Visual device battery state monitoring and emergency power supply adding prediction method and system

Abstract

The invention discloses a visual device battery state monitoring and emergency power supply installation prediction method and system, and the method comprises the steps: S1, obtaining the data of a power transmission line channel visual device due to the fault of a battery, and building a prediction basic data set according to the fault type of the battery of the visual device; s2, constructing a power transmission line channel visualization device battery state monitoring and prediction characteristic quantity based on the prediction basic data set, and preprocessing the characteristic quantity to form a sample set of visualization device battery operation environment and internal physical working condition data; and S3, respectively inputting the sample set into a pre-constructed long-short-term memory network model and a full convolutional neural network model for prediction to obtain a first prediction value and a second prediction value, and carrying out weighted fusion on the first prediction value and the second prediction value to obtain a final fault type prediction result. The method has the advantages of being accurate in prediction, reducing the battery maintenance cost and the like.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of power transmission lines, and in particular to a method and system for visualizing device battery status monitoring and emergency power supply installation prediction. Background Art

[0002] At present, online monitoring of the transmission line channel environment usually uses more mature gun cameras or ball cameras as channel visualization devices, and grasps the on-site conditions of the transmission line channel environment by transmitting pictures or real-time videos through the network to ensure the safe and stable operation of the transmission line. Since the transmission towers are basically located in mountainous areas, their channel visualization online devices usually use the "solar energy + battery" mode to achieve power supply. Therefore, under good weather conditions, this power supply mode basically meets the functional requirements of online visualization devices, picture transmission, and snapshots. However, due to the fact that some transmission line towers in some areas are located in complex "two micro" climate areas such as high altitudes, canyons, and heavy ice, there is a high probability of continuous rainy and humid weather, and the duration is long, resulting in the inability of the visualization device power supply system to effectively and normally supply power, which directly affects the normal operation of the visualization device and brings great safety hazards to the control of the transmission line channel environment.

[0003] With the continuous development of battery capacity and charging and discharging strategy control technology, it provides strong guiding significance for battery status detection and judgment of transmission line channel visualization devices. At present, traditional battery capacity and monitoring are concentrated on the battery production stage. The battery capacity level is predicted by a variety of status data and battery capacity during the battery production stage; the second is to predict the remaining life of the battery. By obtaining the original charging and discharging data, multiple charging and discharging status data are used as training samples to predict the battery life. Both of the above two types of battery status monitoring and judgment do not consider the impact of the actual battery operating environment on its capacity, and the single charging and discharging status data only considers the number of charging and discharging times, and comprehensively considers the changes in battery internal resistance, output voltage and current, and operating time.

[0004] At the same time, the battery of the transmission line channel visualization device determines whether the front camera and network communication device can operate normally and effectively. When the visualization device is found to be offline during the centralized management of the channel visualization device, it needs to be maintained in time. According to maintenance data statistics, more than 90% of the failures are battery failures, that is, due to continuous rainy and bad weather, battery swelling or failure, etc., the equipment cannot be effectively powered. The traditional maintenance treatment method usually adopts a rough battery replacement or equipment replacement, but there are the following problems:

[0005] First, directly replace the battery. Limited by the available space of the device itself, the capacity of the replacement battery is affected by the installation space, and the battery capacity is limited. At the same time, some of the actual batteries are still in a relatively good condition and can continue to work normally after the weather conditions improve. Therefore, this method is very likely to cause an increase in battery replacement and maintenance costs;

[0006] The second is to directly replace the equipment. The new equipment has to go through procurement, grid connection, and debugging, with many processing steps and a long time. The maintenance cost increases significantly. Moreover, after continuous rainy and bad weather, the equipment still cannot meet the requirements of online monitoring of the transmission line channel environment.

[0007] Third, it did not take into account factors such as multiple battery manufacturers, fault types, and the particularities of different tower operating environments. It did not combine the battery status monitoring data of the visualization device and the weather forecast in advance, and carry out "preemptive" maintenance before the visualization equipment was offline, so as to achieve an orderly online transition of the visualization device in severe weather such as continuous rain, heavy fog, and icing. Summary of the invention

[0008] In view of the technical problems existing in the prior art, the present invention provides a method and system for predicting battery status and emergency power supply installation of a visualization device with accurate prediction and reduced battery maintenance costs.

[0009] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0010] A method for visualizing device battery status monitoring and emergency power supply installation prediction, comprising the steps of:

[0011] S1. Obtain data on battery failures in a transmission line channel visualization device, and establish a prediction basic data set based on the type of battery failure in the visualization device;

[0012] S2. Constructing the battery status monitoring and prediction feature quantity of the transmission line channel visualization device based on the prediction basic data set, and preprocessing the feature quantity to form a sample set of the battery operating environment and internal physical working condition data of the visualization device;

[0013] S3. Input the sample set into the pre-built long short-term memory network model and the fully convolutional neural network model for prediction, obtain the first prediction value and the second prediction value, and then perform weighted fusion on the first prediction value and the second prediction value to obtain the final fault type prediction result.

[0014] Preferably, it also includes step S4: on the basis of analyzing the longitudinal prediction of a single device in the time series in steps S2 and S3, constructing the correlation relationship between the fault types of different devices in the horizontal direction, and on the basis of the status monitoring and prediction of a single transmission line channel visualization device, combining the tower coordinates with the five categories of two-micro-region maps, clustering analysis is performed on the correlation of different fault types of multiple visualization devices to form a mapping between the two-micro-region maps and the equipment fault types; the five categories include heavy ice areas, dancing areas, pollution areas, minefields and high altitude areas; the two micros are micro-topography and micro-climate.

[0015] Preferably, after step S4, the method further comprises step S5:

[0016] Calculate the Jaccard distance D based on the fault type and the One-Hot feature vector of the "two micro-region map" jd ;

[0017] For the feature vector constructed in step S2, calculate the correlation coefficient D coo ;

[0018] Comprehensively form the improved K-Means clustering algorithm model distance D, that is: D = ωD jd +(1-ω)D coo ;

[0019] ω is the coefficient;

[0020] The clustering results are obtained based on the improved K-Means clustering algorithm model. The actual on-site maintenance strategy is compared with the clustering results. The number of centroids and the ω coefficient are optimized. Whitelist noise points are removed for cases with large differences or inconsistent types. A mapping relationship between W clusters and W types of actual on-site maintenance overall measures is formed.

[0021] Preferably, before step S2, the M×N data matrix formed by a single device in M ​​hours in step S1 is converted into a 1×L feature vector, where L is the number of feature quantities.

[0022] Preferably, in step S2, a longitudinal analysis is performed on a single device grouped by time, and recursive grouping is performed on the data between the current fault time data and the time when the device is back online for maintenance after the last fault, and feature quantities and data samples are constructed for the single device according to the time series and fault grouping data.

[0023] Preferably, the specific process of constructing the feature quantity and data sample is:

[0024] The sample data features of equal division, maximum value and mean value are retained to discretize the state of charge, operating voltage, operating current, operating temperature data collected by the visualization device battery and whether it is online;

[0025] Reconstruct the battery manufacturer and weather data into feature quantities by one-hot and classification

[0026] The number of charge and discharge times, operating hours, snapshot times, and image upload times are summarized or taken as actual fixed values.

[0027] Preferably, principal component analysis (PCA) and normalization methods are used to preprocess data samples of the feature matrix to form a training sample set of data on the battery operating environment and internal physical conditions of the visualization device.

[0028] Preferably, corresponding to the visualized battery status type, the status types are set to Class A, Class B, Class C, Class D, and Class E; among them, Class A, B, and Class C correspond to the emergency power supply installation strategy, but distinguish the emergency power supply capacity; Class D is included in the direct battery replacement category; Class E is a non-battery failure.

[0029] The present invention also discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are executed.

[0030] The present invention further discloses a visualization device battery status monitoring and emergency power supply installation prediction system, including a memory and a processor connected to each other, wherein a computer program is stored on the memory, and when the computer program is run by the processor, the steps of the method described above are executed.

[0031] Compared with the prior art, the advantages of the present invention are:

[0032] The present invention uses the historical maintenance experience of transmission line channel visualization devices, focuses on monitoring the battery status with a higher failure rate and implementing corresponding emergency power supply strategies. Based on the historical operation data of a large number of channel visualization devices, the present invention reconstructs parameters such as the state of charge (SOC), operating voltage, operating current, operating temperature data, and whether online, and converts them into training samples that combine the operating environment and physical quantities. The principal component analysis method and normalization are used to reduce the impact of noise data, and FCNN model training is adopted to realize the monitoring of the battery status of the visualization device and the proposal of corresponding strategies.

[0033] The present invention fully mines the historical operation data of the visualization device to form battery status monitoring and prediction, and proposes an emergency power supply installation strategy in advance, thus changing the state of inefficient maintenance and lack of target focus; through the prediction and verification of the actual operation data of subsequent equipment, the model is promoted to be tuned and improved, the prediction accuracy and robustness are improved, and the battery status monitoring is made simpler after model iteration and optimization; the battery maintenance cost is reduced, and the battery service life is maximized while ensuring the effective power supply of the equipment, so that the maintenance cost of the visualization battery of the transmission line channel is gradually reduced, which is easy to promote.

[0034] The present invention is based on a deep learning algorithm model and combines the particularity of the operating environment of the channel visualization device on the transmission line tower. It combines different operating environments, battery operation data of different visualization devices and fault types, reconstructs the training sample set, optimizes the matching of the prediction model and the emergency power supply installation strategy, and realizes low-cost and efficient support for the precise operation and maintenance of the transmission line channel visualization device.

[0035] The present invention integrates the factors affecting the normal power supply of the battery based on the battery manufacturer, state of charge (SOC), operating voltage and current, number of charge and discharge times, number of operating hours, number of snapshots, number of picture return times, altitude, meteorological data, etc. of the transmission line channel visualization device, constructs data fusion training samples of different visualization batteries of different towers, sets labels of fault categories for the battery states in the training samples through historical maintenance data, uses data preprocessing and deep learning methods to monitor and judge the health status of the visualization device battery by category, proposes different maintenance strategies for different states of the visualization device battery, proposes to install a circulating external emergency power supply to support the normal power supply of the device under severe weather conditions for the visualization battery that is still in a good category, distinguishes the status and subsequent weather forecast, and provides different emergency power supply capacities; for the visualization battery that is in a bulging or failure category, direct replacement and emergency power supply installation are adopted in combination with subsequent weather forecasts, so as to achieve targeted maintenance and avoid blind and high-cost maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The flowchart of the prediction method of the present invention in an embodiment. DETAILED DESCRIPTION

[0037] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0038] like Figure 1 As shown, the method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply provided by an embodiment of the present invention includes the following steps:

[0039] S1: Collect maintenance data of transmission line channel visualization devices (such as cameras) due to battery failure; according to the battery failure type of the offline visualization device, establish a forecast basic data set of battery manufacturer, state of charge (SOC), operating voltage and current, number of charge and discharge times, number of operating hours, number of snapshots, fixed snapshot interval, number of image return times, whether online, altitude, temperature, and weather; through the collection of historical data of the channel visualization maintenance equipment in operation, according to the current visualization device operation strategy, a single device records a basic data set per hour, that is, a device forms an M×N basic data set matrix for M hours, where N is the number of recorded data, also known as the number of rows.

[0040] S2: Construct battery status monitoring and prediction characteristics of transmission line channel visualization device.

[0041] Considering the difference between continuous and discrete quantities, the M×N data matrix formed by a single device in M ​​hours in step S1 needs to be converted into a 1×L feature vector, where L is the number of feature quantities.

[0042] At the same time, considering that the present invention mainly focuses on the status monitoring of the transmission line channel visualization device, and focuses on the analysis and prediction of battery failures. Therefore, firstly, a longitudinal analysis of the single device is performed by grouping by time, and the data between the current fault time data and the normal maintenance time after the last fault is recursively grouped, and the feature quantity and data sample are constructed for the single device according to the time series and fault grouping data, specifically:

[0043] First, the sample data features are retained by equal division, maximum value and average value, and the state of charge (SOC), operating voltage, operating current, operating temperature data collected by the visualization device battery are converted and processed online, as shown in Table 1:

[0044] Table 1 Discretization of continuous basic data

[0045]

[0046] Second, the battery manufacturers and weather data are reconstructed into feature quantities according to the One-Hot and classification methods, see Table 2 for details.

[0047] Table 2 Battery manufacturer, weather data feature processing

[0048]

[0049] Third, normal quantities such as the number of charge and discharge times, operating hours, number of snapshots, and number of image uploads are summarized or taken as actual fixed values.

[0050] Furthermore, after the feature quantities are discretized and aggregated, the feature matrix becomes sparse and diverse. The principal component analysis (PCA) and normalization method are used to preprocess the feature matrix data samples to form a training sample set of the battery operating environment and internal physical working condition data of the visualization device.

[0051] Corresponding to the visualized battery status types, the status types are initially set as Class A, Class B, Class C, Class D, and Class E; among them, Class A, B, and Class C correspond to the emergency power supply installation strategy, but distinguish the emergency power supply capacity; Class D is included in the direct battery replacement category; Class E is a non-battery failure.

[0052] Through the above feature vector construction, a single device training sample matrix X is formed ij , i is the number of single equipment failures (number of basic data set groups), j is the number of feature vectors constructed by grouping and feature quantities through M×N basic data sets.ij The data feature vector fault state type label matrix Yi.

[0053] S3: A single device is constructed based on the LSTM-FCNN prediction model. Considering that the failure of the transmission line channel visualization device is closely related to the device operation time and the gradual impact of the environment, the long short-term memory network (LSTM) model is used for longitudinal prediction of the time series.

[0054] At the same time, considering that the training sample set constructed in step S2 contains both the strong physical principle correlation of the battery and the operating environment that affects the operating status of the battery, it is a fusion of logical relationships and association relationships. Therefore, the fully convolutional neural network (FCNN) model is simultaneously selected to perform deep learning algorithm training and prediction of the internal relationship of the data.

[0055] First, the training sample set constructed in step S2 is divided into a training set, a validation set, and a test set in a ratio of 6:2:2. Secondly, the LSTM-FCNN prediction model is used to predict the battery status type of the visualization device.

[0056] The prediction model is calibrated and tuned. Considering that FCNN is prone to slow convergence and distortion during training, the cross entropy loss function is used to tune and verify the model parameters.

[0057] Through the fusion prediction of the LSTM and FCNN models, the fault type of a single transmission line channel visualization device is predicted. The prediction value Ypre1 of the long short-term memory network (LSTM) model and the prediction value Ypre2 of the full convolutional neural network model are fused and predicted in a weighted manner. The weighted value Y pre The interval and the set state type are mapped to Class A, Class B, Class C, Class D, and Class E to achieve analysis and judgment of the fault type, that is, the predicted value is: Y pre =αY pre1i +(1-α)Y pre2i ;

[0058] Furthermore, the model is iteratively optimized on the training sample set to enhance the robustness of the model.

[0059] S4: Multiple devices are used to construct characteristic cluster analysis of five types of faults. The basic performance and characteristics of the same type of channel visualization devices from the same manufacturer are not much different. However, due to the particularity of the operating environment of the transmission line towers, there are great differences in the fault types, especially due to the large differences in the terrain and landforms of mountainous areas and transmission line towers in some areas, resulting in the characteristics of "two micro" microclimate areas such as heavy ice areas, dancing areas, pollution areas, minefields, and high altitude areas (referred to as five categories); the two micros are microtopography and microclimate.

[0060] Therefore, the present invention constructs the correlation relationship between the fault types of different devices in the horizontal direction based on the analysis of a single device in the time series longitudinal prediction in step S2 and step S3, and on the basis of the status monitoring and prediction of a single transmission line channel visualization device, combines the tower coordinates with five types of "two micro" regional maps, and performs cluster analysis on the correlation of different fault types of multiple visualization devices to form a mapping between the "two micro" regional map and the equipment fault type, so as to achieve a method and guidance for accurate maintenance of the device in emergency maintenance.

[0061] S5: Multiple devices build an improved K-Means clustering algorithm model based on Jaccard and correlation coefficient.

[0062] Based on step S2, construct a single training sample X ij and fault status type label matrix Y i , matching different types of "two micro" regional maps to form training samples of multiple full-volume devices X rt , r is the number of equipment failures, t is X ij The number of feature vectors constructed by expanding the corresponding regional features based on the basic data set. Among them, the feature vectors of five fault types and five "two micro" regions are constructed through One-Hot, namely 001, 010, 011, 100, and 101 respectively.

[0063] The initial centroids of the K-Means cluster are determined to be W, with a value range of 3 to 5. The corresponding cluster centroids are the types of subsequent emergency power supplies and maintenance strategies. The calculation method of the Euclidean distance of the traditional K-Means algorithm is improved, considering the training sample X rt There are differences in the preprocessing of sample feature quantity before and after construction. Therefore, the Jaccard distance D is calculated for the fault type and the One-Hot feature vector of the “two micro-region map”. jd ;

[0064] For the feature vector constructed in step S2, calculate the correlation coefficient D coo ;

[0065] Comprehensively form the improved K-Means clustering algorithm model distance D, that is: D = ωD jd +(1-ω)D coo .

[0066] The actual on-site maintenance strategy is compared with the clustering results, the number of centroids and the ω coefficient are optimized, and whitelist noise points are removed for cases with large differences or inconsistent types. A mapping relationship between W clusters and W types of actual on-site maintenance overall measures is formed to support the precise implementation of battery operation and maintenance of transmission line channel visualization devices.

[0067] In specific applications, the visualization equipment on the large side of the 220 kV Rongkuang Line #086 of the State Grid Chenzhou Power Supply Company was actually monitored to predict battery problems. During the on-site verification process, bulging and failure were found, and the battery was directly replaced and repaired on site.

[0068] The present invention uses the historical maintenance experience of transmission line channel visualization devices, focuses on monitoring the battery status with a higher failure rate and implementing corresponding emergency power supply strategies. Based on the historical operation data of a large number of channel visualization devices, the present invention reconstructs parameters such as the state of charge (SOC), operating voltage, operating current, operating temperature data, and whether online, and converts them into training samples that combine the operating environment and physical quantities. The principal component analysis method and normalization are used to reduce the impact of noise data, and FCNN model training is adopted to realize the monitoring of the battery status of the visualization device and the proposal of corresponding strategies.

[0069] The present invention fully mines the historical operation data of the visualization device to form battery status monitoring and prediction, and proposes an emergency power supply installation strategy in advance, thus changing the state of inefficient maintenance and lack of target focus; through the prediction and verification of the actual operation data of subsequent equipment, the model is promoted to be tuned and improved, the prediction accuracy and robustness are improved, and the battery status monitoring is made simpler after model iteration and optimization; the battery maintenance cost is reduced, and the battery service life is maximized while ensuring the effective power supply of the equipment, so that the maintenance cost of the visualization battery of the transmission line channel is gradually reduced, which is easy to promote.

[0070] The present invention is based on a deep learning algorithm model and combines the particularity of the operating environment of the channel visualization device on the transmission line tower. It combines different operating environments, battery operation data of different visualization devices and fault types, reconstructs the training sample set, optimizes the matching of the prediction model and the emergency power supply installation strategy, and realizes low-cost and efficient support for the precise operation and maintenance of the transmission line channel visualization device.

[0071] The present invention integrates the factors affecting the normal power supply of the battery based on the battery manufacturer, state of charge (SOC), operating voltage and current, number of charge and discharge times, number of operating hours, number of snapshots, number of picture return times, altitude, meteorological data, etc. of the transmission line channel visualization device, constructs data fusion training samples of different visualization batteries of different towers, sets labels of fault categories for the battery states in the training samples through historical maintenance data, uses data preprocessing and deep learning methods to monitor and judge the health status of the visualization device battery by category, proposes different maintenance strategies for different states of the visualization device battery, proposes to install a circulating external emergency power supply to support the normal power supply of the device under severe weather conditions for the visualization battery that is still in a good category, distinguishes the status and subsequent weather forecast, and provides different emergency power supply capacities; for the visualization battery that is in a bulging or failure category, direct replacement and emergency power supply installation are adopted in combination with subsequent weather forecasts, so as to achieve targeted maintenance and avoid blind and high-cost maintenance.

[0072] The present invention also discloses a computer-readable storage medium, on which a computer program is stored, and the computer program executes the steps of the above method when it is executed by a processor. An embodiment of the present invention further provides a visualization device battery status monitoring and emergency power supply installation prediction system, including a memory and a processor connected to each other, the memory storing a computer program, and the computer program executes the steps of the above method when it is executed by the processor.

[0073] The present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by hardware related to computer program instructions. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned method embodiment when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable storage media include: any entity or device that can carry computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. The memory is used to store computer programs and / or modules, and the processor implements various functions by running or executing computer programs and / or modules stored in the memory, and calling data stored in the memory. The memory may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, an internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0074] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for visualizing device battery status monitoring and emergency power supply installation prediction, characterized in that: Includes steps: S1. Obtain data on battery failure in a transmission line channel visualization device, and establish a prediction basic data set based on the type of battery failure in the visualization device; S2. Constructing the battery status monitoring and prediction feature quantity of the transmission line channel visualization device based on the prediction basic data set, and preprocessing the feature quantity to form a sample set of the battery operating environment and internal physical working condition data of the visualization device; S3. Input the sample set into the pre-built long short-term memory network model and the fully convolutional neural network model for prediction, obtain the first prediction value and the second prediction value, and then perform weighted fusion on the first prediction value and the second prediction value to obtain the final fault type prediction result.

2. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 1, characterized in that: It also includes step S4: on the basis of analyzing the longitudinal prediction of a single device in time series in step S2 and step S3, constructing the correlation relationship between the fault types of different devices in the horizontal direction, and on the basis of the status monitoring and prediction of a single transmission line channel visualization device, combining the tower coordinates with five categories of two-micro-area diagrams, clustering analysis is performed on the correlation of different fault types of multiple visualization devices to form a mapping between the two-micro-area diagrams and the equipment fault types; the five categories include heavy ice areas, dancing areas, pollution areas, minefields and high altitude areas; the two micros are micro-topography and micro-climate.

3. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 2, characterized in that: After step S4, the method further includes step S5: Calculate the Jaccard distance D based on the fault type and the One-Hot feature vector of the "two micro-region map" jd ; For the feature vector constructed in step S2, calculate the correlation coefficient D coo ; Comprehensively form the improved K-Means clustering algorithm model distance D, that is: D = ωD jd +(1-ω)D coo ; ω is the coefficient; The clustering results are obtained based on the improved K-Means clustering algorithm model. The actual on-site maintenance strategy is compared with the clustering results. The number of centroids and the ω coefficient are optimized. Whitelist noise points are removed for cases with large differences or inconsistent types. A mapping relationship between W clusters and W types of actual on-site maintenance overall measures is formed.

4. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 1, 2 or 3, characterized in that: Before step S2, the M×N data matrix formed by a single device in M ​​hours in step S1 is converted into a 1×L feature vector, where L is the number of feature quantities.

5. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 4, characterized in that: In step S2, a longitudinal analysis of a single device is performed by grouping it by time, and recursive grouping is performed on the data between the current fault time data and the time when the maintenance and maintenance are back online after the last fault, and feature quantities and data samples are constructed for a single device according to the time series and fault grouping data.

6. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 5, characterized in that: The specific process of constructing feature quantities and data samples is as follows: The sample data features of equal division, maximum value and mean value are retained to discretize the state of charge, operating voltage, operating current, operating temperature data collected by the visualization device battery and whether it is online; Reconstruct the battery manufacturer and weather data into feature quantities by one-hot and classification The number of charge and discharge times, operating hours, snapshot times, and image upload times are summarized or taken as actual fixed values.

7. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 5, characterized in that: The principal component analysis (PCA) and normalization method are used to preprocess the data samples of the feature matrix to form a training sample set of the battery operating environment and internal physical working condition data of the visualization device.

8. The method for monitoring battery status of a visualization device and predicting the installation of an emergency power supply according to claim 5, characterized in that: Corresponding to the visualized battery status type, the status types are set to Class A, Class B, Class C, Class D, and Class E; among them, Class A, B, and Class C correspond to the emergency power supply installation strategy, but distinguish the emergency power supply capacity; Class D is included in the direct battery replacement category; Class E is non-battery failure.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program performs the steps of the method according to any one of claims 1 to 8.

10. A visualization device battery status monitoring and emergency power supply installation prediction system, comprising a memory and a processor connected to each other, wherein a computer program is stored in the memory, characterized in that: When the computer program is executed by a processor, the computer program performs the steps of the method according to any one of claims 1 to 8.

Images