Method for detecting electric quantity of network camera through dynamic load prediction

Abstract

The invention discloses a method for detecting the electric quantity of a network camera through dynamic load prediction, and relates to the technical field of network cameras. S2, load feature extraction; s3, constructing a dynamic load prediction model; s4, estimating power consumption; and S5, outputting and feeding back a result. According to the network camera electric quantity detection method based on dynamic load prediction, the actual use condition of the network camera is monitored, the dynamic load prediction algorithm is combined, the electric quantity consumption condition of the battery under different load conditions is estimated, the actual use condition of the battery can be accurately reflected, and therefore the electric quantity detection precision is improved; and meanwhile, the electric quantity prediction result is updated in real time by monitoring the use condition of the camera in real time, so that a user can know the change condition of the electric quantity of the battery in time, corresponding measures are taken to manage the electric quantity, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of network cameras, and in particular to a method for detecting power consumption of a network camera through dynamic load prediction. Background Art

[0002] Network cameras are a new generation of cameras that combine traditional cameras with network technology. They not only capture video images but also transmit them to remote users over the network, allowing them to monitor the video through a standard web browser without installing any specialized software. Monitoring the battery level of network cameras is crucial for ensuring stable operation, especially in scenarios where they rely on battery power or require long-term online monitoring.

[0003] IP camera battery level testing refers to the process of monitoring and managing the battery level used by the IP camera. Because IP cameras may need to operate in a variety of environments, including wired and wireless scenarios, battery level becomes a key factor affecting their operating time and stability. Therefore, accurately testing the battery level of the IP camera helps users to promptly understand battery usage and take appropriate measures, such as replacing or charging the battery, to ensure the normal operation of the IP camera. Currently, common testing methods include: direct measurement (by connecting the circuit between the battery and the camera and directly measuring the battery voltage or current to determine the battery level), battery indicator (the camera has a built-in battery indicator function, which displays the remaining power on the display or indicator light), and software monitoring (using the camera's management software or app to monitor the battery level in real time).

[0004] However, in actual testing, existing battery level detection methods are susceptible to various factors. For example, battery aging causes a decrease in battery capacity, but the detection methods cannot accurately reflect this change, resulting in inaccurate battery level indication, which affects the accuracy of battery level detection. Most existing detection methods also suffer from poor real-time performance and are unable to reflect changes in battery level in real time. For example, direct measurement methods require periodic measurement operations to update the battery level, and software monitoring methods may be affected by factors such as network latency. Furthermore, different network camera brands and models often require different battery and power detection methods, resulting in poor compatibility, which undoubtedly increases user difficulty and cost.

[0005] Therefore, there is an urgent need to improve this shortcoming. The present invention studies and improves the existing technology and its shortcomings, and provides a method for detecting the power consumption of a network camera through dynamic load prediction. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for detecting power consumption of a network camera through dynamic load prediction, so as to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: a method for detecting the power consumption of a network camera through dynamic load prediction, comprising the following steps:

[0008] S1. Data collection and preprocessing: Collect the usage data of network cameras, including shooting frequency, resolution, brightness, working hours, etc., and perform preprocessing;

[0009] S2, load feature extraction: extracting load features that can reflect the actual usage of the network camera from the preprocessed data;

[0010] S3. Dynamic load prediction model construction: Based on the extracted load characteristics, select the appropriate prediction algorithm and build a dynamic load prediction model;

[0011] S4, power consumption estimation: Based on the dynamic load prediction model, the power consumption of the network camera under different load conditions is estimated;

[0012] S5. Result output and feedback: Output the power consumption estimation results to the user in the form of charts or reports, collect user feedback, understand the performance of the model in actual applications, and optimize the model based on user feedback.

[0013] Furthermore, in step S1, using sensors or software tools to collect usage data of the network camera and setting data collection parameters specifically includes:

[0014] Sampling frequency: determines the accuracy and real-time nature of data collection. Different sampling frequencies can be set based on actual needs, such as once per second, once per minute, or once per hour.

[0015] Data format: determines the readability and processability of the data. Common data formats include CSV, JSON, XML, etc. Choose an appropriate data format to facilitate subsequent data analysis and processing.

[0016] Data storage: The data storage method determines the data's persistence and accessibility. You can choose to store data on a local computer, remote server, or cloud storage platform. You can also set different storage policies based on data volume and access requirements, such as real-time storage, scheduled backup, or periodic cleanup.

[0017] Data security: Setting access permissions, encrypting transmission and storage, and other measures to protect the confidentiality and integrity of data;

[0018] Other parameters: Based on actual needs, you can also set other parameters, such as the time range for data collection, trigger conditions (such as when a specific event occurs), etc.

[0019] Furthermore, in step S1, the preprocessing of the usage data of the network camera includes data cleaning, format conversion, and outlier processing to ensure the accuracy and consistency of the data;

[0020] 1) Data cleaning: Remove duplicate, invalid or abnormal data. Specific steps:

[0021] Identify duplicate data: Use unique identifiers (such as timestamps, serial numbers, etc.) to identify and delete duplicate data records;

[0022] Remove invalid data: Check the data for empty, missing, or obviously illogical values (such as negative brightness values) and remove or replace them with appropriate default values.

[0023] Handling incomplete data: For missing values in certain fields, use interpolation, regression, or mean filling to complete the data;

[0024] 2) Format conversion: Convert the data into a format suitable for subsequent processing. Specific steps:

[0025] Unified data format: Ensure that the data types of all fields are consistent, such as converting date and time fields to a unified date and time format, and converting numeric fields to floating point or integer types;

[0026] Standardized data units: For data involving different units (such as brightness values may be expressed in different units), convert them into unified units;

[0027] Split or merge fields: Split some composite fields into multiple simple fields, or merge multiple related fields into one composite field according to the needs of subsequent processing;

[0028] 3) Outlier processing: Identify and process outliers in the data. Specific steps:

[0029] Identify outliers: Use statistical methods (such as box plots, Z scores, etc.) or machine learning algorithms (such as isolation forests) to identify outliers in the data;

[0030] Handling outliers: Different strategies are used to handle outliers based on their nature and processing requirements. The methods include:

[0031] Delete outliers: Delete outliers that clearly do not conform to logic or business rules directly;

[0032] Replace outliers: For some outliers, use statistics such as mean, median or mode to replace them;

[0033] Correcting outliers: Attempts to correct outliers due to data entry errors or equipment failures, etc.

[0034] Retain outliers: Retain and give special treatment to outliers that contain important information or patterns.

[0035] Furthermore, in step S2, the specific process of load feature extraction is: analyzing the parameter change pattern in the data, using statistical methods or machine learning algorithms to extract load features, and verifying and optimizing the extracted features to ensure that they can accurately reflect the load conditions of the network camera.

[0036] Furthermore, the analysis parameter variation rules specifically include:

[0037] Shooting frequency: The number of times a surveillance camera shoots or the frame rate within a certain period of time is analyzed, and its changing trends in different time periods (such as daytime, nighttime, weekdays, and weekends) are analyzed to identify periodic changes in shooting frequency, such as an increase or decrease in the number of shots during a fixed time period each day.

[0038] Resolution: Tracks changes in the resolution of video captured by the camera and analyzes its selection in different scenes or conditions to identify patterns in resolution adjustment, such as whether the resolution is automatically adjusted to maintain image quality when lighting changes;

[0039] Brightness: Monitors changes in the brightness of the camera's environment, or the average brightness of the video image, and analyzes the impact of brightness changes on camera performance (such as exposure time and gain), as well as the camera's adaptation strategy to brightness changes.

[0040] Furthermore, the feature extraction using statistical methods or machine learning algorithms specifically includes:

[0041] 1) Statistical methods:

[0042] Use statistics such as mean, median, and standard deviation to describe the distribution of shooting frequency, resolution, and brightness parameters;

[0043] Use time series analysis to identify cyclical changes and trends in parameters;

[0044] Correlation analysis was used to assess the degree of association between different parameters;

[0045] 2) Machine Learning Algorithms:

[0046] Apply clustering algorithms (such as K-means) to divide the shooting frequency, resolution, and brightness parameters into different categories or modes;

[0047] Use classification algorithms (such as decision trees and support vector machines) to identify different camera usage scenarios or states;

[0048] Regression algorithms (such as linear regression and random forest regression) are used to predict the load of cameras in future time periods.

[0049] Furthermore, the specific operation of verifying the features is as follows: by comparing the extracted features with the actual usage of the camera, the accuracy and effectiveness of the features are verified, and a cross-validation method is used to evaluate the performance of the features on different data sets.

[0050] Furthermore, the specific operation of the feature optimization is as follows: according to the verification result, the parameters and methods of feature extraction are adjusted, and new features are introduced or existing features are combined to more comprehensively reflect the load condition of the camera.

[0051] Furthermore, in step S3, the selectable prediction algorithms include time series analysis, neural network, and support vector machine (SVM), and a corresponding prediction model is constructed according to the selected prediction algorithm, as follows:

[0052] 1) If time series analysis is selected, an ARIMA model is built based on historical data;

[0053] Time series analysis predicts future load conditions based on past load data. It can help identify underlying patterns and trends, such as cyclical changes and seasonal fluctuations. Common time series models include autoregressive (AR), moving average (MA), ARMA, and ARIMA models.

[0054] 2) If a neural network is selected, a BP neural network or LSTM network is constructed;

[0055] Neural networks can adaptively adjust model parameters to better adapt to changes in load. Commonly used neural network models in load forecasting include BP neural networks, RNNs (recurrent neural networks), and LSTMs (long short-term memory networks).

[0056] 3) If you choose Support Vector Machine (SVM), you need to build an SVR (Support Vector Regression) model;

[0057] SVM is a machine learning method based on statistical learning theory. It has many unique advantages in solving small sample, nonlinear and high-dimensional pattern recognition problems. In load forecasting, SVM can handle nonlinear problems by introducing kernel functions. Commonly used kernel functions include linear kernel, polynomial kernel, radial basis kernel (RBF), etc.

[0058] Furthermore, in step S4, the specific process of power consumption estimation is as follows: based on the load conditions obtained by the prediction model, the power consumption rate of the battery under different loads is calculated, and the remaining power of the battery is estimated in combination with the initial power and working time of the battery;

[0059] 1) Calculate the battery consumption rate under different loads based on the load conditions obtained by the prediction model: Utilize the established dynamic load prediction model to predict the future load conditions of the network camera. The prediction results include load changes over different time periods, such as shooting frequency and data transmission volume. Then, based on the predicted load conditions and the power consumption characteristics of the network camera (such as power consumption in different operating modes such as shooting and transmission), calculate the battery consumption rate under different loads.

[0060] 2) Estimate the remaining battery capacity based on the battery's initial charge and operating time: Obtain the battery's initial charge through a battery management system (BMS) or direct measurement. The initial charge should be expressed as a percentage or milliampere-hours (mAh) for subsequent calculations. Then, based on the network camera's usage plan or operating time obtained from a prediction model (the operating time should take into account the network camera's different operating modes and load changes), determine the total duration the battery needs to supply power. Finally, use a mathematical model or algorithm to estimate the remaining battery capacity based on the battery's initial charge, power consumption rate, and operating time.

[0061] The present invention provides a method for detecting the power consumption of a network camera through dynamic load prediction, which has the following beneficial effects:

[0062] The present invention monitors the actual usage of network cameras and combines them with a dynamic load prediction algorithm to estimate the power consumption of batteries under different load conditions. It can accurately reflect the actual usage of the battery, thereby improving the accuracy of power detection. At the same time, by monitoring the usage of the camera in real time, the power prediction results are updated in real time, which helps users to promptly understand the changes in battery power and take corresponding measures to manage power and reduce maintenance costs. In addition, the prediction method of the present invention has strong adaptability and flexibility, and can be customized and optimized according to different camera models and usage scenarios, making it applicable to a wider range of scenarios and user needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A schematic flow chart of the steps of a method for detecting power consumption of a network camera through dynamic load prediction according to the present invention;

[0064] Figure 2 The diagram is a prediction algorithm-prediction model diagram of a method for detecting power consumption of a network camera through dynamic load prediction according to the present invention. DETAILED DESCRIPTION

[0065] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0066] like Figure 1-Figure 2 As shown, a method for detecting power consumption of a network camera through dynamic load prediction includes the following steps:

[0067] S1. Data collection and preprocessing: Collect the usage data of network cameras, including shooting frequency, resolution, brightness, working hours, etc., and perform preprocessing;

[0068] In this embodiment, sensors or software tools are used to collect usage data of network cameras, and data collection parameters are set, specifically including:

[0069] Sampling frequency: determines the accuracy and real-time nature of data collection. Different sampling frequencies can be set based on actual needs, such as once per second, once per minute, or once per hour.

[0070] Data format: determines the readability and processability of the data. Common data formats include CSV, JSON, XML, etc. Choose an appropriate data format to facilitate subsequent data analysis and processing.

[0071] Data storage: The data storage method determines the data's persistence and accessibility. You can choose to store data on a local computer, remote server, or cloud storage platform. You can also set different storage policies based on data volume and access requirements, such as real-time storage, scheduled backup, or periodic cleanup.

[0072] Data security: Setting access permissions, encrypting transmission and storage, and other measures to protect the confidentiality and integrity of data;

[0073] Other parameters: Based on actual needs, you can also set other parameters, such as the time range for data collection, trigger conditions (such as when a specific event occurs), etc.

[0074] In this embodiment, the preprocessing of the usage data of the network camera includes data cleaning, format conversion, and outlier processing to ensure the accuracy and consistency of the data;

[0075] 1) Data cleaning: Remove duplicate, invalid or abnormal data. Specific steps:

[0076] Identify duplicate data: Use unique identifiers (such as timestamps, serial numbers, etc.) to identify and delete duplicate data records;

[0077] Remove invalid data: Check the data for empty, missing, or obviously illogical values (such as negative brightness values) and remove or replace them with appropriate default values.

[0078] Handling incomplete data: For missing values in certain fields, use interpolation, regression, or mean filling to complete the data;

[0079] 2) Format conversion: Convert the data into a format suitable for subsequent processing. Specific steps:

[0080] Unified data format: Ensure that the data types of all fields are consistent, such as converting date and time fields to a unified date and time format, and converting numeric fields to floating point or integer types;

[0081] Standardized data units: For data involving different units (such as brightness values may be expressed in different units), convert them into unified units;

[0082] Split or merge fields: Split some composite fields into multiple simple fields, or merge multiple related fields into one composite field according to the needs of subsequent processing;

[0083] 3) Outlier processing: Identify and process outliers in the data. Specific steps:

[0084] Identify outliers: Use statistical methods (such as box plots, Z scores, etc.) or machine learning algorithms (such as isolation forests) to identify outliers in the data;

[0085] Handling outliers: Different strategies are used to handle outliers based on their nature and processing requirements. The methods include:

[0086] Delete outliers: Delete outliers that clearly do not conform to logic or business rules directly;

[0087] Replace outliers: For some outliers, use statistics such as mean, median or mode to replace them;

[0088] Correcting outliers: Attempts to correct outliers due to data entry errors or equipment failures, etc.

[0089] Retain outliers: Retain and give special treatment to outliers that contain important information or patterns;

[0090] S2. Load feature extraction: Extract load features that reflect the actual usage of the network camera from the preprocessed data. The specific process of this step is to analyze the parameter variation patterns in the data, use statistical methods or machine learning algorithms to extract load features, and verify and optimize the extracted features to ensure that they can accurately reflect the load conditions of the network camera.

[0091] In this embodiment, analyzing the parameter variation pattern specifically includes:

[0092] Shooting frequency: The number of times a surveillance camera shoots or the frame rate within a certain period of time is analyzed, and its changing trends in different time periods (such as daytime, nighttime, weekdays, and weekends) are analyzed to identify periodic changes in shooting frequency, such as an increase or decrease in the number of shots during a fixed time period each day.

[0093] Resolution: Tracks changes in the resolution of video captured by the camera and analyzes its selection in different scenes or conditions to identify patterns in resolution adjustment, such as whether the resolution is automatically adjusted to maintain image quality when lighting changes;

[0094] Brightness: Monitors changes in the brightness of the camera's environment, or the average brightness of the video image, and analyzes the impact of brightness changes on camera performance (such as exposure time and gain), as well as the camera's adaptation strategy to brightness changes.

[0095] In this embodiment, using statistical methods or machine learning algorithms to extract features specifically includes:

[0096] 1) Statistical methods:

[0097] Use statistics such as mean, median, and standard deviation to describe the distribution of shooting frequency, resolution, and brightness parameters;

[0098] Use time series analysis to identify cyclical changes and trends in parameters;

[0099] Correlation analysis was used to assess the degree of association between different parameters;

[0100] 2) Machine Learning Algorithms:

[0101] Apply clustering algorithms (such as K-means) to divide the shooting frequency, resolution, and brightness parameters into different categories or modes;

[0102] Use classification algorithms (such as decision trees and support vector machines) to identify different camera usage scenarios or states;

[0103] Use regression algorithms (such as linear regression and random forest regression) to predict the load of cameras in future time periods;

[0104] Verify the specific operation of the feature: By comparing the extracted features with the actual camera usage, verify the accuracy and effectiveness of the features, and use the cross-validation method to evaluate the performance of the features on different data sets;

[0105] Specific operations for feature optimization: Based on the verification results, adjust the parameters and methods of feature extraction, introduce new features or combine existing features to more comprehensively reflect the camera load;

[0106] S3. Dynamic load forecasting model construction: Based on the extracted load characteristics, select an appropriate forecasting algorithm to build a dynamic load forecasting model. The available forecasting algorithms include time series analysis, neural networks, and support vector machines (SVM). Based on the selected forecasting algorithm, a corresponding forecasting model is constructed as follows:

[0107] 1) If time series analysis is selected, an ARIMA model is built based on historical data;

[0108] Time series analysis predicts future load conditions based on past load data. It can help identify underlying patterns and trends, such as cyclical changes and seasonal fluctuations. Common time series models include autoregressive (AR), moving average (MA), ARMA, and ARIMA models.

[0109] 2) If a neural network is selected, a BP neural network or LSTM network is constructed;

[0110] Neural networks can adaptively adjust model parameters to better adapt to changes in load. Commonly used neural network models in load forecasting include BP neural networks, RNNs (recurrent neural networks), and LSTMs (long short-term memory networks).

[0111] 3) If you choose Support Vector Machine (SVM), you need to build an SVR (Support Vector Regression) model;

[0112] SVM is a machine learning method based on statistical learning theory. It has many unique advantages in solving small sample, nonlinear and high-dimensional pattern recognition problems. In load forecasting, SVM can handle nonlinear problems by introducing kernel functions. Commonly used kernel functions include linear kernel, polynomial kernel, radial basis kernel (RBF), etc.

[0113] S4. Power consumption estimation: Based on the dynamic load prediction model, estimate the power consumption of the network camera under different load conditions. The specific process of this step is as follows: Based on the load conditions obtained by the prediction model, calculate the power consumption rate of the battery under different loads, and combine the initial battery power and operating time to estimate the remaining battery power;

[0114] 1) Calculating the battery power consumption rate under different loads based on the load conditions obtained by the prediction model: The established dynamic load prediction model is used to predict the future load conditions of the network camera. The prediction results include load changes in different time periods, such as shooting frequency and data transmission volume. Then, based on the predicted load conditions and the power consumption characteristics of the network camera (such as power consumption in different operating modes such as shooting and transmission), the battery power consumption rate under different loads is calculated. In this embodiment, the power consumption rate is expressed as a percentage of power consumed per unit time (such as per hour) or milliampere hours (mAh);

[0115] 2) Estimate the remaining battery capacity based on the initial battery capacity and operating time: Obtain the initial battery capacity through a battery management system (BMS) or direct measurement. The initial capacity should be expressed as a percentage or milliampere-hours (mAh) for subsequent calculations. Then, based on the network camera's usage plan or operating time obtained from a prediction model (the operating time should take into account the network camera's different operating modes and load changes), determine the total duration the battery needs to supply power. Finally, combine the battery's initial capacity, power consumption rate, and operating time to estimate the remaining battery capacity using a mathematical model or algorithm. In this embodiment, the estimation result can be expressed as a percentage of remaining capacity or remaining operating time.

[0116] S5. Result output and feedback: Output the power consumption estimation results to the user in the form of charts or reports, collect user feedback, understand the performance of the model in actual applications, and optimize the model based on user feedback.

[0117] The embodiments of the present invention are presented for purposes of illustration and description and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those skilled in the art. The embodiments are chosen and described in order to better illustrate the principles of the invention and its practical application and to enable those skilled in the art to understand the invention and design various embodiments with various modifications as suited for specific applications.

Claims

1. A method for detecting power consumption of a network camera through dynamic load prediction, characterized in that: The following steps are involved: S1. Data collection and preprocessing: Collect usage data of network cameras and perform preprocessing; S2, load feature extraction: extracting load features that can reflect the actual usage of the network camera from the preprocessed data; S3. Dynamic load prediction model construction: Based on the extracted load characteristics, select the appropriate prediction algorithm and build a dynamic load prediction model; S4, power consumption estimation: Based on the dynamic load prediction model, the power consumption of the network camera under different load conditions is estimated; S5. Result output and feedback: Output the power consumption estimation results to the user in the form of charts or reports, collect user feedback, and optimize the model based on user feedback.

2. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 1, characterized in that: In step S1, sensors or software tools are used to collect usage data of the network camera, and data collection parameters are set, including sampling frequency, data format, data storage, data security and other parameters.

3. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 1, characterized in that: In step S1, the preprocessing of the usage data of the network camera includes data cleaning, format conversion, and outlier processing; 1) Data cleaning: Remove duplicate, invalid or abnormal data. Specific steps: Identify duplicate data: Use unique identifiers to identify and remove duplicate data records; Remove invalid data: Check the data for empty, missing, or obviously illogical values and remove or replace them with appropriate default values. Handling incomplete data: For fields with missing values, use interpolation, regression, or mean filling to complete the data; 2) Format conversion: Convert the data into a format suitable for subsequent processing. Specific steps: Unified data format: ensure that the data types of all fields are consistent; Standardized data units: For data involving different units, convert them into unified units; Split or merge fields: Split some composite fields into several simple fields, or merge several related fields into one composite field according to the needs of subsequent processing; 3) Outlier processing: Identify and process outliers in the data. Specific steps: Identify outliers: Use statistical methods or machine learning algorithms to identify outliers in the data; Handling outliers: Different strategies are used to handle outliers based on their nature and processing requirements. The methods include: Delete outliers: Delete outliers that clearly do not conform to logic or business rules directly; Replace outliers: For some outliers, use the mean, median or mode to replace them; Correcting outliers: Attempts to correct outliers due to data entry errors or equipment failures; Retain outliers: Retain and give special treatment to outliers that contain important information or patterns.

4. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 1, characterized in that: In step S2, the specific process of load feature extraction is: analyzing the parameter change pattern in the data, using statistical methods or machine learning algorithms to extract load features, and verifying and optimizing the extracted features to ensure that they can accurately reflect the load conditions of the network camera.

5. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 4, characterized in that: The analysis parameter variation rules specifically include: Shooting frequency: Monitor the number of times a camera shoots or its frame rate within a certain period of time and analyze its changing trends over different time periods to identify periodic changes in shooting frequency. Resolution: Track changes in the resolution of video captured by the camera and analyze its selection in different scenarios or conditions to identify patterns in resolution adjustment; Brightness: Monitors changes in the brightness of the camera's environment, or the average brightness of the video image, and analyzes the impact of brightness changes on camera performance, as well as the camera's adaptation strategy to brightness changes.

6. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 4, characterized in that: The feature extraction using statistical methods or machine learning algorithms specifically includes: 1) Statistical methods: Use statistics to describe the distribution of shooting frequency, resolution, and brightness parameters; Use time series analysis to identify cyclical changes and trends in parameters; Correlation analysis was used to assess the degree of association between different parameters; 2) Machine Learning Algorithms: Apply clustering algorithms to classify shooting frequency, resolution, and brightness parameters into different categories or modes; Use classification algorithms to identify different camera usage scenarios or states; A regression algorithm is used to predict the camera load in the future time period.

7. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 4, characterized in that: The specific operation of the feature verification is to verify the accuracy and effectiveness of the feature by comparing the extracted features with the actual usage of the camera, and use the cross-validation method to evaluate the performance of the feature on different data sets.

8. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 4, characterized in that: The specific operation of the feature optimization is as follows: according to the verification results, the parameters and methods of feature extraction are adjusted, and new features are introduced or existing features are combined to more comprehensively reflect the load condition of the camera.

9. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 1, characterized in that: In step S3, the selectable prediction algorithms include time series analysis, neural network, and support vector machine, and a corresponding prediction model is constructed according to the selected prediction algorithm, as follows: 1) If time series analysis is selected, an ARIMA model is built based on historical data; 2) If a neural network is selected, a BP neural network or LSTM network is constructed; 3) If you choose support vector machine, you need to build an SVR model.

10. The method for detecting power consumption of a network camera by dynamic load prediction according to claim 1, characterized in that: In step S4, the specific process of power consumption estimation is as follows: based on the load conditions obtained by the prediction model, the power consumption rate of the battery under different loads is calculated, and the remaining power of the battery is estimated based on the initial power and working time of the battery; 1) Calculate the battery consumption rate under different loads based on the load conditions obtained by the prediction model: Utilize the established dynamic load prediction model to predict the future load conditions of the network camera. The prediction results include the load changes in different time periods. Then, based on the predicted load conditions and the power consumption characteristics of the network camera, calculate the battery consumption rate under different loads. 2) Estimate the remaining battery capacity based on the initial battery charge and operating time: Obtain the initial battery charge through a battery management system or direct measurement. Then, based on the network camera's usage plan or operating time obtained from a prediction model, determine the total battery life. Finally, use a mathematical model or algorithm to estimate the remaining battery capacity based on the initial battery charge, power consumption rate, and operating time.

Images