exceeded, executing the alarm level downward correction, obtaining the alarm level decrement label; S5: according to the load level temperature rise response deviation flag and the alarm level decrement label, judging whether the temperature rise response deviation exists at the same time, and the alarm level after downward adjustment meets the alarm trigger threshold, if the conditions are met, taking the period indicated by the residual trajectory evolution period identification as the trigger point, defining the corresponding period as the abnormal alarm period, implementing the battery abnormal alarm, obtaining the multi-source state joint alarm result; The period state residual set includes a state deviation vector sequence, a period index mapping structure and a parameter deviation classification label, the residual trajectory evolution period identification includes a period number annotation group, a trajectory evolution trend index and a residual growth inflection point index, the load level temperature rise response deviation flag includes a load level identification result, a temperature rise interval deviation identification and a thermal response instability instruction, the alarm level decrement label includes a level downward trigger label, a disturbance tolerance judgment identification and a microcirculation identification sequence, and the multi-source state joint alarm result includes an abnormal period warning number, an alarm state level label and a trigger basis index mapping.

[0021] Please refer to Figure 2 The acquisition step of the period state residual set is specifically: S111: obtaining the battery operation log at the end of the current charging and discharging period, performing data structure splitting on the terminal voltage measurement sequence and the shell temperature sampling sequence recorded in the period, respectively selecting the maximum value and the minimum value in the voltage sequence, the peak value and the valley value in the temperature sequence, and monitoring the maximum amplitude of the current sampling sequence in the period, obtaining the numerical limit characteristics of the monitoring parameters in the period, obtaining the period boundary state index group; The battery operation log at the end of the charge and discharge cycle numbered 20240509-C001 is obtained, which is a structured data file containing a timestamp sequence, an end voltage measurement sequence, a current sampling sequence, and a shell temperature sampling sequence. Then, the end voltage measurement sequence and the shell temperature sampling sequence recorded during the cycle are split into data structures. The complete voltage data stream [3.301, 3.305,..., 4.198, 4.199, 4.195,..., 3.302] (unit: ) and the temperature data stream [25.1, 25.2,..., 38.6, 38.4,..., 28.5] (unit: ℃) are extracted from the log file as two independent one-dimensional arrays. Then, the maximum and minimum values in the voltage sequence and the peak and valley values in the temperature sequence are selected respectively. This selection process is achieved by performing a traversal search on the voltage array. In one traversal, the current maximum value variable (initial value is the first element of the array ) and the minimum value variable (initial value is the first element of the array ) are compared and updated simultaneously, and the voltage maximum value is obtained after the traversal ends. , the voltage minimum value is . Similarly, a traversal search is performed on the temperature array, and the initial peak and valley values are set to the first element of the array . After traversal and update, the temperature peak value is , and the temperature valley value is . At the same time, the maximum amplitude of the current sampling sequence during the cycle is monitored. This monitoring process first takes the absolute value of each element in the current data stream [-49.8, -49.9,..., 99.5, 99.8, 99.6,..., 0] (unit: A, negative value indicates charging, positive value indicates discharging), generating a new amplitude sequence [49.8, 49.9,..., 99.5, 99.8, 99.6,..., 0], and then finding the maximum value in the amplitude sequence through a single traversal, obtaining the maximum current amplitude is . The numerical limit characteristics of the monitoring parameters in the cycle are obtained, and the five characteristic values are combined to obtain the cycle boundary state indicator group.

[0022] S112: Based on the parameter value of each item in the cycle boundary state indicator group, the same parameter item in multiple consecutive charge and discharge cycles in the previous stage is called to construct a multi-cycle boundary state matrix with the same structure. Using the cycle index as a reference, the difference between each state indicator in the current cycle and the corresponding indicator in the historical cycle is calculated to construct a difference vector arranged by parameter dimension, and the cycle state offset information is obtained. The same parameters recorded in the last five consecutive charge and discharge cycles (No. 20240508-C001 to 20240508-C005) in the previous stage are called to construct a multi-cycle boundary state matrix with the same structure. The specific construction process is as follows: retrieve the boundary state index group of the specified cycle from the historical database, and arrange them in the order of cycle index to form a matrix, where the rows represent the cycles, and the columns represent the parameter items. Table 1 Multi-cycle boundary state matrix As shown in Table 1, the table lists the boundary state data of the historical cycles, and uses the cycle index as the basis. Here, the last cycle (20240508-C005) closest to the current cycle (20240509-C001) in the historical cycle is taken as the comparison basis. The difference value of each state index in the current cycle and the corresponding index in the historical cycle is calculated respectively. This calculation is an element-by-element subtraction. For example, the difference value of the maximum voltage is , the difference value of the minimum voltage is , the difference value of the temperature peak value is , the difference value of the temperature valley value is , and the difference value of the maximum current amplitude is . Arrange these five difference values in a fixed order to construct a difference vector arranged by parameter dimension. The final obtained cycle state offset information is .

[0023] S113: According to the arrangement order of the cycle state offset information in the parameter dimension, index code each parameter offset value, and aggregate the offset vector to form a continuous sequence structure. Meanwhile, integrate the mapping relationship between each cycle index and the offset vector in time sequence to generate a cycle state residual set; Index code each parameter offset value. The specific coding method is to assign fixed integer indexes to the five parameter dimensions respectively, so that the offset value corresponds to index 1, corresponds to index 2, and so on. Aggregate the offset vector to form a continuous sequence structure with inherent order. This structure is represented as a key-value pair set {1:0.001, 2:-0.001, 3:0.1, 4:0.1, 5:0} in data storage. Meanwhile, integrate the mapping relationship between each cycle index and the offset vector calculated for that cycle in time sequence, for example, map the cycle index "20240509-C001" with the offset vector ​​The difference calculation in S112 is also performed on the other four historical periods stored in the database (each period is compared with its previous period), and their respective offset vectors are obtained, for example, the offset vector of period 20240508-C005 is , the offset vector of period 20240508-C004 is , and so on. The period indexes and their corresponding offset vectors are collected together to establish a multi-period residual sample set of inter-period evolution trends. The sample set is stored in the form of a database table, each row containing a period index and five offset value fields. All the offset vector data in the sample set are extracted to generate a period state residual set.

[0024] Referring to Figure 3 , the acquisition step of the residual trajectory evolution period identifier is as follows: S211: Based on the five-dimensional vector data frame corresponding to each period in the period state residual set, a sequence trajectory composed of consecutive period vectors is constructed in chronological order. The vector triplets formed by every three consecutive periods are retrieved, the angle structure of the triplets in Euclidean space is called, and an angle evolution sequence is constructed based on the vector dot product and the modulus product to generate residual trajectory angle sequence information. Based on the five-dimensional vector data frame, for example, the vector of the current period 20240509-C001 and the vectors of its previous two periods and , a sequence trajectory composed of consecutive period vectors is constructed in chronological order. The trajectory is represented in multi-dimensional space as a series of time-ordered points , where the coordinates of each point are defined by the corresponding vector. Then, the vector triplets formed by every three consecutive periods are retrieved, and the latest triplet is . The angle structure of the triplet in Euclidean space is called, which specifically calculates the angle between two adjacent vectors. Here, the angle between vectors and is calculated , and the angle between vectors and is calculated . An angle evolution sequence is constructed based on the vector dot product and the modulus product, and the calculation process is as follows: ; First, calculate , the dot product ; The modulus ; The modulus ; but ; Similarly, the calculation yields... The calculated angle values ​​are arranged in chronological order to generate residual trajectory angle sequence information.

[0025] S212: Call the angle sequence data in the residual trajectory angle sequence information, judge the index increment trend of each sequence, extract the subsequence segment with the angle value increase state according to the period index position, and filter the segment path with continuous growth attribute to obtain the angle trend growth index set. Call the angle sequence Each value is associated with a unique periodic index position. An increasing index trend is determined for each sequence by comparing the angle between two adjacent elements in the sequence, starting from the second element of the sequence. Whether it holds true, for example, for a sequence segment The first comparison is The result is true; the second comparison is... If the result is true, based on the period index position, extract the subsequence segment where the angle value increases. The criterion for determining the increase state is that the angle value of the later period is greater than the angle value of the previous period, and the increase exceeds a fixed noise threshold. This noise threshold is set according to the standard deviation of the angle fluctuation under historical stable operating conditions. For example, if the historical fluctuation standard deviation is... The threshold is then set to twice the standard deviation. ,because and Therefore, both of these consecutive growth states were identified as valid increases, and segments with continuous growth attributes were selected. This involved finding sequence segments that satisfied the increase condition for two or more consecutive periods. In this example, the segments were selected from the corresponding period index. Starting from the position, to the corresponding The position ends, forming a continuous increasing segment of length 3. The set of all periodic indices within this segment is obtained, resulting in the angular trend increasing index set.

[0026] S213: Based on the period position number in the angle trend growth index set, combined with the evolution order of the period index, aggregate the tail period position and mark it as the abnormal state evolution point index. At the same time, establish a mapping structure between residual evolution nodes and period numbers to obtain the residual trajectory evolution period identifier. The index set of angle trend growth contains the factors that cause the angle to increase from Growth to The continuous cycle numbers, such as {20240508-C004, 20240508-C005, 20240509-C001}, are combined with the evolution order of the cycle index. The tail cycle position is aggregated and marked as the abnormal state evolution point index. Specifically, the last cycle number in time in the index set, i.e., 20240509-C001, is taken and marked as the end point of this continuous growth event, i.e., the abnormal state evolution point. At the same time, a mapping structure between residual evolution nodes and cycle numbers is established. This structure is a key-value pair, where the key is "abnormal state evolution point" and the value is the cycle number "20240509-C001". This mapping relationship is written to a dedicated state evolution log table, which contains timestamps, evolution point types and corresponding cycle numbers for subsequent module queries. This mapping relationship is extracted to obtain the residual trajectory evolution cycle identifier.

[0027] Please see Figure 4 The specific steps for obtaining the load level temperature rise response offset flag are as follows: S311: Based on the residual trajectory evolution cycle identifier, obtain the average operating current and battery cell power output value in the corresponding cycle. According to the load level standard, based on the relationship between current and power, map the operating current and power output value of the current cycle to the corresponding load level range to obtain the load level classification result corresponding to the current cycle. Based on cycle number 20240509-C001, the average operating current and the power output value of each battery cell were extracted and calculated from the original operation log of this cycle. The average operating current was calculated by taking the arithmetic mean of all current sampling points [-49.8, -49.9, ..., 99.5, 99.8, 99.6, ..., 0] within the cycle. for The method for calculating the power output value of a single battery cell is to multiply the voltage and current values ​​at each time point during the discharge phase, and then average all the power values ​​to obtain the average output power. for According to the preset load level standard, which is determined based on battery design specifications and application scenarios, the specific division is as follows: Level 1 Light Load Range (Operating Current) And power output Secondary intermediate load range (operating current) And power output Level 3 heavy load range (operating current) or power output Based on the relationship between current and power, the operating current and power output values ​​of the current cycle are mapped to the corresponding load level range. This is because the average operating current of the current cycle... In within the interval, and the average output power is in the interval, both conditions meet the definition of the first light load interval, and the load level classification result corresponding to the current period is "first light load".

[0028] S312: According to the load level classification result, the temperature rise rate limit interval and the steady state maintenance time interval corresponding to the load level are called, the temperature rise rate and the temperature stabilization time in the current period are retrieved, and the rate and time interval under the load level are compared to determine whether the current period temperature response deviates from the upper and lower boundaries of the set interval. If the temperature rise response data deviates from the set interval and the deviation of the temperature response exceeds the threshold value, the period is marked as an abnormal period, and a load level temperature rise response deviation flag is generated; According to the load level classification result "first light load", the temperature rise rate limit interval and the steady state maintenance time interval corresponding to the load level are called. The setting of these two intervals is based on the experimental data of a large number of first light load conditions of the battery of this type, and the 5th percentile and the 95th percentile of the distribution of the temperature response characteristics are taken as the lower boundary and the upper boundary of the interval. Specifically, the temperature rise rate limit interval corresponding to the first light load is , and the steady state maintenance time interval is minutes. Then, the temperature rise rate and the temperature stabilization time in the current period 20240509-C001 are retrieved. The temperature rise rate is calculated by dividing the temperature peak value - the temperature valley value by the time used for temperature rise (for example, hours), which is , and the temperature stabilization time is obtained by analyzing the duration of the temperature curve fluctuation less than near the peak value, for example minutes. The calculated values are compared with the rate and time interval under the load level to determine whether the current period temperature response deviates from the upper and lower boundaries of the set interval. The current temperature rise rate is within the interval , and there is no deviation, while the temperature stabilization time minutes is less than the lower boundary of the interval minutes, which deviates. If the temperature rise response data deviates from the set interval and the deviation of the temperature response exceeds the threshold value, the period is marked as an abnormal period. The deviation threshold value here is set according to the relative value of the interval boundary, for example, the time threshold value is 10% of the lower limit of the interval, i.e. minutes, and the current deviation is minutes, which is minutes less than the threshold value minutes, so the period is not marked as abnormal, and the load level temperature rise response deviation flag is finally generated as "no".

[0029] Please see Figure 5 The specific steps for obtaining the alarm level decreasing tag are as follows: S411: Based on the residual trajectory evolution cycle identifier, extract the charge state change amplitude and single cycle duration in the abnormal tendency evolution point cycle, compare the charge state change amplitude and cycle duration to determine whether the micro-circulation judgment threshold is met. If the threshold condition is met, proceed to the micro-circulation judgment process and generate the micro-circulation threshold judgment result. Based on the selected anomalous tendency evolution point period 20240509-C001, the state of charge (SOC) change amplitude and single-cycle duration are extracted from the detailed log of this period. This is achieved by reading the SOC value at the beginning of the period (e.g., ) and the SOC value at the end (e.g. The change in SOC was calculated to be... Simultaneously, by subtracting the start time stamp from the end time stamp of the cycle, the duration of a single cycle is obtained as 28 minutes. The change in state of charge (SOC) is compared with the cycle duration to determine whether it meets the micro-cycle judgment threshold. This threshold is set to identify short and shallow charge-discharge behaviors that have cumulative damage to battery life. Its setting is based on the following: the SOC change threshold is set according to the battery material characteristics; for lithium iron phosphate batteries, below... Shallow charge and discharge cycles can trigger a specific aging mechanism, hence the setting as The cycle duration threshold is set based on the typical idle or adjustment time of the application scenario. For example, in frequency adjustment applications of energy storage power stations, a fast response cycle of less than 30 minutes is considered a micro-cycle, so it is set to 30 minutes, due to the current cycle's SOC change amplitude. Less than If the threshold is met and the cycle duration of 28 minutes is also less than the threshold of 30 minutes, then the microcirculation determination process is initiated, and the microcirculation threshold determination result is "yes".

[0030] S412: Based on the microcirculation threshold determination result, retrieve the cumulative number of microcirculations and the voltage fluctuation range within the current monitoring window, using the following formula: ; Calculate the disturbance level assessment value for the current period, determine whether the disturbance level assessment value for the period is within the allowable disturbance range, and obtain the disturbance range matching status.

[0031] in, This represents the maximum terminal voltage value in the current cycle. This represents the average value of the terminal voltage during the current cycle. This represents the maximum current value in the current cycle. the mean value of the current of the current cycle, the duration of the current cycle, the internal resistance constant of the battery, the thermal effect coefficient of the battery, represents the disturbance degree evaluation value, the allowable disturbance range is a dynamic range set according to the working characteristics of the battery, the load condition and the historical data, which is usually obtained by monitoring the voltage and current change range of the battery under normal working state, and is adjusted in combination with the technical specifications and standards provided by the battery manufacturer. This value usually reflects the acceptable electrical disturbance range of the battery without causing failure or significant performance degradation; the number of micro-cycles accumulated in the current monitoring window, for example, 15 times, and the terminal voltage fluctuation amplitude interval, which is the statistical distribution of the difference between the voltage peak value and the mean voltage in the historical normal micro-cycle period, for example, and the formula is: ; ; the disturbance degree evaluation value of the current cycle is calculated, and the formula quantifies the comprehensive disturbance of the micro-cycle to the battery by normalizing and multiplying the peak value of the voltage and the current, and multiplying a correction factor related to time, internal resistance and thermal effect. The first term quantifies the deviation of the voltage peak value from the average voltage, the second term quantifies the deviation of the current peak value from the average current, and the third term is a dimensionless stress factor that integrates the cycle duration , internal resistance and thermal effect coefficient , reflecting the imbalance between the joule heat generated by the internal resistance and the heat dissipation capacity of the battery in a given time, and the multiplication of the three aims to comprehensively evaluate the comprehensive severity of electrical disturbance and potential thermal stress in a single micro-cycle event. The benefit of the formula is that by introducing the normalized voltage and current deviation terms and , the evaluation result is not affected by the absolute voltage or current level of the battery, and the comparability under different working conditions is enhanced. At the same time, by introducing the square root term containing the cycle duration , the internal resistance constant and the thermal effect coefficient , the electrical disturbance is related to the physical and chemical properties and thermodynamic processes inside the battery, so that the evaluation value not only reflects the surface electrical signal fluctuation, but also indirectly reveals the potential accelerated aging risk caused by internal loss and heat accumulation. For the calculation of the evaluation value ​, first get the parameter values: the maximum terminal voltage is extracted from the log of cycle 20240509-C001 , the average terminal voltage is obtained by averaging the voltage data in the cycle , the maximum current is extracted , the average current is obtained by averaging the current data in the cycle , the cycle duration , the internal resistance constant of the battery is obtained by testing the battery with a 1C current pulse for 10 seconds in the laboratory at 25°C and 50% SOC, recording the voltage change and calculating it, and its value is , the thermal effect coefficient of the battery is an empirical constant determined by accelerated aging experiments, used to correlate the electro-thermal effect and the aging rate, with a unit of , for this type of battery, its value is , the value is set based on applying different characteristic micro-cycle working conditions to multiple groups of the same type of battery until their capacity decays by 10%, recording the total ampere-hour throughput and cycle characteristics, establishing the relationship between cycle parameters and life decay through multivariate regression analysis, and then fitting it out, and the numerical value is brought into the formula for calculation: ; The disturbance degree evaluation value is calculated to be 1.2595, and it is judged whether the value is within the allowable disturbance range. The allowable disturbance range is determined by statistical analysis of the values of a large number of historical normal micro-cycle periods, and the upper limit of the 99% confidence interval is taken as the setting of the dynamic threshold, for example, the allowable disturbance range here is Since the calculated value 1.2595 is within the interval, the disturbance range matching state is "matched".

[0032] S413: Based on the disturbance range matching state, if the micro-cycle judgment criteria are met and the disturbance degree evaluation value does not exceed the allowable disturbance range, the alarm level is downgraded, the alarm level of the current cycle is corrected, and an alarm level decrement tag is generated. The alarm level of the current period is corrected. The correction process is as follows: according to the initial alarm level given by the original monitoring device (such as voltage anomaly monitoring and temperature anomaly monitoring), if the current period is identified as an "abnormal state evolution point" and is assigned an initial alarm level of "level 2 - attention", the downshift operation is performed according to the preset correction rule library. The rule library defines: "if microcirculation is 'yes' and disturbance degree assessment is'match', then alarm level is reduced by one level". According to this, "level 2 - attention" is downshifted to "level 3 - prompt". If the initial level is "level 1 - serious", it is downshifted to "level 2 - attention". If the initial level is the lowest level "level 4 - normal", it remains unchanged. In this example, the alarm level is corrected from "level 2 - attention" to "level 3 - prompt", and a data tag containing the level information before and after correction is generated, i.e. an alarm level decrease tag.

[0033] Please refer to Figure 6 The acquisition step of the multi-source state joint alarm result is as follows: S511: Based on the load level temperature rise response offset flag and the alarm level decrease tag, the temperature rise response data and alarm level information of the current period are acquired, and it is judged whether the temperature rise response offset exists at the same time and the downshifted alarm level meets the alarm trigger threshold. If the conditions are met, the corresponding period is taken as the trigger point of the abnormal alarm period, and an abnormal alarm period trigger identification is generated; Based on the load level temperature rise response offset flag (its value is "no") and the alarm level decrease tag (which contains the downshifted alarm level "level 3 - prompt"), the temperature rise response data and alarm level information of the current period are acquired, and it is judged whether the temperature rise response offset exists at the same time and the downshifted alarm level meets the alarm trigger threshold. The setting of the alarm trigger threshold is based on the risk management strategy, which specifies the minimum alarm level that requires human intervention or system automatic intervention. For example, it is set to "level 2 - attention" and above, that is, only when the alarm level is "level 2 - attention" or "level 1 - serious", the trigger condition is met. The judgment process is to perform a logical AND operation: (temperature rise response offset flag = "yes") AND (downshifted alarm level ≤ alarm trigger threshold). In this example, since the temperature rise response offset flag is "no", the first condition is not met, and the downshifted alarm level "level 3 - prompt" is higher (i.e. the severity is lower) than the alarm trigger threshold "level 2 - attention", so the second condition is also not met. Therefore, the result of logical AND is false, which does not meet the trigger condition, and the current period is not taken as the trigger point of the abnormal alarm period. The generated abnormal alarm period trigger identification is "no".

[0034] S512: According to the abnormal alarm period trigger identification, the battery abnormal alarm is implemented, the battery abnormal state information is sent to the management personnel, and the alarm time and feedback processing result are recorded to obtain the multi-source state joint alarm result; According to the abnormal alarm cycle trigger identification, if the value is "yes" (here is another setting example, the setting temperature rise response offset is "yes" and the corrected alarm level is "second level - attention", which meets the trigger condition), the battery abnormal alarm is implemented. The implementation process is to generate a structured alarm data packet with the content {"timestamp": "2025-11-03T06:54:39Z", "battery_id": "LFP-100Ah-007", "cycle_id": "20240509-C001", "alarm_level": 2, "alarm_code": "TC-D-02", "message": "Joint state anomaly: temperature rise response offset and residual trajectory evolution", "data": {"temp_deviation": "5min", "disturbance_D": 1.2595}} and send this JSON format data packet to the preset remote monitoring center API endpoint through HTTPS POST request. In this way, the battery abnormal state information is sent to the monitoring platform of the management personnel, at the same time, the complete content of the alarm data packet and the sending state (success / failure) are written into the "alarm_log" table of the local database, and a field is opened for recording the feedback processing result of the management personnel through the system interface, such as "has been dispatched for inspection" or "confirmed as false alarm". Finally, a multi-source state joint alarm result containing complete alarm information, sending record and pending state is obtained.

[0035] A battery abnormal state intelligent alarm system, the battery abnormal state intelligent alarm system is used to realize the above-mentioned battery abnormal state intelligent alarm method, and the system comprises: A cycle residual construction module, which acquires the battery operation log at the end of the current charge and discharge cycle, extracts the maximum terminal voltage, the minimum terminal voltage, the shell temperature peak value, the shell temperature valley value and the working current peak value, calculates the cycle deviation value item by item, constructs the state residual feature vector of the current cycle, and generates a cycle state residual set; A residual evolution identification module, which calculates the angle evolution trend of any adjacent three cycle vectors in the Euclidean space trajectory based on the cycle state residual set, judges whether there is a continuous growth section, locates the cycle index corresponding to the end point of the section, identifies it as an abnormal tendency evolution point, and obtains the residual trajectory evolution cycle identification; A load level division module, which calls the residual trajectory evolution cycle identification, matches the corresponding load interval according to the working current mean value and the battery monomer power output value, detects whether the current cycle parameters deviate from the temperature rise rate limit interval and the steady state maintenance time interval range at the same time, and obtains the load level temperature rise response offset flag; The alarm level correction module extracts the state of charge variation amplitude and single period duration in the evolution point period of abnormal tendency according to the residual trajectory evolution period identifier, and performs alarm level downward correction if the microcirculation determination standard is met and the disturbance is not out of limit, to obtain an alarm level decreasing label; The alarm implementation processing module judges whether the temperature rise response offset exists and the alarm level after being adjusted downward meets the alarm trigger threshold according to the load level temperature rise response offset flag and the alarm level decreasing label, and if the conditions are met, takes the period indicated by the residual trajectory evolution period identifier as the trigger point, defines the corresponding period as an abnormal alarm period, and implements the battery abnormal alarm to obtain the multi-source state joint alarm result.

[0036] The above is only a preferred embodiment of the present application, and does not limit the present application in other forms. Any person skilled in the art can use the disclosed technical content to make changes or modifications to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application without departing from the technical solution content of the present application still falls within the protection scope of the present application technical solution.

Claims

1. A method for intelligent alarming of abnormal state of a battery, characterized in that, The method comprises the following steps: S1: obtaining the battery operation log at the end of the current charge and discharge cycle, extracting the maximum terminal voltage, minimum terminal voltage, shell temperature peak value, shell temperature valley value and working current peak value, calculating the cycle deviation value item by item, constructing the state residual feature vector of the current cycle, and generating the cycle state residual set; S2: based on the cycle state residual set, the angle evolution trend of any adjacent three cycle vectors in the Euclidean space trajectory is calculated, it is judged whether there is a continuously increasing section, and the section end point corresponding to the cycle index is located, which is marked as an abnormal tendency evolution point, and the residual trajectory evolution cycle index is obtained; S3: calling the residual trajectory evolution cycle index, matching the corresponding load interval of the working current mean value and the battery monomer power output value, detecting whether the current cycle parameters deviate from the temperature rise rate limit interval and the steady state maintenance time interval range at the same time, and obtaining the load grade temperature rise response deviation flag; S4: according to the residual trajectory evolution cycle index, the state of charge change amplitude and single cycle duration in the abnormal tendency evolution point cycle are extracted, if the microcirculation judgment standard is met and the disturbance is not over limit, the alarm level is corrected downward, and the alarm level decrement label is obtained.

2. The battery abnormal state intelligent alarming method according to claim 1, wherein The cycle state residual set includes state offset vector sequence, cycle index mapping structure and parameter deviation classification label, the residual trajectory evolution cycle index includes cycle number annotation group, trajectory evolution trend index and residual growth inflection point index, the load grade temperature rise response deviation flag includes load grade identification result, temperature rise interval deviation identification and thermal response instability instruction, and the alarm level decrement label includes grade down trigger label, disturbance tolerance judgment identification and microcirculation identification sequence.

3. The battery abnormal state intelligent alarming method according to claim 2, wherein The acquisition step of the cycle state residual set is specifically: S111: obtaining the battery operation log at the end of the current charge and discharge cycle, splitting the data structure of the terminal voltage measurement sequence and the shell temperature sampling sequence recorded in the cycle, respectively selecting the maximum value and the minimum value in the voltage sequence, the peak value and the valley value in the temperature sequence, and monitoring the maximum amplitude of the current sampling sequence in the cycle, obtaining the numerical limit characteristics of the monitoring parameters in the cycle, and obtaining the cycle boundary state index group; S112: based on each parameter value in the cycle boundary state index group, calling the same parameter item in multiple continuous charge and discharge cycles in the previous stage, constructing a multi-cycle boundary state matrix with the same structure, and using the cycle index as the benchmark, respectively calculating the difference value of each state index in the current cycle and the corresponding index in the history cycle, constructing a difference vector arranged in parameter dimension, and obtaining the cycle state offset information; S113: according to the arrangement order of the cycle state offset information in the parameter dimension, the index coding of each parameter offset value is carried out, the offset vector is aggregated to form a continuous sequence structure, and the mapping relationship between each cycle index and the offset vector is integrated in time sequence, and the cycle state residual set is generated.

4. The battery abnormal state intelligent alarming method according to claim 3, wherein, The acquisition step of the residual trajectory evolution cycle index is specifically: S211: Construct a sequence track composed of consecutive cycle vectors in time sequence based on the five-dimensional vector data frame corresponding to each cycle in the cycle state residual set, retrieve a vector triplet formed by every three consecutive cycles, call the included angle structure of the triplet in Euclidean space, and construct an included angle evolution sequence according to the vector dot product and modulus product to generate residual track included angle sequence information; S212: Call the included angle sequence data in the residual track included angle sequence information, judge the index increasing trend of each sequence, extract the subsequence section with an included angle value amplitude state according to the cycle index position, and screen the section path with a continuous growth attribute to obtain an included angle trend growth index set; S213: According to the cycle position number in the included angle trend growth index set, combine the evolution sequence of the cycle index, aggregate the tail cycle position and identify it as an abnormal state evolution point index, simultaneously establish a mapping structure of residual evolution nodes and cycle numbers, and obtain a residual track evolution cycle identifier.

5. The battery abnormal state intelligent alarming method according to claim 4, wherein The obtaining step of the load level temperature rise response offset flag is specifically: S311: Based on the residual track evolution cycle identifier, obtain the working current mean value and battery monomer power output value in the corresponding cycle, map the working current and power output value of the current cycle to the corresponding load level interval according to the relationship between the current and the power based on the load level standard, and obtain the load level classification result corresponding to the current cycle; S312: According to the load level classification result, call the temperature rise rate limit interval and steady state maintenance time interval corresponding to the load level, retrieve the temperature rise rate and temperature stabilization time in the current cycle, compare them with the rate and time interval under the load level, judge whether the current cycle temperature response deviates from the upper and lower boundaries of the set interval, and if the temperature rise response data deviates from the set interval and the deviation of the temperature response exceeds the threshold value, mark the cycle as an abnormal cycle and generate a load level temperature rise response offset flag.

6. The battery abnormal state intelligent alarming method according to claim 5, wherein The obtaining step of the alarm level decrement label is specifically: S411: According to the residual track evolution cycle identifier, extract the state of charge change amplitude and single cycle duration in the abnormal tendency evolution point cycle, compare the state of charge change amplitude with the cycle duration, judge whether the microcirculation determination threshold is met, and if the threshold condition is met, enter the microcirculation determination process to generate a microcirculation threshold determination result; S412: According to the microcirculation threshold determination result, call the microcirculation cumulative number and terminal voltage fluctuation amplitude interval in the current monitoring window, calculate the disturbance degree evaluation value of the current cycle, judge whether the disturbance degree evaluation value of the cycle is within the allowable disturbance range, and obtain a disturbance range matching state; S413: Based on the disturbance range matching state, if the microcirculation determination standard is met and the disturbance degree evaluation value does not exceed the allowable disturbance range, the alarm level is downgraded, the alarm level of the current cycle is corrected, and an alarm level decrement label is generated.

7. The battery abnormal state intelligent alarming method according to claim 6, wherein The formula for calculating the disturbance degree evaluation value of the current cycle is specifically: ； wherein, represents the maximum value of the terminal voltage of the current cycle, represents the average value of the terminal voltage of the current cycle, represents the maximum value of the current of the current cycle, represents the average value of the current of the current cycle, is the duration of the current cycle, is the internal resistance constant of the battery, is the thermal effect coefficient of the battery, represents the disturbance degree evaluation value.

8. The battery abnormal state intelligent alarming method according to claim 7, wherein, The method further comprises the following steps: S5: According to the load level temperature rise response offset flag and the alarm level decrement tag, it is judged whether the temperature rise response offset exists at the same time and the alarm level after being lowered meets the alarm trigger threshold. If the conditions are met, the period indicated by the residual trajectory evolution period identifier is taken as the trigger point, the corresponding period is defined as the abnormal alarm period, the battery abnormal alarm is implemented, and the multi-source state joint alarm result is obtained. The multi-source state joint alarm result includes an abnormal period warning number, an alarm state level label, and a trigger basis index mapping.

9. The battery abnormal state intelligent alarming method according to claim 8, wherein, The acquisition step of the multi-source state joint alarm result is specifically: S511: Based on the load level temperature rise response offset flag and the alarm level decrement tag, the temperature rise response data and alarm level information of the current period are obtained, it is judged whether the temperature rise response offset exists at the same time and the alarm level after being lowered meets the alarm trigger threshold. If the conditions are met, the corresponding period is taken as the trigger point of the abnormal alarm period, and an abnormal alarm period trigger identifier is generated; S512: According to the abnormal alarm period trigger identifier, the battery abnormal alarm is implemented, the battery abnormal state information is sent to the management personnel, and the alarm time and feedback processing result are recorded, and the multi-source state joint alarm result is obtained.

10. A battery abnormal state intelligent alarm system, characterized in that, The system is used to implement the battery abnormal state intelligent alarm method of any one of claims 1-9, and the system comprises: A period residual construction module obtains the battery operation log at the end of the current charge and discharge period, extracts the maximum terminal voltage, the minimum terminal voltage, the shell temperature peak value, the shell temperature valley value, and the working current peak value, calculates the period deviation value item by item, constructs the state residual feature vector of the current period, and generates a period state residual set; A residual evolution identification module calculates the evolution trend of the included angle of any adjacent three period vectors in the Euclidean space trajectory based on the period state residual set, judges whether there is a continuously increasing section, locates the period index corresponding to the end of the section, identifies it as an abnormal tendency evolution point, and obtains a residual trajectory evolution period identifier; A load level division module calls the residual trajectory evolution period identifier, matches the load interval of the corresponding level according to the working current mean value and the battery monomer power output value, detects whether the current period parameters deviate from the temperature rise rate limit interval and the steady-state maintenance time interval range at the same time, and obtains a load level temperature rise response offset flag; An alarm level correction module extracts the state of charge change amplitude and single period duration in the abnormal tendency evolution point period according to the residual trajectory evolution period identifier. If the microcirculation judgment standard is met and the disturbance is not out of limit, the alarm level is corrected downward, and an alarm level decrement tag is obtained; An alarm implementation processing module judges whether the temperature rise response offset exists at the same time according to the load level temperature rise response offset flag and the alarm level decrement tag, and whether the alarm level after being lowered meets the alarm trigger threshold. If the conditions are met, the period indicated by the residual trajectory evolution period identifier is taken as the trigger point, the corresponding period is defined as the abnormal alarm period, the battery abnormal alarm is implemented, and the multi-source state joint alarm result is obtained.

Images