UPS-based utility power failure prediction system

Abstract

The application relates to the technical field of fault prediction, and particularly discloses a power failure prediction system based on UPS, which comprises a fault collection module, a monitoring analysis module, a model construction module and a model determination module; the fault collection module is used for quantifying the accurate state of fault judgment, obtaining a monitoring inaccuracy value and judging the monitoring accuracy of the UPS device; the monitoring analysis module is used for acquiring a load peak working period and a UPS inaccuracy judgment period when the power failure is not accurate, and calculating the period similarity; the model construction module is used for extracting power parameters to construct a training set when the period similarity is obtained, and constructing a power failure prediction model by using linear regression, a decision tree, a neural network and a long short-term memory network model algorithm; and the model determination module is used for monitoring inaccuracy value analysis of different models, obtaining a model adaptation value and determining an optimal algorithm; the application can improve the accuracy of power failure prediction and provides guarantee for stable operation of the UPS device.

Description

Technical Field

[0001] This invention relates to the field of fault prediction technology, and more specifically to a mains power fault prediction system based on a UPS. Background Technology

[0002] UPS (Uninterruptible Power Supply) is a power supply device that provides a stable power supply to load equipment when the mains power is interrupted or abnormal. However, the occurrence of mains power failure brings many challenges to the stable operation of UPS.

[0003] A Chinese patent application with publication number CN115859058B discloses a UPS fault prediction method and system based on a wide learning network, including: preprocessing historical data of UPS power supply batteries under different states; determining the value range of parameters such as feature nodes, enhancement nodes, and feature window number in the wide learning network; introducing a simulated annealing algorithm to improve the mutation operation in the genetic algorithm, resulting in an improved genetic algorithm; optimizing the wide learning parameters using the improved genetic algorithm; determining the threshold range of normal UPS power supply batteries based on a grid search algorithm and a Ksigma model; and comprehensively applying GA+BLS+K-sigma to obtain the final prediction and diagnostic model, and monitoring the status of UPS power supply batteries in real time.

[0004] The above technology uses a wide learning network and determines the threshold range through an improved genetic algorithm and a grid search algorithm. The above technology constructs a general UPS power supply battery status monitoring model. However, it does not fully consider the relationship between mains power failure and UPS inaccuracy judgment cycle and peak load cycle. If it is necessary to determine whether the UPS inaccuracy judgment cycle and peak load cycle are similar, and if they are similar, different model algorithms can be used to construct the model, which is beneficial for predicting peak loads when mains power failure occurs.

[0005] The aforementioned technology is currently used for real-time monitoring of UPS power supply battery status, using diagnostic models to determine battery status. However, it does not predict mains power failures, nor does it establish a monitoring and evaluation system. Calculating the false alarm rate and false miss rate to obtain the monitoring inaccuracy value would be beneficial for evaluating the accuracy of UPS equipment in monitoring mains power failures. Currently, the technology lacks a quantitative evaluation of monitoring accuracy and a comprehensive evaluation system for model performance. Calculating the effective cycle ratio and inaccuracy improvement ratio to obtain the model fitness value and determine the optimal algorithm would facilitate timely identification of problems in the monitoring process and optimization of the model.

[0006] Therefore, the present invention provides a mains power failure prediction system based on UPS. Summary of the Invention

[0007] The purpose of this invention is to provide a UPS-based mains power failure prediction system to solve the problems mentioned above.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] The UPS-based mains power failure prediction system includes the following modules:

[0010] Fault acquisition module: used to acquire the accurate status of the mains power fault judgment of the UPS equipment, perform quantitative analysis, obtain the monitoring inaccuracy value, and determine whether the UPS equipment is accurate in monitoring mains power faults;

[0011] Monitoring and Analysis Module: If the UPS equipment's monitoring of mains power failures is inaccurate, it is used to obtain the peak operating cycle of the load equipment, construct a peak cycle group, obtain the UPS equipment's inaccuracy judgment cycle, construct an inaccuracy cycle group, and determine whether the UPS equipment's inaccuracy judgment cycle is similar to the load's peak operating cycle.

[0012] Model building module: If similar, obtain the mains parameters of the UPS equipment that mislabels mains power faults, and use different model algorithms to build mains power fault prediction models;

[0013] Model determination module: Construct mains power fault prediction models based on different model algorithms, perform numerical analysis on the monitoring inaccuracies output by different model algorithms to obtain model fitness values, and determine the optimal algorithm for mains power fault prediction models.

[0014] As a further technical solution of the present invention: the method for obtaining the monitoring misalignment value is as follows:

[0015] Obtain the mains power fault status monitored by the UPS equipment from the UPS equipment log;

[0016] Within the monitoring period, the number of times the UPS device marked the mains power as faulty is obtained, thus obtaining the fault marking count;

[0017] The number of times a fault was falsely marked is obtained from the number of times a UPS fault was marked. The ratio of the number of times a fault was falsely marked to the number of times a fault was marked is calculated to obtain the false alarm rate.

[0018] The number of times the UPS failed to recognize the mains power failure is obtained, thus determining the number of missed failures.

[0019] The fault false alarm rate is obtained by comparing the number of missed faults with the number of actual faults obtained in advance.

[0020] The monitoring inaccuracy value is obtained by weighted summation based on the false alarm rate and the false missed rate.

[0021] If the monitoring inaccuracy value of the UPS equipment is lower than the preset monitoring accuracy threshold, it indicates that the UPS equipment is not accurately monitoring mains power failures.

[0022] As a further technical solution of the present invention: the method for determining whether the failure judgment period of the UPS equipment is similar to the peak working period of the load is as follows:

[0023] The monitoring period in which the UPS equipment monitoring inaccuracy value is lower than the preset monitoring accuracy threshold is obtained to determine the inaccuracy judgment period;

[0024] Obtain the inaccuracy determination periods from all monitoring periods and construct an inaccuracy period group;

[0025] Obtain the peak operating cycles of the load devices throughout all monitoring periods and construct peak cycle groups;

[0026] Based on the inaccurate periodic group and the peak periodic group, the periodic similarity is obtained using the periodic matching window group and entropy calculation method.

[0027] If the period similarity is higher than the preset similarity threshold, then the inaccurate period group and the peak period group are considered to be similar in the time dimension.

[0028] As a further technical solution of the present invention: the method for obtaining the periodic similarity is as follows:

[0029] The monitoring period is divided into multiple period matching windows by using the period matching window group cutting method, and a period matching window group is constructed.

[0030] Based on the misalignment period group, obtain the matching window group for each period, which includes the number of misalignment judgment periods;

[0031] Based on the peak period group, obtain the matching window group for each period, which includes the number of peak working periods;

[0032] The frequency formula is used to obtain the matching window group for each cycle, including the frequency of the peak working cycle and the frequency of the inaccuracy judgment cycle.

[0033] Numerical calculations were performed on the frequency of peak working cycles and the frequency of inaccuracy judgment cycles to obtain cycle similarity.

[0034] As a further technical solution of the present invention: the frequency of the peak working cycle and the frequency of the inaccuracy judgment cycle are calculated numerically as follows:

[0035] Based on the frequency of peak working cycles and the frequency of inaccuracy judgment cycles, the periodic entropy of the inaccuracy cycle group and the periodic entropy of the peak cycle group are obtained through the entropy formula.

[0036] Periodic similarity is obtained by using a ratio formula based on the periodic entropy of the inaccurate periodic group and the periodic entropy of the peak periodic group.

[0037] As a further technical solution of the present invention: the mains power fault prediction model is constructed as follows:

[0038] Obtain the mains parameters of the UPS equipment that mislabel mains power faults, and construct a training set for the mains power fault prediction model;

[0039] Based on the training set of the mains power failure prediction model, a mains power failure prediction model is constructed using linear regression model algorithm, decision tree model algorithm, neural network model algorithm, and long short-term memory network model algorithm.

[0040] As a further technical solution of the present invention: the method for obtaining the mains power parameters of the mains power fault mislabeling is as follows:

[0041] If the failure judgment period of the UPS equipment is similar to the peak working period of the load, the mains power parameters within the failure judgment period of the UPS equipment can be extracted from the UPS working log.

[0042] The mains power parameters include: voltage, frequency, and phase;

[0043] Based on the mains power parameters, the voltage fluctuation range, mains frequency deviation, and voltage harmonic content are extracted within the inaccuracy judgment period.

[0044] As a further technical solution of the present invention: the method for determining the preferred algorithm of the mains power fault prediction model is as follows:

[0045] Different model algorithms are used to construct a mains power fault prediction model, and the model fitness value is output.

[0046] The model fitness values ​​of different model algorithms are sorted in descending order, and the model algorithm with the largest fitness value is selected as the preferred algorithm for the mains power fault prediction model.

[0047] As a further technical solution of the present invention: the method for obtaining the model fitness value is as follows:

[0048] Based on different model algorithms, a mains power failure prediction model is constructed to obtain the monitoring inaccuracy value of UPS equipment equipped with the mains power failure prediction model during the peak working period, as well as the monitoring inaccuracy value of UPS equipment without the mains power failure prediction model during the peak working period.

[0049] The monitoring inaccuracy value of UPS equipment equipped with a mains power failure prediction model during peak working hours and the monitoring inaccuracy value of UPS equipment without a mains power failure prediction model during peak working hours are compared to obtain the monitoring inaccuracy difference.

[0050] Obtain the monitoring misalignment of the UPS equipment during peak operating cycles and determine the positive or negative nature of the monitoring misalignment;

[0051] If the monitoring inaccuracy is negative, then the peak working period is marked as an effective optimization period;

[0052] The effective optimization cycle is quantitatively analyzed to obtain the effective cycle ratio;

[0053] The ratio of monitoring inaccuracy difference within the effective optimization period is calculated to obtain the inaccuracy improvement ratio.

[0054] The effective period ratio and the inaccuracy improvement ratio are weighted and summed to obtain the model fitness value.

[0055] As a further technical solution of the present invention: the effective cycle ratio is obtained as follows:

[0056] Within multiple peak work cycles, obtain the number of effective optimization cycles and the number of peak work cycles, and then calculate the ratio of the number of effective optimization cycles to the number of peak work cycles to obtain the effective cycle ratio.

[0057] The method for obtaining the misalignment improvement ratio is as follows:

[0058] The monitoring misalignment error within the effective optimization period is obtained. The absolute value of the monitoring misalignment error is processed and compared with the monitoring misalignment value of the UPS equipment without a mains power failure prediction model during the peak working period to obtain the misalignment improvement ratio.

[0059] The beneficial effects of this invention are:

[0060] (1) The fault acquisition module performs quantitative analysis on the mains fault judgment of UPS equipment, obtains the monitoring inaccuracy value, and determines whether the UPS is accurate in monitoring the mains fault. If the monitoring is inaccurate, the monitoring analysis module can construct a load peak cycle group and a UPS inaccuracy cycle group, calculate the cycle similarity, not only can it identify the high-incidence fault cycle, but also focus on monitoring and prediction during these periods, which improves the pertinence of mains fault prediction, allowing relevant personnel to take more targeted preventive measures, which is conducive to reducing equipment abnormalities caused by mains faults;

[0061] (2) When the UPS inaccuracy judgment cycle is similar to the peak load working cycle, the model building module extracts the mains power parameters to build a training set and uses a variety of model algorithms to build a mains power failure prediction model. The model determination module analyzes the monitoring inaccuracy values ​​of different models during the peak working cycle, calculates the model fitness value, selects the optimal algorithm, and uses the most suitable model to predict mains power failure, which improves the reliability of UPS in predicting mains power failure, ensures that UPS equipment can operate stably when there is a mains power failure, and helps to reduce the risk caused by prediction errors. Attached Figure Description

[0062] The invention will now be further described with reference to the accompanying drawings.

[0063] Figure 1 This is a block diagram of the UPS-based mains power failure prediction system of the present invention;

[0064] Figure 2 This is a flowchart illustrating the method for obtaining model fitness values ​​in the UPS-based mains power fault prediction system of this invention. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Example 1

[0067] Please see Figure 1 As shown, this invention is a UPS-based mains power failure prediction system, comprising the following modules:

[0068] Fault Acquisition Module: Acquires the number of mains power faults detected by the UPS equipment, performs quantitative analysis on the accuracy of fault detection, obtains the monitoring inaccuracy value, and determines whether the UPS equipment is accurate in monitoring mains power faults.

[0069] In some embodiments, the mains power fault status monitored by the UPS equipment is obtained from the UPS equipment log;

[0070] It should be noted that the UPS equipment can detect whether the mains power is normal in real time. When the mains power fails, such as power outage, abnormal voltage, abnormal frequency, etc., the UPS will detect it in time and record the mains power failure in the equipment log.

[0071] Within the monitoring period, the number of times the UPS device marked the mains power as faulty is obtained, thus obtaining the fault marking count;

[0072] The number of times a fault was falsely marked is obtained from the number of times a UPS fault was marked. The ratio of the number of times a fault was falsely marked to the number of times a fault was marked is calculated to obtain the false alarm rate.

[0073] The number of times the UPS failed to recognize the mains power failure is obtained, thus determining the number of missed failures.

[0074] The fault false alarm rate is obtained by comparing the number of missed faults with the number of actual faults obtained in advance.

[0075] It should be noted that the actual number of UPS failures is obtained by summing the number of UPS failures correctly marked and the number of failures missed, while the number of UPS failures correctly marked is obtained by subtracting the number of failures marked from the number of failures that were incorrectly marked.

[0076] The monitoring inaccuracy value is obtained by weighted summation based on the false alarm rate and the false missed rate.

[0077] It should be noted that the number of times a fault was incorrectly marked and the number of times a fault was missed in the UPS fault marking count were obtained by technical personnel in the field through analysis of the UPS equipment's operating logs.

[0078] The inaccuracy value of the UPS equipment is compared with the preset accuracy threshold to determine whether the UPS equipment is accurate in monitoring mains power failures.

[0079] If the monitoring inaccuracy value of the UPS equipment is lower than the preset monitoring accuracy threshold, it indicates that the UPS equipment is not accurately monitoring mains power failures.

[0080] If the UPS device's monitoring inaccuracy value is higher than the preset monitoring accuracy threshold, it indicates that the UPS device's monitoring of mains power failures is relatively accurate.

[0081] Monitoring and Analysis Module: If the UPS equipment's monitoring of mains power failures is not accurate enough, it is used to obtain the peak operating cycle of the load equipment, construct a peak cycle group, obtain the UPS equipment's inaccuracy judgment cycle, construct an inaccuracy cycle group, and determine whether the UPS equipment's inaccuracy judgment cycle is similar to the load's peak operating cycle.

[0082] If the UPS equipment's monitoring of mains power failures is not accurate enough, the monitoring period in which the UPS equipment's monitoring inaccuracy value is lower than the preset monitoring accuracy threshold is obtained to determine the inaccuracy judgment period.

[0083] Obtain the inaccuracy determination periods from all monitoring periods and construct an inaccuracy period group;

[0084] In some embodiments, the peak operating cycle of the load equipment is obtained from the monitoring data of the power grid monitoring system;

[0085] Obtain the peak operating cycles of the load devices throughout all monitoring periods and construct peak cycle groups;

[0086] It should be noted that by determining the peak working cycle of the load equipment, we can identify the period during which the load equipment has the greatest demand for mains power within the entire monitoring period. During the peak working cycle, the mains power system is under greater pressure and is prone to problems such as voltage fluctuations and frequency anomalies, thereby increasing the probability of mains power failures. By constructing a peak cycle group, it is helpful to identify the high-incidence period of failures and improve the pertinence and accuracy of mains power failure prediction.

[0087] Based on the inaccurate periodic group and the peak periodic group, the periodic similarity is obtained using the periodic matching window group and entropy calculation method.

[0088] The monitoring period is divided into multiple period matching windows using the period matching window group shearing method, and a period matching window group Z is constructed. i ={z1, z2, ..., z n}, where i is the number of the periodic matching window group, and the value of i ranges from [1, n];

[0089] Based on the misalignment period group, obtain the matching window group Z for each period. i The number of misjudgment periods, m 1i ;

[0090] Through the formula: Obtain each periodic matching window group, containing the frequency f of the misalignment determination period. 1i , where n is the total number of periodic matching window groups;

[0091] Based on the peak period group, obtain the matching window group Z for each period. i The number of peak work cycles, m 2i ;

[0092] Through the formula: Get the matching window group Z for each period i The frequency f, which includes peak duty cycles 2i ;

[0093] Through the formula: Obtain the periodic entropy H1 of the misaligned periodic group;

[0094] Through the formula: Obtain the periodic entropy H2 of the peak period group;

[0095] Through the formula: Obtain the periodic similarity Xs;

[0096] It should be noted that when the period similarity Xs=1, |H1-H2|=0, that is, the period entropy H1 of the inaccurate period group and the period entropy H2 of the peak period group are the same, indicating that the period similarity between the inaccurate period group and the peak period group is the highest in the period matching window group.

[0097] The period similarity Xs is analyzed with a preset similarity threshold to determine whether the inaccurate period group and the peak period group are similar in the time dimension.

[0098] If the periodic similarity Xs is higher than the preset similarity threshold, then the inaccurate periodic group and the peak periodic group are considered to be similar in the time dimension; otherwise, they are not similar.

[0099] The technical solution of this embodiment is as follows: The number of mains power failures detected by the UPS equipment is obtained, the accuracy of the failure detection is quantitatively analyzed to obtain the monitoring inaccuracy value, and the accuracy of the UPS equipment's monitoring of mains power failures is determined. If the UPS equipment's monitoring of mains power failures is not accurate enough, the peak working cycle of the load is obtained, a peak cycle group is constructed, the inaccuracy detection cycle of the UPS equipment is obtained, an inaccuracy cycle group is constructed, and it is determined whether the inaccuracy detection cycle of the UPS equipment is similar to the peak working cycle of the load. By calculating the cycle similarity, not only can the high-incidence failure cycle be identified, but also the focus of monitoring and prediction during these periods can be improved, thus enhancing the pertinence of mains power failure prediction and enabling relevant personnel to take more targeted preventive measures, which is beneficial to reducing equipment abnormalities caused by mains power failures.

[0100] Example 2

[0101] like Figure 1 As shown, the UPS-based mains power failure prediction system also includes the following modules:

[0102] Model building module: If the UPS equipment's failure judgment cycle is similar to the peak working cycle of the load, obtain the mains parameters that the UPS equipment mislabels as mains faults, and use different model algorithms to build a mains fault prediction model;

[0103] If the failure judgment period of the UPS equipment is similar to the peak working period of the load, the mains power parameters within the failure judgment period of the UPS equipment can be extracted from the UPS working log.

[0104] The mains power parameters include: voltage, frequency, and phase;

[0105] Based on the mains power parameters, the voltage fluctuation range, mains power frequency deviation, and voltage harmonic content are extracted within the inaccuracy judgment period.

[0106] It should be noted that the UPS equipment has the ability to monitor and record mains power parameter data, including voltage fluctuation range, mains frequency deviation, and voltage harmonic content, which can be extracted from the UPS equipment's operating logs.

[0107] The voltage fluctuation range refers to the difference between the maximum and minimum values ​​of the mains voltage within the inaccuracy judgment period. For example, if the mains voltage reaches a maximum of 230V and a minimum of 210V within a certain inaccuracy judgment period, the voltage fluctuation range is 230-210=20V.

[0108] If the mains voltage waveform is not an ideal sine wave, harmonics will be generated. The voltage harmonic content refers to the percentage of the root mean square value of each harmonic voltage to the root mean square value of the fundamental voltage.

[0109] For example, if the effective value of the fundamental voltage is 220V and the effective value of a certain harmonic voltage is 22V, then the voltage harmonic content of that harmonic is (22÷220)×100%=10%.

[0110] Obtain the voltage fluctuation range, mains frequency deviation, and voltage harmonic content at the time when the UPS equipment mis-marks a mains power fault;

[0111] The voltage fluctuation range, mains frequency deviation, and voltage harmonic content within the mislabeling state and inaccuracy judgment period of the mains fault are used as the training set for the mains fault prediction model.

[0112] Based on the training set of the mains power failure prediction model, a mains power failure prediction model is constructed using linear regression model algorithm, decision tree model algorithm, neural network model algorithm, and long short-term memory network model algorithm.

[0113] It should be noted that the linear regression model algorithm uses voltage fluctuation range, mains frequency deviation, and voltage harmonic content as independent variables and whether a mains fault occurs as the dependent variable. It calculates the linear relationship between the independent and dependent variables and calculates the linear combination between mains parameters and mains faults to predict the probability of a mains fault occurring.

[0114] The decision tree model algorithm constructs a tree structure by recursively partitioning the feature space to make decisions. The decision tree will gradually divide the data into different nodes according to different values ​​of the mains power parameters, and finally determine whether a mains power failure will occur.

[0115] Neural network model algorithms are composed of layers of multiple neurons stacked together. Through the connections and weights between neurons, they learn complex patterns and nonlinear relationships in the data. The neural network can take the mains power parameters as input, and after nonlinear transformation through multiple hidden layers, it finally outputs the probability of mains power failure.

[0116] The Long Short-Term Memory (LSTM) network model algorithm processes time-series data of mains power parameters and combines the correlation between mains power parameters at different time points to predict the occurrence of mains power failures.

[0117] Specifically, the mains power failure prediction model is constructed as follows:

[0118] S1. Mains power data collection and preprocessing;

[0119] The dataset of the mains power fault prediction model is preprocessed. The missing values ​​in the dataset are filled with mean imputation, and outliers in the dataset are removed using the three-times-standard-deviation principle.

[0120] S2. Label and divide the dataset;

[0121] In the dataset of the mains power fault prediction model, the voltage fluctuation range, mains frequency deviation, and voltage harmonic content of the mislabeled mains power fault state are used as labels.

[0122] For example, the label value for voltage fluctuation range, mains frequency deviation, and voltage harmonic content in the case of a mains power fault mislabeling state is 0, and the label value for the other two states is 1.

[0123] The dataset of the mains power fault prediction model was divided into training and testing sets in an 8:2 ratio;

[0124] S3. Use different model algorithms for training;

[0125] The dataset of the divided mains power fault prediction model is input into the linear regression model algorithm, decision tree model algorithm, neural network model algorithm, and long short-term memory network model algorithm for training.

[0126] Model determination module: Construct mains power fault prediction models based on different model algorithms, perform numerical analysis on the monitoring inaccuracies output by different model algorithms to obtain model fitness values, and determine the optimal algorithm for mains power fault prediction models;

[0127] Based on different model algorithms, a mains power failure prediction model is constructed to obtain the monitoring inaccuracy value of UPS equipment equipped with the mains power failure prediction model during the peak working period, as well as the monitoring inaccuracy value of UPS equipment without the mains power failure prediction model during the peak working period.

[0128] like Figure 2 As shown, numerical analysis is performed on the monitoring inaccuracies to obtain the model adaptation values;

[0129] Specifically, the model fitness values ​​are obtained as follows:

[0130] The monitoring inaccuracy value of UPS equipment equipped with mains power failure prediction model during peak working period and the monitoring inaccuracy value of UPS equipment without mains power failure prediction model during peak working period are obtained by difference processing to obtain the monitoring inaccuracy difference.

[0131] It should be noted that the larger the absolute value of the monitoring misalignment, the more significant the impact of the model on the monitoring accuracy. A large absolute value of negative misalignment indicates that the model is effective in reducing monitoring misalignment; while a large absolute value of positive misalignment warns of serious problems with the model, which may lead to even more inaccurate monitoring results.

[0132] Obtain the monitoring misalignment of the UPS equipment during peak operating cycles and determine the positive or negative nature of the monitoring misalignment;

[0133] If the monitoring inaccuracy is negative, the peak working period is marked as an effective optimization period; otherwise, it is marked as an invalid optimization period.

[0134] Within multiple peak work cycles, obtain the number of effective optimization cycles and the number of peak work cycles, and then calculate the ratio of the number of effective optimization cycles to the number of peak work cycles to obtain the effective cycle ratio.

[0135] The monitoring misalignment error within the effective optimization period is obtained. The absolute value of the monitoring misalignment error is processed and compared with the monitoring misalignment value of the UPS equipment without the mains power failure prediction model during the peak working period to obtain the misalignment improvement ratio.

[0136] The effective period ratio and the inaccuracy improvement ratio are weighted and summed to obtain the model fitness value;

[0137] Different model algorithms are used to construct a mains power fault prediction model. The model fitness values ​​of the output model algorithms are sorted in descending order. The model algorithm with the largest model fitness value is selected as the preferred algorithm for the mains power fault prediction model.

[0138] It should be noted that the effective period ratio reflects the stability and reliability of the model in practical applications. The higher the effective period ratio, the more frequently the model plays an active role in peak working periods, and the greater its continuous contribution to improving monitoring accuracy.

[0139] The accuracy improvement ratio measures the degree to which different model algorithms can improve the monitoring accuracy within an effective optimization period. It reflects the model's ability to improve monitoring accuracy. The larger the accuracy improvement ratio, the more significant the effect of the model on reducing the monitoring accuracy within an effective optimization period.

[0140] The model fitness value reflects the overall performance of the model in terms of monitoring accuracy. The higher the model fitness value, the higher the value of the model algorithm in UPS prediction of mains power failures, and the more likely it is to become the preferred algorithm. Conversely, the lower the model fitness value, the worse the overall performance of the model, and the more necessary it is to optimize the model algorithm or select another model.

[0141] The technical solution of this embodiment is as follows: If the UPS equipment's inaccuracy judgment cycle is similar to the peak working cycle of the load, the mains parameters of the UPS equipment that mis-marked the mains power failure are obtained, and a mains power failure prediction model is constructed using different model algorithms. Based on the different model algorithms, the monitoring inaccuracy values ​​output by different model algorithms are numerically analyzed to obtain the model fitness value. The optimal algorithm for the mains power failure prediction model is determined, the model fitness value is calculated, and the optimal algorithm is selected. By using the most suitable model for mains power failure prediction, the reliability of the UPS in predicting mains power failure is improved, ensuring that the UPS equipment can operate stably during mains power failures, which helps to reduce the risks caused by prediction errors.

[0142] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A UPS-based mains power failure prediction system, characterized in that: Includes the following modules: Fault acquisition module: used to acquire the accurate status of the mains power fault judgment of the UPS equipment, perform quantitative analysis, obtain the monitoring inaccuracy value, and determine whether the UPS equipment is accurate in monitoring mains power faults; Monitoring and Analysis Module: If the UPS equipment's monitoring of mains power failures is inaccurate, it is used to obtain the peak operating cycle of the load equipment, construct a peak cycle group, obtain the UPS equipment's inaccuracy judgment cycle, construct an inaccuracy cycle group, and determine whether the UPS equipment's inaccuracy judgment cycle is similar to the load equipment's peak operating cycle. Model building module: If similar, it is used to obtain the mains parameters that cause the UPS equipment to mislabel mains faults, and to build mains fault prediction models using different model algorithms; Model determination module: Based on the mains power fault prediction models constructed with different model algorithms, numerical analysis is performed on the monitoring inaccuracies output by different model algorithms to obtain the model fitness value and determine the optimal algorithm for the mains power fault prediction model; The method for obtaining the monitoring inaccuracy value is as follows: Obtain the mains power fault status monitored by the UPS equipment from the UPS equipment log; Within the monitoring period, the number of times the UPS device marked the mains power as faulty is obtained, thus obtaining the fault marking count; The number of times a fault was falsely marked is obtained from the number of times a UPS fault was marked. The ratio of the number of times a fault was falsely marked to the number of times a fault was marked is calculated to obtain the false alarm rate. The number of times the UPS failed to recognize the mains power failure is obtained, thus determining the number of missed failures. The fault false alarm rate is obtained by comparing the number of missed faults with the number of actual faults obtained in advance. The monitoring inaccuracy value is obtained by weighted summation based on the false alarm rate and the false missed rate. If the monitoring inaccuracy value of the UPS equipment is lower than the preset monitoring accuracy threshold, it indicates that the UPS equipment is not accurately monitoring mains power failures. The mains power fault prediction model is constructed as follows: Obtain the mains parameters of the UPS equipment that mislabel mains power faults, and construct a training set for the mains power fault prediction model; Based on the training set of the mains power failure prediction model, a mains power failure prediction model is constructed using linear regression model algorithm, decision tree model algorithm, neural network model algorithm, and long short-term memory network model algorithm. The optimal algorithm for the mains power fault prediction model is determined as follows: Different model algorithms are used to construct a mains power fault prediction model, and the model fitness value is output. The model fitness values ​​of different model algorithms are sorted in descending order, and the model algorithm with the largest model fitness value is determined as the preferred algorithm for the mains power fault prediction model. The monitoring period in which the UPS equipment monitoring inaccuracy value is lower than the preset monitoring accuracy threshold is obtained to obtain the inaccuracy judgment period; the inaccuracy judgment period in all monitoring periods is obtained to construct the inaccuracy period group.

2. The UPS-based mains power failure prediction system according to claim 1, characterized in that: The method for determining whether the UPS equipment's failure detection period is similar to the peak load cycle is as follows: Obtain the peak operating cycles of the load devices within all monitoring periods and construct peak cycle groups; Based on the inaccurate periodic group and the peak periodic group, the periodic similarity is obtained using the periodic matching window group and periodic entropy calculation method. If the period similarity is higher than the preset similarity threshold, then the inaccurate period group and the peak period group are considered to be similar in the time dimension.

3. The UPS-based mains power failure prediction system according to claim 2, characterized in that: The method for obtaining the periodic similarity is as follows: The monitoring period is divided into multiple period matching windows by using the period matching window group cutting method, and a period matching window group is constructed. Based on the misalignment period group, obtain the matching window group for each period, which includes the number of misalignment judgment periods; Based on the peak period group, obtain the matching window group for each period, which includes the number of peak working periods; The frequency formula is used to obtain the matching window group for each cycle, including the frequency of the peak working cycle and the frequency of the inaccuracy judgment cycle. Numerical calculations were performed on the frequency of peak working cycles and the frequency of inaccuracy judgment cycles to obtain cycle similarity.

4. The UPS-based mains power failure prediction system according to claim 3, characterized in that: The frequency of the peak working cycle and the frequency of the inaccuracy judgment cycle are calculated numerically as follows: Based on the frequency of peak working cycles and the frequency of inaccuracy judgment cycles, the periodic entropy of the inaccuracy cycle group and the periodic entropy of the peak cycle group are obtained through the entropy formula. Periodic similarity is obtained by using a ratio formula based on the periodic entropy of the inaccurate periodic group and the periodic entropy of the peak periodic group.

5. The UPS-based mains power failure prediction system according to claim 4, characterized in that: The method for obtaining the mains power parameters for the mislabeled mains power fault is as follows: If the failure judgment period of the UPS equipment is similar to the peak working period of the load, the mains power parameters within the failure judgment period of the UPS equipment can be extracted from the UPS working log. The mains power parameters include: voltage, frequency, and phase; Based on the mains power parameters, the voltage fluctuation range, mains frequency deviation, and voltage harmonic content are extracted within the inaccuracy judgment period.

6. The UPS-based mains power failure prediction system according to claim 5, characterized in that: The model fitness value is obtained in the following way: Based on different model algorithms, a mains power failure prediction model is constructed to obtain the monitoring inaccuracy value of UPS equipment equipped with the mains power failure prediction model during the peak working period, as well as the monitoring inaccuracy value of UPS equipment without the mains power failure prediction model during the peak working period. The monitoring inaccuracy value of UPS equipment equipped with a mains power failure prediction model during peak working hours and the monitoring inaccuracy value of UPS equipment without a mains power failure prediction model during peak working hours are compared to obtain the monitoring inaccuracy difference. Obtain the monitoring misalignment of the UPS equipment during peak operating cycles and determine the positive or negative nature of the monitoring misalignment; If the monitoring inaccuracy is negative, then the peak working period is marked as an effective optimization period; The effective optimization cycle is quantitatively analyzed to obtain the effective cycle ratio; The ratio of monitoring inaccuracy difference within the effective optimization period is calculated to obtain the inaccuracy improvement ratio. The effective period ratio and the inaccuracy improvement ratio are weighted and summed to obtain the model fitness value.

7. The UPS-based mains power failure prediction system according to claim 6, characterized in that: The effective period ratio is obtained as follows: Within multiple peak work cycles, obtain the number of effective optimization cycles and the number of peak work cycles, and then calculate the ratio of the number of effective optimization cycles to the number of peak work cycles to obtain the effective cycle ratio. The method for obtaining the misalignment improvement ratio is as follows: The monitoring misalignment error within the effective optimization period is obtained. The absolute value of the monitoring misalignment error is processed and compared with the monitoring misalignment value of the UPS equipment without a mains power failure prediction model during the peak working period to obtain the misalignment improvement ratio.

Images