Real-time monitoring method for abnormal operation state of energy storage power generation system

By collecting and analyzing the operating and environmental parameters of the battery pack in the energy storage system, using cloud platform and machine learning models for fault prediction and dynamic adjustment, the problem of insufficient threshold judgment in the existing technology is solved, and the accuracy of the energy storage system is achieved Monitor and efficient operation and maintenance.

CN119030151AInactive Publication Date: 2024-11-26BEIJING WARMLAND ENERGY SERVICE CO LTD
View PDF 0 Cites 10 Cited by

Patent Information

Application Number
CN202411363750.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-28
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing energy storage system monitoring methods rely on fixed threshold judgments, are difficult to adapt to complex working conditions and potential failures, lack of fine distinction and prediction capabilities, resulting in insufficient accuracy and timeliness of early warning signals.

Method used

By collecting the operating parameters and environmental parameters of the battery pack, using the cloud platform to conduct preliminary judgment and fault prediction, introducing dynamic control range and machine learning models, issuing different levels of early warning signals and adjustment suggestions, and dynamically adjusting the operating parameters and heat dissipation parameters of the battery pack .

Benefits of technology

It realizes accurate monitoring and effective response to abnormal operating status of the energy storage system, improves the safety and stability of the system, reduces unplanned downtime, extends the service life of the battery pack, and reduces operation and maintenance costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119030151A_ABST
    Figure CN119030151A_ABST
Patent Text Reader

Abstract

The invention provides a real-time monitoring method for an abnormal operation state of an energy storage power generation system, and relates to the technical field of energy storage monitoring, and the method comprises the steps: collecting operation parameters of a battery pack through a first sensor group; acquiring environmental parameters of the battery pack through a second sensor group; uploading the key parameters and the environmental parameters to a platform; the cloud platform performs first judgment on the operation parameters through a control model of each parameter to obtain an abnormal mark; according to the operation parameters, a fault is predicted through a cloud platform and a fault prediction model, and a fault prediction result is obtained; according to the abnormal mark and / or the fault prediction result, an early warning is given, and an energy storage system adjustment suggestion is given; according to the method, operation parameters and environmental parameters of the battery pack can be comprehensively considered, fault prediction is carried out by dynamically adjusting the control range and utilizing machine learning, and specific adjustment suggestions are given, so that the safety and reliability of the energy storage system are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage monitoring, and in particular to a method for real-time monitoring of abnormal operating status of an energy storage power generation system. Background Art

[0002] With the rapid development of renewable energy, energy storage systems are becoming increasingly important as a key technology to balance grid supply and demand and improve energy efficiency. However, energy storage systems, especially battery energy storage systems, may face a variety of complex operating conditions and potential failures during operation, such as battery overheating, overcharging, over-discharging, increased internal resistance, etc. These problems will not only affect the performance of the energy storage system, but may also cause safety accidents.

[0003] Most existing energy storage system monitoring methods rely on fixed threshold judgments, that is, when a certain operating parameter exceeds a preset threshold, the system issues an alarm or warning signal. However, this method has many limitations: first, the threshold setting is often based on experience or design specifications, which is difficult to adapt to the complex changes in actual operation; second, it lacks the ability to finely distinguish and predict different types of faults, resulting in insufficient accuracy and timeliness of warning signals; third, it fails to fully utilize historical data and real-time data for in-depth analysis and mining to extract more valuable information.

[0004] Therefore, a more intelligent and comprehensive real-time monitoring method for the abnormal operating status of energy storage power generation systems is needed to improve the safety and reliability of energy storage systems. Summary of the invention

[0005] Based on the above problems, the present invention provides a real-time monitoring method for the abnormal operating status of an energy storage power generation system. The operating parameters and environmental parameters of the battery pack are collected and uploaded to the cloud platform for preliminary judgment and fault prediction. Different levels of early warning signals and adjustment suggestions are issued according to abnormal marks and prediction results. Advanced technologies such as dynamic control range, machine learning model and parameter weight calculation are introduced to achieve accurate monitoring and effective response to the abnormal operating status of the energy storage system.

[0006] The purpose of the present invention is achieved by the following technical solutions: The present invention provides a method for real-time monitoring of abnormal operation status of an energy storage power generation system, the method comprising: S1. Collecting the operating parameters of the battery pack through the first sensor group; acquiring the environmental parameters of the battery pack through the second sensor group; S2. Upload the key parameters and environmental parameters to the platform; the cloud platform performs a first determination on the operating parameters through the control model of each parameter to obtain an abnormality mark; the control model includes a first control range, a second control range and a third control range; the abnormality mark includes a first abnormality, a second abnormality and a third abnormality; S3, predicting the fault through a fault prediction model on the cloud platform according to the operating parameters to obtain a fault prediction result; S4. Based on the abnormal markings and / or fault prediction results, an early warning is issued and suggestions for adjusting the energy storage system are given. The adjustment suggestions include adjusting the battery pack operating parameters and heat dissipation parameters.

[0007] Preferably, the S1 includes: Collecting operating parameters of the battery pack during charging or discharging by means of a first sensor group installed in the battery pack, wherein the operating parameters include temperature, energy consumption, internal resistance, voltage, current, and charging or discharging rate; The environmental parameters of the battery pack are acquired through the second sensor group; the environmental parameters include environmental temperature and environmental humidity.

[0008] Preferably, the S2 includes: Through design specifications and historical data, control models for each operating parameter are established; The upper and lower limits of each key parameter are obtained through design specifications as the first control range.

[0009] Preferably, the second control range includes a second control range for a charging process and a second control range for a discharging process; The method for obtaining the second control range of the charging process is as follows: Obtaining a curve of a parameter changing with charging time during a single charging process of the battery pack under historical normal operation as a first curve; Establishing a first model of the parameter by using a plurality of first curves, environmental parameters corresponding to the first curves, and the number of charging times; According to the current charging time, environmental parameters, load and charging times, based on the first model of the parameter, a range of the parameter under the charging time is obtained as a second control range of the charging process of the parameter; The method for obtaining the second control range of the discharge process is as follows: Obtaining a curve of a parameter changing with discharge time during a single discharge of the battery pack under historical normal operation as the second curve; Establishing a second prediction model for the parameter through a plurality of second curves, environmental parameters corresponding to the second curves, loads, and discharge times; Through the current discharge duration, environmental parameters, load and number of charges, based on the second model of the parameter, a range of the parameter under the discharge duration is obtained as the second range of the discharge process of the parameter.

[0010] Preferably, the third control range includes: The difference between two adjacent samples of each parameter during the charging process or the discharging process is obtained through historical data as a first difference; and a third control range is obtained through multiple first differences; The third control range is: DeltaCmini=DeltaCai- (k+max(Fc / Fmax,T c / Tmax))*σidelta DeltaCmaxi=DeltaCai+( k+ max(Fc / Fmax,T c / Tmax))*σidelta Where DeltaCmini is the minimum value of the third control range of the i-th parameter, DeltaCmaxi is the maximum value of the third control range of the i-th parameter, DeltaCai is the mean of multiple first differences of the i-th parameter; σidelta is the standard deviation of multiple first differences of the i-th parameter, Fc is the current load of the system; Fmax is the maximum load of the system, Tc is the current temperature of the battery pack; Tmax is the maximum allowable temperature of the battery pack, k is the adjustment coefficient, 1 <k≤2。

[0011] Preferably, the S2 further includes: If any parameter exceeds the first control range, a first abnormality is marked; If any parameter exceeds the second control range, a second abnormality is marked; If any parameter exceeds the third control range, a third abnormality is flagged.

[0012] Preferably, S3 includes: Obtaining fault data of each fault in historical data, wherein the fault data includes type, fault occurrence stage, fault occurrence frequency, and duration; wherein the fault occurrence stage includes a charging stage and a discharging stage; Obtain operating parameters, environmental parameters and time series data of the battery pack related to the fault data; The fault prediction results are obtained through machine learning based on the fault data, the operating parameters of the battery pack related to the fault data, the environmental parameters and the time series data.

[0013] Preferably, the S3 further includes: Compare the parameters corresponding to the fault with the normal operating data to obtain the second coefficient; The second coefficient is: Dei2=abs(Dfaei- Dnai) / Dnai Among them, Dei2 is the second coefficient, Dfaei is the mean value of the ith parameter of the eth fault type at the time of the fault; Dnai is the mean value of the ith parameter under normal operation; abs() is the absolute value; Perform correlation analysis on each parameter and obtain the first correlation coefficient; If the second difference or the first correlation coefficient is greater than a preset threshold, the corresponding parameter is used as an input parameter of the machine learning model; Obtaining the weight of each input parameter through the second coefficient and the first correlation coefficient of each input parameter; Among them, the weights of the input parameters are:

[0014] in, The weights of the input parameters for the i-th machine learning model, is the first correlation coefficient of the i-th parameter of the e-th fault type; E is the fault type, I is the total number of input parameters in the machine learning model; y1 and y2 are the second weight coefficients.

[0015] Preferably, the S4 includes: issuing a first alarm for the first abnormality; issuing a second warning for the second abnormality; issuing a third warning for the third abnormality; According to the forecast results, the fourth warning is issued; If the temperature is abnormal, adjust the heat dissipation parameters in the heat dissipation device.

[0016] The present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any method described in the present invention when executing the computer program.

[0017] The beneficial effects of the present invention include: by real-time monitoring of the operating parameters and environmental parameters of the battery pack, and abnormality determination based on multi-level control ranges (first control range, second control range, third control range), potential operating abnormalities can be discovered and warned in time, thereby avoiding system failures or accidents caused by overheating, overcharging, over-discharging, etc., and improving the safety and stability of the entire energy storage power generation system. The machine learning algorithm is used to deeply analyze the historical fault data, and the system faults are accurately predicted in combination with the current operating parameters and environmental parameters. This prediction mechanism can identify potential fault risks in advance and give corresponding adjustment suggestions, thereby preventing faults, reducing unplanned downtime, and improving the availability and operation efficiency of the system. The centralized processing and analysis of data through the cloud platform not only improves the efficiency and accuracy of data processing, but also realizes the intelligent operation and maintenance management of the energy storage power generation system. The operation and maintenance personnel can quickly respond to and handle system abnormalities based on the warning information and adjustment suggestions provided by the cloud platform, reducing the difficulty and cost of operation and maintenance. By real-time monitoring and adjusting the operating parameters and heat dissipation parameters of the battery pack, the battery pack can be finely managed according to the actual operation conditions and its performance can be optimized. This helps to extend the service life of the battery pack, improve energy storage efficiency, and reduce overall operating costs. When an abnormality is found or a fault is predicted, the system can quickly issue an early warning and give adjustment suggestions, so that the operation and maintenance personnel can take immediate measures to deal with it. This rapid response mechanism helps to reduce system downtime and improve system reliability and stability. By dynamically adjusting the control range, this method can more accurately reflect the operating status of the battery pack under different working conditions, avoiding the false alarm and missed alarm problems caused by fixed threshold judgment. At the same time, combined with the machine learning model for fault prediction, potential faults can be discovered in advance, improving the timeliness and accuracy of the early warning. By introducing the second coefficient and the first correlation coefficient, and determining the weights of the machine learning model input parameters accordingly, the accuracy of fault identification is improved, the model input is optimized, the model prediction accuracy is improved, and the model generalization ability is enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of a method for real-time monitoring of abnormal operating status of an energy storage power generation system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0020] See also Figure 1 Some embodiments of the present invention provide a method for real-time monitoring of abnormal operation status of an energy storage power generation system, the method comprising: S1. Collecting the operating parameters of the battery pack through the first sensor group; acquiring the environmental parameters of the battery pack through the second sensor group; S2. Upload the key parameters and environmental parameters to the platform; the cloud platform performs a first determination on the operating parameters through the control model of each parameter to obtain an abnormality mark; the control model includes a first control range, a second control range and a third control range; the abnormality mark includes a first abnormality, a second abnormality and a third abnormality; S3, predicting the fault through a fault prediction model on the cloud platform according to the operating parameters to obtain a fault prediction result; S4. Based on the abnormal markings and / or fault prediction results, an early warning is issued and suggestions for adjusting the energy storage system are given. The adjustment suggestions include adjusting the battery pack operating parameters and heat dissipation parameters.

[0021] In some embodiments, the S1 includes: Collecting operating parameters of the battery pack during charging or discharging by means of a first sensor group installed in the battery pack, wherein the operating parameters include temperature, energy consumption, internal resistance, voltage, current, and charging or discharging rate; The environmental parameters of the battery pack are acquired through the second sensor group; the environmental parameters include environmental temperature and environmental humidity.

[0022] The working principle and effect of the above technical solution are: through the first sensor group installed on the battery pack, the key operating parameters of the battery pack during the charging or discharging process are collected in real time. These parameters include but are not limited to temperature, energy consumption, internal resistance, voltage, current, and charging or discharging rate. These parameters directly reflect the current working state and performance of the battery pack. At the same time, the environmental parameters of the environment where the battery pack is located, such as ambient temperature and ambient humidity, are obtained through the second sensor group. Environmental parameters have an important impact on the performance and working state of the battery pack, and are therefore also an important basis for monitoring and evaluating the health status of the battery pack. The collected key operating parameters and environmental parameters are uploaded to the cloud platform. As the center of data processing and analysis, the cloud platform is responsible for receiving, storing and processing these data. The cloud platform uses a pre-established control model to make a first judgment on the uploaded operating parameters. The control model includes three control ranges: the first control range, the second control range, and the third control range.

[0023] The cloud platform uses the fault prediction model to further analyze the operating parameters and predict possible future faults. The fault prediction model is trained and optimized through machine learning algorithms based on historical fault data, relevant operating parameters, environmental parameters, and time series data. The prediction results include information such as fault type, probability of occurrence, and expected time of occurrence.

[0024] Based on the abnormal markings and / or fault prediction results, the cloud platform issues a warning signal and gives suggestions for adjusting the energy storage system. The adjustment suggestions are intended to reduce the risk of failure or alleviate abnormal conditions that have occurred by adjusting the operating parameters of the battery pack (such as charging / discharging rate, voltage, etc.) and heat dissipation parameters (such as fan speed, heat sink temperature, etc.). Operation and maintenance personnel can take corresponding measures based on these suggestions to ensure the safe and stable operation of the energy storage system.

[0025] In some embodiments, the S2 includes: Through design specifications and historical data, control models for each operating parameter are established; The upper and lower limits of each key parameter are obtained through design specifications as the first control range.

[0026] In some embodiments, the second control range includes a second control range for a charging process and a second control range for a discharging process; The method for obtaining the second control range of the charging process is as follows: Obtaining a curve of a parameter changing with charging time during a single charging process of the battery pack under historical normal operation as a first curve; Establishing a first model of the parameter by using a plurality of first curves, environmental parameters corresponding to the first curves, and the number of charging times; According to the current charging time, environmental parameters, load and charging times, based on the first model of the parameter, a range of the parameter under the charging time is obtained as a second control range of the charging process of the parameter; The method for obtaining the second control range of the discharge process is as follows: Obtaining a curve of a parameter changing with discharge time during a single discharge of the battery pack under historical normal operation as the second curve; Establishing a second prediction model for the parameter through a plurality of second curves, environmental parameters corresponding to the second curves, loads, and discharge times; Through the current discharge duration, environmental parameters, load and number of charges, based on the second model of the parameter, a range of the parameter under the discharge duration is obtained as the second range of the discharge process of the parameter.

[0027] The working principle of the above technical solution is: first, according to the design specifications of the battery pack (such as the parameter range provided by the manufacturer, safety standards, etc.), the upper and lower limits of each key parameter are directly determined. These upper and lower limits constitute the first control range, which is the basis for judging whether the battery pack is in an extremely abnormal state.

[0028] Although the upper and lower limits of the parameters have been defined in the design specifications, in actual applications, the actual operating parameters of the battery pack may fluctuate within a certain range due to the influence of various factors such as environment, load, and usage habits. Therefore, by collecting and analyzing historical normal operation data, the parameter range in the design specifications can be further verified and adjusted, but this step is usually performed in the early stage of system deployment to calibrate and optimize the first control range.

[0029] From the historical normal operation data, the variation curves of a certain parameter (such as voltage, current, temperature, etc.) during a single charging process of the battery pack with the charging time are extracted, and these curves are used as the first curves.

[0030] Based on the multiple first curves, corresponding environmental parameters (such as ambient temperature, humidity), load conditions, and charging times, a first model of the parameter is established using a statistical method or a machine learning algorithm. This model can reflect the change pattern of the parameter with charging time, environment, load, and charging times.

[0031] In the real-time operation process, according to the current charging time, environmental parameters, load and charging times, the first model is used to calculate the reasonable range of the parameter under the current charging time, that is, the second control range of the charging process. If the actual parameter value exceeds this range, it is marked as the second abnormality.

[0032] Similarly, for the discharge process, the second prediction model is established by collecting the discharge parameter change curve in the historical normal operation data. This model is used to reflect the change law of the parameters with the discharge time, environment, load and discharge times.

[0033] In the real-time operation process, according to the current discharge time, environmental parameters, load and (note that this is the number of discharges, not the number of charges), the reasonable range of the parameter under the current discharge time is calculated based on the second model, that is, the second control range of the discharge process. If the actual parameter value exceeds this range, it is also marked as the second abnormality.

[0034] Through the above method, not only the basic parameter control range (first control range) is established using the design specifications, but also a more refined and dynamic second control range of the charging and discharging process is established through in-depth analysis of historical data and current operating conditions. Such a control model can more accurately reflect the actual operating status of the battery pack, detect potential anomalies in a timely manner, and provide strong support for subsequent early warnings and adjustment suggestions.

[0035] In some embodiments, a first model of the parameter is established through machine learning using multiple first curves, environmental parameters corresponding to the first curves, and the number of charges; and a second prediction model of the parameter is established through machine learning using multiple second curves, environmental parameters corresponding to the second curves, loads, and the number of discharges.

[0036] In some other embodiments, the first model is as follows: Cminijc = (w1 * Caijc + w2 * Ci(j - 1)c) - (k + xijc) * σijc Cmaxijc = (w1 * Caijc + w2 * Ci(j - 1)c) + (k + xijc) * σijc xijc = max(xijc1, xijc2) / (xijc1 + xijc2) xijc1 = max(pvi * (pvjc - pvjac) / pvjac) xijc2 = (Fc - Fa) / Fmax + (Ncc - Nac) / Nmaxc Wherein, Cminijc is the minimum value of the second control range of the i-th operating parameter at the j-th charging duration node; Cmaxijc is the maximum value of the second control range of the i-th operating parameter at the j-th charging duration node; Caijc is the historical mean of the i-th operating parameter during charging at the j-th charging duration node; σijc is the standard deviation of the i-th operating parameter during charging at the j-th charging duration node in the historical data; Ci(j - 1)c is the value of the i-th operating parameter at the (j - 1)-th charging duration node; w1 and w2 are both the first weight coefficients, k is an adjustment coefficient, 1 < k ≤ 2, xijc is the adjustment factor of the i-th operating parameter at the j-th charging duration node; xijc1 is the first adjustment factor of the i-th parameter at the j-th charging duration node, and xijc2 is the second adjustment factor of the i-th parameter at the j-th charging duration node; pvi is the correlation coefficient between the v-th environmental parameter and the i-th operating parameter; pvjc is the value of the v-th environmental parameter at the j-th charging duration node; pvjac is the historical mean of the v-th environmental parameter at the j-th charging duration node; max(pvi * (pvjc - pvjac) / pvjac) is the maximum value among v pvi * (pvjc - pvjac) / pvjac; Fc is the current load of the system; Fmax is the maximum load of the system; Ncc is the current charging times of the battery pack, Nac is the average charging times obtained from historical data; Nmaxc is the calibrated maximum charging times of the battery pack.

[0037] The second model is similar to the first model; the second model is as follows: Cminijf = (w1 * Caijf + w2 * Ci(j - 1)f) - (k + xijf) * σijf Cmaxijf = (w1 * Caijf + w2 * Ci(j - 1)f) + (k + xijf) * σijf xijf = max(xijf1, xijf2) / (xijf1 + xijf2) xijf1 = max(pvi * (pvjf - pvjaf) / pvjaf) xijf2 = (Fc - Fa) / Fmax + (Ncf - Naf) / Nmaxf Where, Cminijf is the minimum value of the second control range of the i-th operating parameter at the j-th discharge duration node; Cmaxijf is the maximum value of the second control range of the i-th operating parameter at the j-th discharge duration node; Caijf is the historical mean value of the i-th operating parameter at the j-th discharge duration node during the discharge process; σijf is the standard deviation of the i-th operating parameter at the j-th discharge duration node in the historical data during the discharge process; Ci(j - 1)f is the value of the i-th operating parameter at the (j - 1)-th discharge duration node; w1 and w2 are both the first weight coefficients, k is the adjustment coefficient, 1 < k ≤ 2, xijf is the adjustment factor of the i-th operating parameter at the j-th discharge duration node; xijf1 is the first adjustment factor of the i-th parameter at the j-th discharge duration node, xijf2 is the second adjustment factor of the i-th parameter at the j-th discharge duration node; pvi is the correlation coefficient between the v-th environmental parameter and the i-th operating parameter; pvjf is the value of the v-th environmental parameter at the j-th discharge duration node; pvjaf is the historical mean value of the v-th environmental parameter at the j-th discharge duration node; max(pvi * (pvjf - pvjaf) / pvjaf) is the maximum value among v pvi * (pvjf - pvjaf) / pvjaf; Fc is the current load of the system; Fmax is the maximum load of the system; Ncf is the current number of discharges of the battery pack, Naf is the average number of discharges obtained from historical data; Nmaxf is the calibrated maximum number of discharges of the battery pack.

[0038] The principle and effect of the above technical solution are as follows: This model is mainly used to predict and control the dynamic range of operating parameters during the charging process to ensure the safety and efficiency of the charging process. The model dynamically adjusts the control range of these operating parameters by combining historical data, current environmental parameters, system load, and the charging state of the battery pack.

[0039] During the charging process: Cminijc and Cmaxijc: These two variables represent the minimum and maximum values ​​of the second control range of the i-th operating parameter at the j-th charging duration node. This range is weighted based on the historical mean (Caijc) and the value at the previous time point (Ci(j-1)), and takes into account a dynamic adjustment factor (xijc) and standard deviation (σij) to adapt to changes in current conditions. The adjustment factor xijc consists of two parts (xijc1 and xijc2), which reflect the impact of environmental parameter changes and battery pack charging status on operating parameters. xijc1 captures the immediate changes in environmental parameters (such as temperature, humidity, etc.), while xijc2 takes into account the long-term impact of system load and battery pack charging times.

[0040] Xijc1: Evaluate the immediate impact of environmental parameters on operating parameters by calculating the deviation of the current environmental parameter value (pvjc) from the historical mean (pvjac) and multiplying it by the correlation coefficient (pvi). Take the maximum value of all v environmental parameter effects to ensure that the most significant environmental changes are taken into account.

[0041] xijc2: Combines the ratio of the system's current load (Fc) to the maximum load (Fmax), and the comparison of the current battery pack charge count (Ncc) to the average charge count (Nac) and the maximum charge count (Nmaxc) to evaluate the impact of system load and battery aging on operating parameters.

[0042] By dynamically adjusting the control range of operating parameters, the model can ensure that operating parameters remain within a safe range when environmental parameters change or system load increases, thereby avoiding potential risks such as overcharging and overheating.

[0043] The model takes into account the charging state and aging of the battery pack and optimizes the charging process by adjusting the control range, reducing unnecessary energy loss and improving charging efficiency.

[0044] By taking into account multiple factors (including historical data, current environmental parameters, system load and battery status), the model can more accurately predict and control the operating parameters during the charging process, making the system more stable and reliable.

[0045] In some embodiments, the third control range includes: Using historical data, a difference between two adjacent samples of each parameter during a charging process or a discharging process is obtained as a first difference; and a third control range is obtained through multiple first differences; The third control range is: DeltaCmini=DeltaCai- (k+max(Fc / Fmax,T c / Tmax))*σidelta DeltaCmaxi=DeltaCai+( k+ max(Fc / Fmax,T c / Tmax))*σidelta Where DeltaCmini is the minimum value of the third control range of the i-th parameter, DeltaCmaxi is the maximum value of the third control range of the i-th parameter, DeltaCai is the mean of multiple first differences of the i-th parameter; σidelta is the standard deviation of multiple first differences of the i-th parameter, Fc is the current load of the system; Fmax is the maximum load of the system, Tc is the current temperature of the battery pack; Tmax is the maximum allowable temperature of the battery pack, k is the adjustment coefficient, 1 <k≤2。

[0046] The working principle and effect of the above technical solution are as follows: First, by collecting the historical data of various parameters of the battery pack during charging or discharging, the difference between two adjacent samples (i.e., the first difference) is calculated. These differences reflect the natural fluctuation of the parameters. Then, the mean (DeltaCai) and standard deviation (σidelta) of these first differences are calculated. The mean represents the average level of parameter changes, while the standard deviation measures the degree of dispersion of such changes.

[0047] Taking into account the impact of the current system load (Fc) and battery pack temperature (Tc) on the parameter variation range, the control range is dynamically adjusted by introducing the adjustment coefficient k and the maximum values ​​of the maximum load ratio (Fc / Fmax) and temperature ratio (Tc / Tmax). This ensures that when the load is heavy or the temperature is high, the control range can be appropriately relaxed to cope with possible parameter fluctuations; when the load is light or the temperature is low, a tighter control range is maintained to improve the stability and safety of the system. Finally, based on the parameters calculated above, the third control range of each parameter (i.e. DeltaCmini to DeltaCmaxi) is determined. This range takes into account the natural fluctuations of the parameters and combines the current operating conditions of the system, providing a scientific basis for the monitoring and protection of the battery pack.

[0048] By dynamically adjusting the control range, we can better adapt to changes in system operating conditions and reduce system instability caused by excessive parameter fluctuations.

[0049] In some embodiments, the S2 includes: If any parameter exceeds the first control range, a first abnormality is marked; If any parameter exceeds the second control range, a second abnormality is marked; If any parameter exceeds the third control range, a third abnormality is flagged.

[0050] The working principle and effect of the above technical solution are: Based on the design specifications and historical data of the battery pack, the upper and lower limits of each key parameter are set. If any parameter exceeds this range, it is marked as the first abnormality, indicating that the battery pack may be in a serious abnormal state. This usually means that immediate attention and possible corrective measures are required.

[0051] Dynamically adjust the control range of each parameter based on historical normal operation data. For the charging and discharging processes, establish a second control range respectively. By considering factors such as the current charging / discharging time, environmental parameters, load, and charging / discharging times, calculate the reasonable range of the current parameter value. If the parameter value exceeds this range, it is marked as a second abnormality, indicating that the battery pack may be in a slightly abnormal or potentially risky state. Further observation or preventive measures may be required.

[0052] The third control range is set based on the difference between two adjacent samples of each parameter in the historical data. This range is used to detect sudden changes in parameter values. If the change in parameter value exceeds this range, it is marked as the third abnormality, indicating that the battery pack may be affected by external interference or internal faults. The third control range is set based on the difference between two adjacent samples of each parameter in the historical data. The third control range is used to capture larger parameter changes. Although these changes may not necessarily cause system failures immediately, they may indicate that the system is approaching its operating limits or that there are potential problems. When the parameter value exceeds the third control range, it indicates that the battery pack may be affected by external interference or internal faults. The system is marked as the third abnormality, which may require long-term monitoring or planned maintenance.

[0053] By setting multiple control ranges, the system can issue early warnings when abnormal changes occur in parameters, thereby providing users or maintenance personnel with enough time to take countermeasures to avoid the occurrence or expansion of faults. Different abnormality markers help users or systems understand the current status more accurately and take corresponding management measures according to the severity of the abnormality. For example, for the first abnormality, it may be necessary to shut down the machine for maintenance immediately; while for the second and third abnormalities, it may only require enhanced monitoring and planned maintenance.

[0054] By monitoring parameter changes over a long period of time and recording abnormality marks, the operating trends and potential problems of the system can be analyzed, thereby optimizing maintenance plans and improving system reliability and service life. Timely abnormality detection and response can reduce the impact of system failures on users and improve user satisfaction and trust in the product. At the same time, precise management and optimized maintenance can also reduce maintenance costs and improve overall economic benefits.

[0055] In summary, by setting multiple control ranges and monitoring parameter changes in real time, the system can sense and respond to potential problems to varying degrees, thereby ensuring the stable operation of the battery pack and system, and improving user experience and economic benefits.

[0056] In some embodiments, the S3 includes: Obtaining fault data of each fault in historical data, wherein the fault data includes type, fault occurrence stage, fault occurrence frequency, and duration; wherein the fault occurrence stage includes a charging stage and a discharging stage; Obtain operating parameters, environmental parameters and time series data of the battery pack related to the fault data; The fault prediction results are obtained through machine learning based on the fault data, the operating parameters of the battery pack related to the fault data, the environmental parameters and the time series data.

[0057] The working principle and effect of the above technical solution are: First, collect detailed information about each fault in the historical data, including the type of fault, the specific stage of the fault (charging stage or discharging stage), the frequency of the fault, and the duration of each fault. This information is the basis for understanding the fault mode and making predictions. At the same time, obtain the battery pack operating parameters related to the fault data, such as voltage, current, temperature, etc. These parameters may change significantly before and after the fault occurs and are important inputs to the prediction model. Environmental parameters such as temperature and humidity may also affect the performance and failure mode of the battery pack, so they also need to be collected and recorded. Time series data records the changes in the battery pack status over time and is the key to capturing fault precursors and making predictions.

[0058] Clean the collected data to remove noise and outliers. Standardize or normalize the data to ensure that data of different dimensions can be fairly compared in the same model. It may also be necessary to extract and select features from the data to build a more representative feature set.

[0059] Choose appropriate machine learning algorithms (such as random forests, gradient boosted trees, neural networks, etc.) that can handle complex data relationships and perform pattern recognition. Use the preprocessed data as input to train the machine learning model. During the training process, the model learns the complex relationships between fault data, operating parameters, environmental parameters, and time series data, and attempts to identify the patterns and regularities that lead to faults. Once the model is trained, new operating parameters, environmental parameters, and time series data can be used as input to predict faults. The model outputs a prediction result indicating the probability or risk level of a certain fault in the battery pack within a certain period of time in the future.

[0060] Fault prediction through machine learning models can issue early warning signals before the actual fault occurs. This provides users or maintenance personnel with enough time to take preventive measures or perform maintenance to avoid the impact of the fault on the battery pack performance and system stability. Since faults can be predicted in advance and corresponding measures can be taken, downtime and maintenance costs caused by sudden failures can be reduced. In addition, preventive maintenance based on prediction results can also extend the service life of the battery pack and improve its reliability. By continuously monitoring and predicting the failure risk of the battery pack, potential problems can be discovered and resolved in a timely manner, thereby improving the reliability and stability of the entire system. Fault prediction results can also provide strong support for the optimization of operating strategies. For example, according to the prediction results, the charging and discharging strategies can be adjusted, the energy management scheme can be optimized, etc., to further improve the system efficiency and economic benefits.

[0061] In some embodiments, S3 further includes: Compare the parameters corresponding to the fault with the normal operating data to obtain the second coefficient; The second coefficient is: Dei2=abs(Dfaei- Dnai) / Dnai Among them, Dei2 is the second coefficient, Dfaei is the mean value of the ith parameter of the eth fault type at the time of the fault; Dnai is the mean value of the ith parameter under normal operation; abs() is the absolute value; Perform correlation analysis on each parameter and obtain a first correlation coefficient; If the second difference or the first correlation coefficient is greater than a preset threshold, the corresponding parameter is used as an input parameter of the machine learning model; Obtaining the weight of each input parameter through the second coefficient and the first correlation coefficient of each input parameter; Among them, the weights of the input parameters are:

[0062] in, The weights of the input parameters for the i-th machine learning model, is the first correlation coefficient of the i-th parameter of the e-th fault type; E is the fault type, I is the total number of input parameters in the machine learning model; y1 and y2 are the second weight coefficients.

[0063] The working principle of the above technical solution is: For each fault type, the absolute difference between the mean value of the corresponding parameter at the time point when the fault occurs (Dfaei) and the mean value of the parameter under normal operation (Dnai) is calculated, and divided by the mean value under normal operation to obtain the second coefficient (Dei2). This coefficient reflects the degree to which the parameter deviates from the normal state when the fault occurs.

[0064] Correlation analysis is performed between the parameters to identify which parameters show stronger correlation when a fault occurs. This correlation is measured by the first correlation coefficient (Rei), which indicates the degree of mutual influence between the parameters when a fault occurs.

[0065] Based on whether the second coefficient or the first correlation coefficient is greater than the preset threshold, it is determined which parameters should be used as input parameters of the machine learning model. This step aims to screen out the parameters that have the greatest impact on fault prediction in order to reduce the complexity of the model and improve the prediction accuracy.

[0066] For the selected input parameters, the weights (Qi) are calculated based on their second coefficients and first correlation coefficients. The weight calculation takes into account the contribution of each parameter under all fault types, and adjusts the relative importance of the second coefficient and the first correlation coefficient in the weight calculation through the second weight coefficient (y1, y2). The allocation of weights allows the model to pay more attention to those parameters that are more critical to fault prediction during prediction.

[0067] Using the selected and weighted input parameters, the machine learning model is trained. During the training process, the model learns the complex relationship between these parameters and the fault types, and tries to extract useful features from the input data to predict future faults.

[0068] Once the model is trained, new input data (i.e., real-time or recent battery pack operating parameters and environmental parameters) can be used to predict failures. The model calculates the probability or risk level of various future failure types based on the current values ​​of the input parameters and their weights.

[0069] The effects of the above technical solution are: By calculating the second coefficient (i.e., the deviation ratio between the parameter value when the fault occurs and the parameter mean value during normal operation), the degree of abnormality of each parameter when the fault occurs can be quantitatively evaluated. At the same time, combined with the first correlation coefficient (i.e., the degree of correlation between each parameter and the fault type), the parameters closely related to the specific fault type can be more accurately screened out. This dual screening mechanism improves the accuracy and reliability of fault identification. Parameters highly related to the fault type are screened out according to the second coefficient and the first correlation coefficient. This can reduce the input dimension of the model, reduce the complexity of the model, and improve the training efficiency and prediction performance of the model. By assigning weights to the input parameters in the machine learning model, the model can pay more attention to those parameters that have a greater impact on the prediction results when predicting. This weighting mechanism helps to improve the accuracy and stability of the model prediction, making the prediction results more in line with the actual situation. Since the screening and weight allocation of the model input parameters are based on the correlation analysis of historical data and fault types, this method can make the model better adapt to different types of faults and different operating conditions. This helps to improve the generalization ability of the model, so that the model can also make accurate predictions when facing new fault types or operating conditions.

[0070] By calculating the weight of each parameter, we can clearly understand which parameters play a key role when a fault occurs. This helps operation and maintenance personnel quickly locate the cause of the fault and take appropriate measures to deal with it. At the same time, it also provides strong support for subsequent fault analysis and diagnosis.

[0071] In some embodiments, S3 further includes: Acquiring multimodal data, wherein the multimodal data includes a vibration signal, a sound signal, and image data of the device; The fault prediction results are obtained through machine learning based on the fault data, operating parameters of the battery pack related to the fault data, environmental parameters, time series data and multimodal data.

[0072] The working principle and effect of the above technical solution are as follows: In this embodiment, the fault prediction system (or S3 system) integrates multimodal data processing and machine learning technology to achieve early prediction of battery pack or other complex equipment failures. Its principle mainly includes the following steps: Vibration data can be obtained through vibration sensors installed on the equipment, which can reflect the mechanical state and potential faults of the equipment, such as imbalance, looseness or wear. A microphone or sound sensor is used to capture the sound generated when the equipment is running. The specific frequency or pattern in the sound may be related to the health status of the equipment. The image data can be used to detect visual signs of failure such as appearance damage, corrosion, and leakage by taking images of the appearance or internal key components of the equipment through a camera.

[0073] The acquired multimodal data (vibration signals, sound signals, image data) are integrated with fault data, operating parameters of the battery pack (such as current, voltage, temperature), environmental parameters (such as temperature, humidity, pressure), and time series data (recording changes in equipment status over time). The data is preprocessed by cleaning, denoising, feature extraction, etc. to extract useful information for fault prediction. The preprocessed data is used to train machine learning models. These models can be supervised learning models (such as random forests, gradient boosting trees, neural networks, etc.), or semi-supervised or unsupervised learning models, depending on the availability of data and the nature of the problem. The model automatically discovers patterns and regularities in the data by learning the association between historical fault data and other related data, so that it can predict possible future faults. When new multimodal data, operating parameters, environmental parameters, and time series data are input, the trained machine learning model will generate fault prediction results based on these data. The prediction results may include the type of fault, the probability of occurrence, the expected time of occurrence, etc., providing decision support for equipment maintenance and maintenance.

[0074] The fusion of multimodal data utilizes information from different sources that may be complementary, thereby improving the accuracy and reliability of fault prediction.

[0075] In some embodiments, the S4 includes: issuing a first alarm for the first abnormality; issuing a second warning for the second abnormality; issuing a third warning for the third abnormality; According to the forecast results, the fourth warning is issued; If the temperature is abnormal, adjust the heat dissipation parameters in the heat dissipation device.

[0076] The working principle and effect of the above technical solution are: for those abnormalities that directly affect the safe operation of the equipment and may cause immediate shutdown or serious damage (such as short circuit, overcurrent, etc.), the system will immediately trigger the first alarm, so that the operator can respond quickly and take measures.

[0077] For abnormalities that may indicate future failures but do not directly affect equipment operation (such as performance degradation, minor abnormalities, etc.), the system will issue a second or third warning. The purpose is to remind operators to pay attention to the health of the equipment and consider preventive maintenance.

[0078] Based on the prediction results of machine learning or other predictive models, the system will issue a fourth warning. This warning is usually used to predict possible future failures and notify operators in advance so that they can prepare.

[0079] If the system detects an abnormal temperature (such as overheating of the device), the system will automatically adjust the cooling parameters in the cooling device. This may include increasing the fan speed, adjusting the flow or temperature of the coolant, etc. to ensure that the device remains within a safe operating temperature range.

[0080] Through immediate alarm and timely warning, the S4 system can help operators quickly identify and respond to potential safety hazards, thereby reducing the risk of equipment failure and accidents.

[0081] The present invention further provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of any one of the methods of the present invention when executing the computer program.

[0082] The present invention is explained from the viewpoints of purpose of use, efficiency, progress and novelty, and has complied with the functional enhancement and use requirements emphasized by the Patent Law. The above description and drawings of the present invention are only preferred embodiments of the present invention, and are not intended to limit the present invention. Therefore, all structures, devices, features, etc. that are similar or identical to the present invention, that is, all equivalent replacements or modifications made according to the scope of the patent application of the present invention, should fall within the scope of protection of the patent application of the present invention.

Claims

1. A real-time monitoring method for abnormal operation status of an energy storage power generation system, characterized in that: The method comprises: S1. Collecting the operating parameters of the battery pack through the first sensor group; acquiring the environmental parameters of the battery pack through the second sensor group; S2. Upload the key parameters and environmental parameters to the platform; the cloud platform performs a first determination on the operating parameters through the control model of each parameter to obtain an abnormality mark; the control model includes a first control range, a second control range and a third control range; the abnormality mark includes a first abnormality, a second abnormality and a third abnormality; S3, predicting the fault through a fault prediction model on the cloud platform according to the operating parameters to obtain a fault prediction result; S4. Based on the abnormal markings and / or fault prediction results, an early warning is issued and suggestions for adjusting the energy storage system are given. The adjustment suggestions include adjusting the battery pack operating parameters and heat dissipation parameters.

2. The method according to claim 1, characterized in that The S1 includes: Collecting operating parameters of the battery pack during charging or discharging by means of a first sensor group installed in the battery pack, wherein the operating parameters include temperature, energy consumption, internal resistance, voltage, current, and charging or discharging rate; The environmental parameters of the battery pack are acquired through the second sensor group; the environmental parameters include environmental temperature and environmental humidity.

3. The method according to claim 1, characterized in that The S2 includes: Through design specifications and historical data, control models for each operating parameter are established; The upper and lower limits of each key parameter are obtained through design specifications as the first control range.

4. The method according to claim 3, characterized in that The second control range includes a second control range for a charging process and a second control range for a discharging process; The method for obtaining the second control range of the charging process is as follows: Obtaining a curve of a parameter changing with charging time during a single charging process of the battery pack under historical normal operation as a first curve; Establishing a first model of the parameter by using a plurality of first curves, environmental parameters corresponding to the first curves, and the number of charging times; According to the current charging time, environmental parameters, load and charging times, based on the first model of the parameter, a range of the parameter under the charging time is obtained as a second control range of the charging process of the parameter; The method for obtaining the second control range of the discharge process is as follows: Obtaining a curve of a parameter changing with discharge time during a single discharge of the battery pack under historical normal operation as the second curve; Establishing a second prediction model of the parameter through a plurality of second curves, environmental parameters corresponding to the second curves, loads, and discharge times; Through the current discharge duration, environmental parameters, load and number of charges, based on the second model of the parameter, a range of the parameter under the discharge duration is obtained as the second range of the discharge process of the parameter.

5. The method according to claim 3, characterized in that: The third control range includes: The difference between two adjacent samples of each parameter during the charging process or the discharging process is obtained through historical data as a first difference; and a third control range is obtained through multiple first differences; The third control range is: DeltaCmini=DeltaCai- (k+max(Fc / Fmax,T c / Tmax))*σidelta DeltaCmaxi=DeltaCai+( k+ max(Fc / Fmax,T c / Tmax))*σidelta Where DeltaCmini is the minimum value of the third control range of the i-th parameter, DeltaCmaxi is the maximum value of the third control range of the i-th parameter, DeltaCai is the mean of multiple first differences of the i-th parameter; σidelta is the standard deviation of multiple first differences of the i-th parameter, Fc is the current load of the system; Fmax is the maximum load of the system, Tc is the current temperature of the battery pack; Tmax is the maximum allowable temperature of the battery pack, k is the adjustment coefficient, 1 <k≤2。 6. The method according to claim 3, characterized in that The S2 further includes: If any parameter exceeds the first control range, a first abnormality is marked; If any parameter exceeds the second control range, a second abnormality is marked; If any parameter exceeds the third control range, a third abnormality is flagged.

7. The method according to claim 1, characterized in that The S3 includes: Obtaining fault data of each fault in historical data, wherein the fault data includes type, fault occurrence stage, fault occurrence frequency, and duration; wherein the fault occurrence stage includes a charging stage and a discharging stage; Obtain operating parameters, environmental parameters and time series data of the battery pack related to the fault data; The fault prediction results are obtained through machine learning based on the fault data, the operating parameters of the battery pack related to the fault data, the environmental parameters and the time series data.

8. The method according to claim 7, characterized in that The S3 further includes: Compare the parameters corresponding to the fault with the normal operating data to obtain the second coefficient; The second coefficient is: Dei2=abs(Dfaei- Dnai) / Dnai Among them, Dei2 is the second coefficient, Dfaei is the mean value of the ith parameter of the eth fault type at the time of the fault; Dnai is the mean value of the ith parameter under normal operation; abs() is the absolute value; Perform correlation analysis on each parameter and obtain a first correlation coefficient; If the second difference or the first correlation coefficient is greater than a preset threshold, the corresponding parameter is used as an input parameter of the machine learning model; Obtaining the weight of each input parameter through the second coefficient and the first correlation coefficient of each input parameter; Among them, the weights of the input parameters are: in, The weights of the input parameters for the i-th machine learning model, is the first correlation coefficient of the i-th parameter of the e-th fault type; E is the fault type, I is the total number of input parameters in the machine learning model; y1 and y2 are the second weight coefficients.

9. The method according to claim 1, characterized in that: The S4 includes: issuing a first alarm for the first abnormality; issuing a second warning for the second abnormality; Producing a third warning for the third abnormality; According to the forecast results, the fourth warning is issued; If the temperature is abnormal, adjust the heat dissipation parameters in the heat dissipation device.

10. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the steps of the method according to any one of claims 1 to 9 when executing the computer program.

Citation Information

Cited By

  • Abnormal state detection method of liquid-cooled energy storage system

    CN119560696A

  • Abnormal State Detection Method for Liquid-Cooled Energy Storage System

    CN119560696B

  • Lithium battery circuit charging and discharging protection method and charging and discharging protection system

    CN119675186A

  • Heating ventilation air conditioner monitoring system based on cloud platform

    CN119802805A

  • Fault monitoring and diagnosing system for energy storage battery pack based on artificial intelligence

    CN119881650A