5g communication base station lithium battery intelligent monitoring and remote diagnosis method

Abstract

The application relates to the technical field of battery state detection, and discloses a 5G communication base station lithium battery intelligent monitoring and remote diagnosis method, which comprises the following steps: continuously detecting battery state parameters, establishing a state parameter sequence of the battery; identifying a rhythm period of the battery, and reconstructing the state parameter sequence based on the rhythm period to obtain a real-time evolution track of the battery state parameters; acquiring historical state data of the battery, and constructing a reference evolution track and a fault evolution track of the battery state parameters; performing abnormal detection on the battery based on the real-time evolution track and the reference evolution track; identifying a real-time evolution direction of the battery state parameters based on the real-time evolution track and the reference evolution track, and identifying a fault type of the battery based on the real-time evolution direction and the fault evolution track; the application can improve the accuracy of abnormal detection and fault diagnosis of the lithium battery state by identifying the rhythm period of the battery state parameters and constructing the evolution track.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery state detection, and in particular to a 5G communication base station lithium battery intelligent monitoring and remote diagnosis method. BACKGROUND

[0002] The core power supply system of a 5G communication base station relies on a lithium battery pack to realize power failure backup, load peak shaving, and emergency protection. Compared with traditional networks, 5G base stations have higher power density and more frequent load rhythm. The aging process of lithium batteries does not simply present a linear rule, but rather shows complex characteristics such as structural evolution, rhythmic fluctuation, and multi-source triggered degradation.

[0003] Existing monitoring methods are mostly based on timed collection and numerical threshold judgment, focusing on the static difference between real-time parameters and health parameters, and are suitable for identifying significant abnormal events, but cannot capture implicit performance degradation patterns, which can easily lead to missed detection of battery degradation. Current fault diagnosis schemes also have difficulty in predicting the evolution path of abnormal faults, and can only determine the existence of a fault, but cannot distinguish the fault type.

[0004] Existing solutions also cannot distinguish the influence of different base station operation rhythms on battery performance evaluation results, which can easily lead to misdiagnosis or missed diagnosis of surface normal but substantial degradation. In base stations with different deployment environments, the discharge rhythm cycle of lithium batteries differs. Existing monitoring solutions use a unified time scale (such as daily or weekly) for data processing and analysis, which makes it difficult to effectively align battery state parameters with different rhythm cycles, and cannot accurately evaluate the degree of battery degradation at the same rhythm position, reducing the relevance and accuracy of monitoring and diagnosis.

[0005] A kind of isolated base station mixed power supply system battery SOH online monitoring method and system are disclosed in Chinese patent with authorization announcement No.CN112798974B, which samples and processes the discharge current of battery in each discharge process to obtain the discharge depth in the cycle process;According to the discharge depth in the cycle process, the equivalent cycle number under the corresponding discharge depth is calculated, the sum of equivalent cycle numbers under different discharge depths is accumulated, and the real-time health status of battery is calculated by combining the linear coefficient of battery health status and equivalent cycle number, to realize online detection of battery health status.

[0006] A patent application with publication number CN109581226A discloses a base station backup battery online monitoring system, which comprises a plurality of detection modules arranged at discharge ends of different base station backup batteries, a bus interface connected with each detection module, and an upper computer connected with the bus interface through a bus; the detection module comprises a voltage detection sensor, a signal conditioning module connected with the voltage detection sensor for processing a detection signal output by the voltage detection sensor, and an A / D conversion module connected with the signal conditioning module; the A / D conversion module is connected with the bus interface. The scheme detects the power of different base station backup batteries through a plurality of voltage detection sensors, transmits the detection signals to the bus interface, and transmits the detection signals to the upper computer through the bus. The monitoring personnel can know the power of each base station backup battery in real time through the upper computer, without monitoring the power of each base station backup battery one by one, thereby improving the monitoring efficiency.

[0007] The above technical solutions all have the problem that the influence of the running rhythm of the base station on the battery performance evaluation result cannot be identified.

[0008] The information disclosed in this Background section is only intended to increase an understanding of the general context in which the present application can be practiced. It is not admitted that any of the information provided in this section is prior art or that it is used to determine the scope of the application. SUMMARY

[0009] The technical problem to be solved by the present application is to overcome the defects of the prior art and provide a 5G communication base station lithium battery intelligent monitoring and remote diagnosis method. The rhythm period of the battery state parameter is identified and an evolution track thereof is constructed, so as to improve the accuracy of abnormal detection and fault diagnosis of the lithium battery state.

[0010] To solve the above technical problems, the present application provides the following technical solutions:

[0011] A 5G communication base station lithium battery intelligent monitoring and remote diagnosis method comprises the following steps:

[0012] Continuously detecting a battery state parameter to establish a state parameter sequence of the battery;

[0013] Identifying a rhythm period of the battery based on the state parameter sequence, and reconstructing the state parameter sequence based on the rhythm period to obtain a real-time evolution track of the battery state parameter;

[0014] Obtaining historical state data of the battery, and constructing a reference evolution track and a fault evolution track of the battery state parameter based on the historical state data;

[0015] Identifying a degradation trend of the battery state parameter based on the real-time evolution track and the reference evolution track, and triggering abnormal detection of the battery based on the degradation trend.

[0016] identify a real-time evolution direction of the battery state parameter based on the real-time evolution trajectory and the reference evolution trajectory, and identify a fault type of the battery based on the real-time evolution direction and the fault evolution trajectory.

[0017] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method, the battery state parameter at least includes a discharge current; and the state parameter sequence is a time sequence composed of state parameter vectors at different time points.

[0018] The state parameter sequence is established, and specifically includes: setting a current threshold; continuously collecting each battery state parameter at each time point at a preset sampling period; if the discharge current is greater than the current threshold, each battery state parameter at the corresponding time point is combined into a state parameter vector, and the state parameter vector at the corresponding time point is filled into the state parameter sequence as an element.

[0019] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method, the rhythm period of the battery is identified, and specifically includes:

[0020] extracting a current time sequence from the state parameter sequence; the current time sequence includes discharge currents at different time points;

[0021] performing frequency domain conversion on the current time sequence to obtain a frequency spectrum of the discharge current;

[0022] extracting a frequency component with the largest energy proportion in the frequency spectrum as a main frequency of the discharge current;

[0023] calculating a change period length of the discharge current based on the main frequency, as a length of the rhythm period of the battery.

[0024] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method, the real-time evolution trajectory includes state parameter vectors corresponding to different rhythm periods; and the state parameter sequence is reconstructed based on the rhythm period, and specifically includes:

[0025] setting a reference parameter sequence; the reference parameter sequence is a time sequence composed of reference parameter vectors at different time points; a reference parameter vector at any time point contains each reference state parameter at the corresponding time point; and the reference state parameter is a reference value of the battery state parameter;

[0026] cutting out a reference sequence segment from the reference parameter sequence by using a sliding window; the length of the sliding window is equal to the period length of the rhythm period;

[0027] M times of sliding interception are performed on the state parameter sequence through the sliding window, and a sliding step is 1, and each time of sliding interception obtains a sequence segment in the state parameter sequence; M is a length of the sliding window;

[0028] Similarities of the M sequence segments and the reference sequence segment are respectively calculated; a sequence segment with the largest similarity with the reference sequence segment is marked as a first rhythm period;

[0029] The first rhythm period is taken as a starting point, and continuous sliding interception is performed on the state parameter sequence through the sliding window, and a sliding step is M, and each time of sliding interception obtains a rhythm period, until the state parameter sequence is completely divided into different rhythm periods, and the real-time evolution track is obtained.

[0030] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method described in the application, the historical state data includes the reference parameter sequence;

[0031] A method for constructing the reference evolution track of the battery state parameter is as follows: the reference sequence segment is taken as a starting point, and continuous sliding interception is performed on the reference parameter sequence through the sliding window, and a sliding step is M, and each time of sliding interception obtains a rhythm period, until the reference parameter sequence is completely divided into different rhythm periods, and the reference evolution track is obtained.

[0032] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method described in the application, the historical state data further includes an abnormal state parameter; the abnormal state parameter includes a battery state parameter when a battery state is abnormal;

[0033] A method for constructing the fault evolution track is as follows:

[0034] The abnormal state parameter is sorted into an abnormal parameter sequence; the abnormal parameter sequence is a time sequence composed of abnormal parameter vectors at different times; an abnormal parameter vector at any time includes each abnormal state parameter at the corresponding time;

[0035] The abnormal parameter sequence is reconstructed based on the rhythm period, and an abnormal evolution track of the abnormal state parameter is obtained; the abnormal evolution track includes abnormal parameter vectors corresponding to different rhythm periods;

[0036] Each abnormal evolution track is clustered, and the abnormal evolution tracks are divided into different cluster clusters;

[0037] A mean sequence of each abnormal evolution track in each cluster cluster is calculated as a fault evolution track of each cluster cluster.

[0038] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method described in the application, wherein the identification of the degradation trend of the battery state parameter specifically comprises:

[0039] N consecutive rhythm cycles are intercepted from the real-time evolution track to obtain a real-time evolution sequence; N consecutive rhythm cycles are intercepted from the reference evolution track to obtain a reference evolution sequence; N is a positive integer;

[0040] M×N vector groups are extracted from the real-time evolution sequence and the reference evolution sequence; any vector group contains a state parameter vector and a reference parameter vector, and in the same vector group, the index of the state parameter vector in the real-time evolution sequence is the same as the index of the reference parameter vector in the reference evolution sequence;

[0041] The Euclidean distance between the state parameter vector and the reference parameter vector in each vector group is calculated as the deviation value of each vector group;

[0042] The deviation value of each vector group is normalized and accumulated to obtain a degradation value of the battery state parameter; if the degradation value is greater than a preset degradation threshold, the battery state parameter has a degradation trend.

[0043] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method described in the application, wherein the anomaly detection specifically comprises: if the battery state parameter has a degradation trend, triggering the anomaly detection of the battery, the method is as follows:

[0044] The deviation value of each vector group is arranged into a deviation value sequence;

[0045] The deviation value sequence is linearly fitted to obtain a fitting slope; the variance of each deviation value in the deviation value sequence is calculated to obtain a deviation fluctuation value;

[0046] An abnormal tension index is calculated based on the fitting slope and the deviation fluctuation value; the abnormal tension index is the product of the fitting slope and the deviation fluctuation value;

[0047] If the abnormal tension index is greater than a preset abnormal tension threshold, the battery has an anomaly.

[0048] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method described in the application, wherein the real-time evolution direction is represented by an evolution direction vector of the real-time evolution track; if the battery has an anomaly, triggering the identification of the real-time evolution direction of the battery state parameter, specifically comprising:

[0049] The difference vector between the state parameter vector and the reference parameter vector in each vector group is calculated as the direction vector of each vector group;

[0050] Calculate the mean vector of the direction vectors of all vector groups as the evolution direction vector of the real-time evolution trajectory.

[0051] As a preferred scheme of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method described in the application, wherein: the real-time evolution direction and the fault evolution trajectory are used to identify the fault type of the battery, specifically including:

[0052] Respectively, N consecutive rhythm cycles are intercepted from each fault evolution trajectory to obtain the fault evolution sequence corresponding to each fault evolution trajectory.

[0053] Based on the fault evolution sequence corresponding to each fault evolution trajectory and the reference evolution sequence, the evolution direction vector of each fault evolution trajectory is calculated.

[0054] The cosine similarity between the evolution direction vector of each fault evolution trajectory and the evolution direction vector of the real-time evolution trajectory is calculated as the matching degree of the real-time evolution trajectory and each fault evolution trajectory. The fault evolution trajectory corresponding to the maximum value of the matching degree is marked as the similar fault trajectory.

[0055] If the matching degree of the real-time evolution trajectory and the similar fault trajectory is greater than the preset matching degree threshold, the fault type of the battery is the fault type corresponding to the similar fault trajectory.

[0056] Compared with the prior art, the application has the following beneficial effects:

[0057] The application reconstructs the state parameter sequence by rhythm cycle, maps and aligns the parameters of different dates and different discharge events according to the position in the rhythm cycle, rather than relying on the natural time stamp, eliminates the adverse effects of discontinuous discharge of the base station lithium battery on the battery state trend analysis, makes the battery state change rule easier to capture, and ensures that the parameters of each rhythm cycle in the trajectory correspond to the same discharge stage of the battery, avoiding parameter mispositioning caused by cycle misjudgment.

[0058] The application constructs the reference evolution trajectory and the real-time evolution trajectory, which can timely identify the implicit degradation trend even if the battery state parameters slowly deviate, calculates the evolution direction vector of the real-time evolution trajectory, and calculates the matching degree between the direction vectors of each fault evolution trajectory, which can assist in judging the fault type based on the existing abnormal data, and further helps to improve the fault repair efficiency of the base station lithium battery. BRIEF DESCRIPTION OF DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort. Among them:

[0060] Figure 1 The flow chart of the 5G communication base station lithium battery intelligent monitoring and remote diagnosis method provided by the present application is shown in the figure.

[0061] Figure 2 The method flow chart of reconstructing the state parameter sequence based on the rhythm cycle provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0062] The technical solutions of the present application will be described in detail below with the help of the drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solutions of the present application, but not limitations of the technical solutions of the present application. The technical features in the embodiments and the embodiments can be combined with each other without conflict.

[0063] The present embodiment introduces a 5G communication base station lithium battery intelligent monitoring and remote diagnosis method, which is described with reference to Figure 1 The method comprises the following steps:

[0064] Continuously detecting the battery state parameters to establish the state parameter sequence of the battery;

[0065] The battery state parameters at least include the discharge current;

[0066] The state parameter sequence is a time sequence composed of state parameter vectors at different times;

[0067] Establishing the state parameter sequence, specifically including: setting the current threshold; continuously collecting each battery state parameter at each time with a preset sampling period; if the discharge current is greater than the current threshold, then each battery state parameter at the corresponding time is composed into a state parameter vector, and the state parameter vector at the corresponding time is filled as an element into the state parameter sequence.

[0068] In the present embodiment, the state parameter sequence only contains the state parameter vector at the actual discharge time of the battery. When it is detected that the battery is actually discharging at a certain time, i.e. the discharge current is greater than the current threshold, the vector composed of multiple battery state parameters at the corresponding time is filled as an element to the end of the state parameter sequence, i.e. as the latest element of the state parameter sequence, so that the elements in the state parameter sequence can all reflect the discharge state of the battery, reducing the invalid redundant data.

[0069] The present scheme is used for long-term monitoring of the battery state and identifying the degradation trend, thus mainly capturing the macroscopic change trend of the state parameters of the battery, without focusing on the millisecond-level or more subtle high-frequency fluctuations of the state parameters, and thus the sampling period can be set in minutes. Alternatively, a high-frequency sampling period of seconds or milliseconds can also be set, and a sliding window with a length of several minutes is set, and the mean value of the battery state parameters in each window is calculated as the battery state parameters at the corresponding time of each macroscopic sampling period, so as to form a macroscopic sampling period with a period length of minutes, and improve the accuracy of the detection of the battery state parameters.

[0070] Alternatively, the battery state parameters further include a discharge voltage, a battery internal resistance, a battery temperature, an SOH (state of health), and the like; by monitoring the multi-dimensional battery state parameters, the comprehensiveness and accuracy of the battery abnormality identification and fault diagnosis are improved.

[0071] Based on the rhythm period of the battery, the state parameter sequence is reconstructed to obtain a real-time evolution trajectory of the battery state parameters.

[0072] The identification of the rhythm period of the battery specifically includes:

[0073] The current time sequence is extracted from the state parameter sequence; the current time sequence includes discharge currents at different times.

[0074] In the present embodiment, each element in the state parameter sequence is a state parameter vector including a discharge current, other battery state parameters in the state parameter vector at each time are excluded, and only the discharge current is retained, so as to obtain the current time sequence. Compared with other battery state parameters, the discharge current is directly affected by the base station communication load, and periodic communication service fluctuations will cause rhythmic changes in the discharge current.

[0075] The current time sequence is subjected to frequency domain conversion to obtain a frequency spectrum of the discharge current.

[0076] The frequency component with the largest energy proportion in the frequency spectrum is extracted as the main frequency of the discharge current.

[0077] The change period length of the discharge current is calculated based on the main frequency, as the length of the rhythm period of the battery.

[0078] The change period length of the discharge current is the reciprocal of the main frequency. The period length of the rhythm period is represented by the number of sampling periods contained in the rhythm period.

[0079] The real-time evolution trajectory includes state parameter vectors corresponding to different rhythm periods; as shown in Figure 2 The reconstruction of the state parameter sequence based on the rhythm period specifically includes:

[0080] setting a reference parameter sequence; the reference parameter sequence is a time sequence composed of reference parameter vectors at different time points; a reference parameter vector at any time point comprises each reference state parameter at the corresponding time point; the reference state parameter is a reference value of a battery state parameter;

[0081] In this embodiment, the reference state parameter is a battery state parameter when the battery state is good; optionally, at the initial stage of using the battery, a time point when the battery starts discharging in a day is selected as a monitoring starting point, and the battery state parameter at each time point is continuously recorded as the reference state parameter at each time point.

[0082] a reference sequence segment is intercepted from the reference parameter sequence by using a sliding window; the length of the sliding window is equal to the cycle length of the rhythm cycle;

[0083] Optionally, the starting point of the sliding window is aligned with the starting point of the reference parameter sequence to intercept the reference sequence segment.

[0084] M times of sliding interception are performed on the state parameter sequence by using the sliding window, and the sliding step is 1, and each time of sliding interception obtains a sequence segment in the state parameter sequence; M is the length of the sliding window;

[0085] The similarity of the M sequence segments and the reference sequence segment is calculated respectively; the sequence segment with the maximum similarity with the reference sequence segment is marked as the first rhythm cycle;

[0086] Optionally, the similarity of each sequence segment and the reference sequence segment is determined by calculating the Euclidean distance or the cosine similarity;

[0087] The first rhythm cycle is taken as a starting point, and continuous sliding interception is performed on the state parameter sequence by using the sliding window, and the sliding step is M, and each time of sliding interception obtains a rhythm cycle, until the state parameter sequence is completely divided into different rhythm cycles, to obtain the real-time evolution trajectory.

[0088] In this embodiment, each rhythm cycle is a sequence segment with a length of M, i.e., a state parameter vector at M time points. When the historical state data is processed subsequently, the position of the first rhythm cycle is also determined by using the reference sequence segment, so that different state parameter sequences can be aligned based on the rhythm cycle. Different base stations and different power supply strategies will make the discharge of the battery present different rhythm characteristics, for example, a base station with intermittent discharge alternation between the battery and the power grid will present a relatively stable rhythm cycle, such as a rhythm cycle of 24 hours; if the base station adopts a discharge strategy of load peak clipping and valley filling, the battery will be discharged multiple times a day, presenting a shorter rhythm cycle. The present scheme does not preset the rhythm cycle as a fixed value, but automatically identifies the actual rhythm cycle by performing frequency domain analysis on the discharge current, to ensure the objectivity of constructing the real-time evolution trajectory.

[0089] The communication base station lithium battery participates in system operation in the form of auxiliary power supply, and the battery state parameters are discontinuously distributed on the time axis. The rhythm cycle is used to reconstruct the state parameter sequence in the embodiment, which can eliminate the adverse effects of fragmented sampling on trend modeling, enable the battery state parameters of different dates and different discharge events to be naturally spliced, map the natural timestamp of the battery state parameter to the position in the rhythm cycle, enable subsequent analysis to be based on the position of the battery state parameter in the rhythm cycle instead of the natural timestamp, facilitate comparison of the battery state of the same rhythm on different dates, and solve the problem in the prior art that remote monitoring of the base station lithium battery is based on a unified time step, which easily conceals rhythm-related aging characteristics or periodic load fatigue.

[0090] obtain historical state data of the battery, and construct a reference evolution track and a fault evolution track of the battery state parameter based on the historical state data;

[0091] The historical state data includes the reference parameter sequence and an abnormal state parameter; the abnormal state parameter includes the battery state parameter when the battery state is abnormal.

[0092] The method for constructing the reference evolution track of the battery state parameter is as follows: taking the reference sequence segment as a starting point, continuously performing sliding interception in the reference parameter sequence through the sliding window, and the sliding step is M, and a rhythm cycle is obtained each time the sliding interception is performed, until the reference parameter sequence is completely divided into different rhythm cycles, and the reference evolution track is obtained.

[0093] The method for constructing the fault evolution track is as follows:

[0094] The abnormal state parameter is arranged into an abnormal parameter sequence; the abnormal parameter sequence is a time sequence composed of abnormal parameter vectors at different times; and the abnormal parameter vector at any time includes each abnormal state parameter at the corresponding time.

[0095] The abnormal parameter sequence is reconstructed based on the rhythm cycle to obtain an abnormal evolution track of the abnormal state parameter; and the abnormal evolution track includes abnormal parameter vectors corresponding to different rhythm cycles.

[0096] In the embodiment, the specific method for arranging the abnormal state parameter into the abnormal parameter sequence refers to the method for establishing the state parameter sequence of the battery based on the battery state parameter; and the method for reconstructing the abnormal parameter sequence to obtain the abnormal evolution track of the abnormal state parameter refers to the method for reconstructing the state parameter sequence to obtain the real-time evolution track. The abnormal state parameters are collected under different times and different abnormal states, and multiple abnormal evolution tracks are constructed, which provide a reference for subsequent battery fault diagnosis.

[0097] Each abnormal evolution track is clustered, and the abnormal evolution tracks are divided into different cluster clusters.

[0098] Calculate the mean sequence of each abnormal evolution trajectory in each cluster as the fault evolution trajectory of each cluster.

[0099] In this embodiment, each cluster contains multiple abnormal evolution trajectories; each abnormal evolution trajectory is a sequence with a length of M, i.e., contains M abnormal parameter vectors; the mean sequence is also a sequence with a length of M, contains M mean vectors, and the mean vector at each position in the mean sequence is the mean vector of the abnormal parameter vectors at the corresponding position in each abnormal evolution trajectory in the cluster. Each cluster obtained by clustering corresponds to a type of battery failure, and the change of the battery state parameter under each type of failure is represented by the corresponding abnormal evolution trajectory. Optionally, the DBSCAN algorithm is used to cluster the multiple abnormal evolution trajectories.

[0100] Identify the degradation trend of the battery state parameter based on the real-time evolution trajectory and the reference evolution trajectory, and trigger the abnormal detection of the battery based on the degradation trend;

[0101] The identification of the degradation trend of the battery state parameter specifically includes:

[0102] Obtain a real-time evolution sequence by intercepting N consecutive rhythm cycles from the real-time evolution trajectory; obtain a reference evolution sequence by intercepting N consecutive rhythm cycles from the reference evolution trajectory; N is a positive integer; preferably, the real-time evolution sequence contains the N closest rhythm cycles to the current time.

[0103] Extract MxN vector groups from the real-time evolution sequence and the reference evolution sequence; any vector group contains a state parameter vector and a reference parameter vector, and in the same vector group, the index of the state parameter vector in the real-time evolution sequence is the same as the index of the reference parameter vector in the reference evolution sequence; for example, the ith state parameter vector in the real-time evolution sequence and the ith reference parameter vector in the reference evolution sequence form the ith vector group; i is a positive integer less than MxN.

[0104] Calculate the Euclidean distance between the state parameter vector and the reference parameter vector in each vector group as the deviation value of each vector group;

[0105] Normalize and accumulate the deviation value of each vector group to obtain the degradation value of the battery state parameter; if the degradation value is greater than a preset degradation threshold, the battery state parameter has a degradation trend.

[0106] In this embodiment, the degradation value represents the structural difference between the real-time evolution sequence and the reference evolution sequence. The reference evolution sequence records the battery state parameter in a healthy state, i.e., the reference state parameter, and the greater the structural difference between the real-time evolution sequence and the reference evolution sequence, the more obvious the degradation of the battery state parameter.

[0107] The abnormality detection specifically includes: if there is a degradation trend in the battery state parameter, triggering the abnormality detection of the battery, and the method is as follows:

[0108] The deviation values of each vector group are sorted into a deviation value sequence; optionally, the deviation values of each vector group are sorted into a deviation value sequence according to the indexes of the state parameter vectors in the vector group, so as to ensure that the deviation values in the deviation value sequence are arranged in chronological order.

[0109] Linear fitting is performed on the deviation value sequence to obtain a fitting slope; the variance of each deviation value in the deviation value sequence is calculated to obtain a deviation fluctuation value;

[0110] An abnormal tension index is calculated based on the fitting slope and the deviation fluctuation value; the abnormal tension index is the product of the fitting slope and the deviation fluctuation value;

[0111] If the abnormal tension index is greater than a preset abnormal tension threshold, the battery has an abnormality.

[0112] In this embodiment, the abnormal tension index is calculated based on the fitting slope and the deviation fluctuation value; the fitting slope reflects the upward trend of the deviation value, and the deviation fluctuation value reflects the stability of the deviation value; when the abnormal tension index is greater than a preset abnormal tension threshold, the trend of the battery state deviating from the healthy state is significant.

[0113] Based on the real-time evolution trajectory and the reference evolution trajectory, the real-time evolution direction of the battery state parameter is identified, and based on the real-time evolution direction and the fault evolution trajectory, the fault type of the battery is identified.

[0114] The real-time evolution direction is represented by an evolution direction vector of the real-time evolution trajectory; if the battery has an abnormality, the real-time evolution direction of the battery state parameter is triggered, specifically including:

[0115] The difference vector between the state parameter vector and the reference parameter vector in each vector group is calculated as the direction vector of each vector group;

[0116] The mean vector of the direction vectors of all vector groups is calculated as the evolution direction vector of the real-time evolution trajectory.

[0117] The state parameter vector and the reference parameter vector both include a vector of n elements, and n is the number of battery state parameters; the difference vector between the two is also a vector including n elements, and each element value is the difference between the corresponding position of the battery state parameter and the reference state parameter. The direction vector of any vector group represents the evolution direction of the battery state parameter at the corresponding time, that is, the deviation direction compared with the reference state parameter; the mean vector of the direction vectors of all vector groups, that is, the evolution direction vector, represents the overall evolution direction of the battery state parameter at continuous multiple times.

[0118] identify the fault type of the battery based on the real-time evolution direction and the fault evolution trajectory, specifically comprising:

[0119] respectively intercept N continuous rhythm cycles from each fault evolution trajectory to obtain a fault evolution sequence corresponding to each fault evolution trajectory;

[0120] respectively calculate an evolution direction vector of each fault evolution trajectory based on the fault evolution sequence corresponding to each fault evolution trajectory and the reference evolution sequence;

[0121] respectively calculate the cosine similarity between the evolution direction vector of each fault evolution trajectory and the evolution direction vector of the real-time evolution trajectory as the matching degree of the real-time evolution trajectory and each fault evolution trajectory; mark the fault evolution trajectory corresponding to the maximum of the matching degree as a similar fault trajectory;

[0122] If the matching degree of the real-time evolution trajectory and the similar fault trajectory is greater than a preset matching degree threshold, the fault type of the battery is the fault type corresponding to the similar fault trajectory.

[0123] Each fault evolution trajectory corresponds to a cluster of abnormal evolution trajectories, and each cluster corresponds to a fault type; when the matching degree of the real-time evolution trajectory and the similar fault trajectory is greater than a preset matching degree threshold, the evolution direction of the real-time evolution trajectory of the battery and the fault evolution trajectory in the cluster is close, and it is determined that the fault type of the battery is the fault type corresponding to the cluster.

[0124] The way to calculate the evolution direction vector of each fault evolution trajectory refers to the method for calculating the evolution direction vector of the real-time evolution trajectory. When calculating the evolution direction vector of the fault evolution trajectory, each vector group includes an abnormal parameter vector and a reference parameter vector. In the calculation of each vector group, the difference vector between the abnormal parameter vector and the reference parameter vector in each vector group is taken as the direction vector of each vector group, and the mean vector of the direction vectors of all vector groups is taken as the evolution direction vector of the fault evolution trajectory.

[0125] Existing battery remote monitoring and diagnosis is mainly based on threshold judgment, which is difficult to identify structural drift in battery performance evolution. The present application can identify whether the evolution direction of the battery state parameter is biased towards the healthy state or the fault state by establishing real-time evolution trajectories and reference evolution trajectories, abnormal evolution trajectories, thereby reducing the missed detection and misjudgment of battery faults.

[0126] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code.

[0127] The embodiments of the present application are described above with reference to the accompanying drawings, but the present application is not limited to the above-described specific embodiments, and the above-described specific embodiments are merely illustrative, not restrictive, and those of ordinary skill in the art can make many forms under the inspiration of the present application without departing from the purpose of the present application and the scope of protection, which are all within the scope of protection of the present application.

Claims

1. A method for intelligent monitoring and remote diagnosis of a lithium battery of a 5G communication base station, characterized in that: The method comprises the following steps: continuously detecting battery state parameters to establish a state parameter sequence of the battery; the battery state parameters at least include a discharge current; the state parameter sequence is a time sequence composed of state parameter vectors at different time points; establishing the state parameter sequence specifically comprises: setting a current threshold; continuously collecting each battery state parameter at each time point at a preset sampling period; if the discharge current is greater than the current threshold, then each battery state parameter at the corresponding time point is composed into a state parameter vector, and the state parameter vector at the corresponding time point is filled into the state parameter sequence as an element; based on the state parameter sequence, identifying a rhythm period of the battery, and based on the rhythm period, reconstructing the state parameter sequence to obtain a real-time evolution track of the battery state parameters; the identification of the rhythm period of the battery specifically comprises: extracting a current time sequence from the state parameter sequence; the current time sequence includes discharge currents at different time points; performing frequency domain conversion on the current time sequence to obtain a frequency spectrum of the discharge current; extracting a frequency component with the largest energy proportion in the frequency spectrum as a main frequency of the discharge current; based on the main frequency, calculating a change period length of the discharge current as a length of the rhythm period of the battery; the real-time evolution track includes state parameter vectors corresponding to different rhythm periods; based on the rhythm period, reconstructing the state parameter sequence specifically comprises: setting a reference parameter sequence; the reference parameter sequence is a time sequence composed of reference parameter vectors at different time points; a reference parameter vector at any time point contains each reference state parameter at the corresponding time point; the reference state parameter is a reference value of the battery state parameter; using a sliding window to intercept a reference sequence segment in the reference parameter sequence; the length of the sliding window is equal to the period length of the rhythm period; through the sliding window, performing M times of sliding interception in the state parameter sequence, and the sliding step is 1; each time of sliding interception obtains a sequence segment in the state parameter sequence; M is the length of the sliding window; respectively calculating the similarity of M sequence segments and the reference sequence segment; marking the sequence segment with the largest similarity to the reference sequence segment as the first rhythm period; taking the first rhythm period as a starting point, through the sliding window, continuously performing sliding interception in the state parameter sequence, and the sliding step is M; each time of sliding interception obtains a rhythm period, until the state parameter sequence is completely divided into different rhythm periods, to obtain the real-time evolution track; obtaining historical state data of the battery, and based on the historical state data, constructing a reference evolution track and a fault evolution track of the battery state parameters; based on the real-time evolution track and the reference evolution track, identifying a degradation trend of the battery state parameters, and based on the degradation trend, triggering an abnormal detection of the battery; based on the real-time evolution track and the reference evolution track, identifying a real-time evolution direction of the battery state parameters, and based on the real-time evolution direction and the fault evolution track, identifying a fault type of the battery. 2.The 5G communication base station lithium battery intelligent monitoring and remote diagnosis method of claim 1, wherein: the historical state data includes the reference parameter sequence; The method for constructing the reference evolution track of the battery state parameter is as follows: taking the reference sequence segment as a starting point, continuously performing sliding interception in the reference parameter sequence through the sliding window, and the sliding step is M, and one rhythm period is obtained each time the sliding interception is performed, until the reference parameter sequence is completely divided into different rhythm periods, and the reference evolution track is obtained. 3.The 5G communication base station lithium battery intelligent monitoring and remote diagnosis method of claim 2, wherein: The historical state data further includes an abnormal state parameter; the abnormal state parameter includes a battery state parameter when a battery state is abnormal; The method for constructing the fault evolution track is as follows: The abnormal state parameter is sorted into an abnormal parameter sequence; the abnormal parameter sequence is a time sequence composed of abnormal parameter vectors at different times; an abnormal parameter vector at any time includes each abnormal state parameter at the corresponding time; The abnormal parameter sequence is reconstructed based on the rhythm period, and an abnormal evolution track of the abnormal state parameter is obtained; the abnormal evolution track includes abnormal parameter vectors corresponding to different rhythm periods; Each abnormal evolution track is clustered, and the abnormal evolution tracks are divided into different clustering clusters; The mean sequence of each abnormal evolution track in each clustering cluster is calculated, as a fault evolution track of each clustering cluster. 4.The 5G communication base station lithium battery intelligent monitoring and remote diagnosis method of claim 3, wherein: The deterioration trend of the battery state parameter is identified, and specifically includes: N consecutive rhythm periods are intercepted from the real-time evolution track, and a real-time evolution sequence is obtained; N consecutive rhythm periods are intercepted from the reference evolution track, and a reference evolution sequence is obtained; N is a positive integer; M×N vector groups are extracted from the real-time evolution sequence and the reference evolution sequence; any vector group includes a state parameter vector and a reference parameter vector, and the index of the state parameter vector in the real-time evolution sequence is the same as the index of the reference parameter vector in the reference evolution sequence in the same vector group; The Euclidean distance between the state parameter vector and the reference parameter vector in each vector group is calculated as the deviation value of each vector group; The deviation value of each vector group is normalized and accumulated to obtain a deterioration value of the battery state parameter; if the deterioration value is greater than a preset deterioration threshold, the battery state parameter has a deterioration trend. 5.The 5G communication base station lithium battery intelligent monitoring and remote diagnosis method of claim 4, wherein: The abnormal detection specifically includes: if the battery state parameter has a deterioration trend, triggering the abnormal detection of the battery, and the method is as follows: The deviation value of each vector group is sorted into a deviation value sequence; The deviation value sequence is linearly fitted to obtain a fitting slope; the variance of each deviation value in the deviation value sequence is calculated to obtain a deviation fluctuation value; An abnormal tension index is calculated based on the fitting slope and the deviation fluctuation value; the abnormal tension index is the product of the fitting slope and the deviation fluctuation value; If the abnormal tension index is greater than a preset abnormal tension threshold, the battery is abnormal. 6.The 5G communication base station lithium battery intelligent monitoring and remote diagnosis method of claim 5, wherein: The real-time evolution direction is represented by an evolution direction vector of the real-time evolution track; if the battery is abnormal, the real-time evolution direction of the battery state parameter is identified, and specifically includes: The difference vector between the state parameter vector and the reference parameter vector in each vector group is calculated as the direction vector of each vector group; The mean vector of the direction vectors of all vector groups is calculated as the evolution direction vector of the real-time evolution track.

7. The 5G communication base station lithium battery intelligent monitoring and remote diagnosis method of claim 6, wherein: Identify the fault type of the battery based on the real-time evolution direction and the fault evolution trajectory, specifically comprising: Respectively, from each fault evolution trajectory, N consecutive rhythm cycles are intercepted to obtain the fault evolution sequence corresponding to each fault evolution trajectory; Respectively, based on the fault evolution sequence corresponding to each fault evolution trajectory and the reference evolution sequence, the evolution direction vector of each fault evolution trajectory is calculated; Respectively, the cosine similarity between the evolution direction vector of each fault evolution trajectory and the evolution direction vector of the real-time evolution trajectory is calculated as the matching degree of the real-time evolution trajectory and each fault evolution trajectory; The fault evolution trajectory corresponding to the maximum value of the matching degree is marked as the similar fault trajectory; If the matching degree of the real-time evolution trajectory and the similar fault trajectory is greater than the preset matching degree threshold, the fault type of the battery is the fault type corresponding to the similar fault trajectory.

Images