Analysis method for predicting residual life of wide-fixed network equipment based on artificial intelligence algorithm

Abstract

The invention belongs to the technical field of Wide-Fixed Network equipment, and particularly relates to an analysis method for predicting the residual life of Wide-Fixed Network equipment based on an artificial intelligence algorithm, and the method comprises the following steps: S1) data collection: collecting the performance index data, environmental stress data and hardware aging data of the equipment in real time through a sensor monitored by a system or carried by the equipment; s2) data preprocessing; s3) feature selection: key features related to the service life of the equipment are selected from the acquired data types; forming a time sequence data set of the key features; s4) constructing and training an LSTM model: selecting a long short-term memory network model, and constructing a multi-layer LSTM architecture; and S5) LSTM model application: inputting the new time sequence data into the LSTM model, and predicting the residual life of the equipment. The data quality is improved through data preprocessing, and key features with large correlation are selected as model input by adopting a Pearson's correlation coefficient; and the prediction performance of the LSTM model is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of broadband and solid network equipment, and in particular, relates to a method for analyzing the remaining life of broadband and solid network equipment based on an artificial intelligence algorithm. Background Art

[0002] Broadband network equipment infrastructure is built in industries such as power, transportation, and petrochemicals, where the stable operation of broadband network equipment is crucial. However, due to the complex operating environment of the equipment and the diverse aging mechanisms, accurately predicting its remaining life has always been a challenge.

[0003] On the one hand, existing methods rely on regular maintenance or judge the remaining life based on failure interval experience, which has low prediction accuracy and easily leads to over-maintenance or under-maintenance.

[0004] On the other hand, in the analysis process based on machine learning algorithms, the time series analysis method using a single feature is difficult to capture the complex interactive relationship between the remaining life of the equipment and environmental factors, resulting in inaccurate prediction results; and the lack of an effective feature selection mechanism, redundant or irrelevant information will interfere with model training, further reducing prediction performance. Summary of the invention

[0005] In order to solve the technical problem that the existing methods lack an effective feature selection mechanism to capture the key features of the remaining life of the equipment, resulting in inaccurate prediction results, the present invention provides an analysis method for predicting the remaining life of broadband solid-state network equipment based on an artificial intelligence algorithm.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm includes the following steps:

[0008] S1) Data collection: Real-time collection of equipment performance indicator data, environmental stress data, and hardware aging data through system monitoring or the equipment’s own sensors;

[0009] The performance indicator data specifically includes throughput, latency or packet loss rate; the environmental stress data specifically includes temperature or vibration intensity; and the hardware aging data specifically includes capacitor ESR value.

[0010] S2) Data preprocessing: Data cleaning of the collected time series data, including filling missing values ​​and abnormal data processing, to ensure data quality;

[0011] Preferably, the method for filling missing data values ​​is to use linear interpolation or polynomial interpolation to fill missing values, thereby improving the continuity of data.

[0012] Preferably, the method flow of abnormal data processing includes:

[0013] Use statistical methods (such as z-score or ) to detect outliers in time series data;

[0014] Use smoothing methods to correct outliers or remove obviously erroneous outliers.

[0015] Preferably, it also includes: standardizing the time series data, scaling the feature data of different dimensions to the same range, and preventing features with large values ​​from having too great an impact on model training.

[0016] S3) Feature selection: Select key features related to equipment life from the collected data types; compose the time series data set of key features;

[0017] The key features include at least three of throughput, latency, packet loss rate, temperature, vibration intensity, and capacitor ESR value.

[0018] Preferably, in step S3), the Pearson correlation coefficient is used to select key features related to the equipment life from the collected data types; the specific calculation steps include:

[0019] Calculate the remaining life of the equipment based on known data, denoted as Y;

[0020] Using the preprocessed time series data, the key features of each time point are paired with the corresponding remaining life value to form a vector data pair (X i , Y);

[0021] For each feature X i , where i = 1, 2, ..., n; calculate the Pearson correlation coefficient ρ between it and the remaining life Y; the calculation formula is:

[0022]

[0023] In the formula, X ij is the i-th eigenvalue of the j-th key feature; is the mean of the i-th feature; is Y i The remaining life value of the jth key feature; is the mean of the remaining life; j is the type of key feature;

[0024] Sort the Pearson correlation coefficients of all key features from large to small, and take the absolute value of the Pearson correlation coefficient;

[0025] A threshold is set and key features whose Pearson correlation coefficient is greater than the threshold are selected as the input of the model.

[0026] Preferably, the method for calculating the remaining life of the equipment based on known data is:

[0027] Based on the correlation between the performance index of the equipment and its life span, a performance threshold is defined. When the performance of the equipment is lower than the threshold, the equipment is considered to be failed.

[0028] Based on the known degradation time curve of the performance indicators of the equipment, the covariance value is used to calculate the remaining life of the equipment at a certain time point.

[0029] Preferably, the threshold is set to 0.6-0.7.

[0030] S4) LSTM model construction and training: A long short-term memory network (LSTM) model is selected and a multi-layer LSTM architecture is constructed: the input layer is a time series data set with key features.

[0031] The LSTM layer contains multiple layers of LSTM units to capture long-term dependencies in time series data. The activation function of the LSTM unit is tanh or relu function. The Dropout rate is set to prevent overfitting.

[0032] The fully connected layer is used to map the output of the LSTM layer to the predicted value of the remaining life of the equipment; the activation function of the fully connected layer unit selects a linear function;

[0033] The output layer outputs the remaining life prediction results of the equipment;

[0034] Model training: Divide the time series data set into training set, validation set and test set, and use the training set data to train the LSTM model; optimize by adjusting the model's hyperparameters; use the mean square error (MSE) loss function and Adam optimizer to update the model parameters;

[0035] Preferably, in the division of the time series data, the time of the training set data is earlier than that of the validation set data, and the time of the validation set data is earlier than that of the test set data, so as to ensure continuity in time.

[0036] Preferably, the method further includes: using validation set data to monitor the loss function value of the model, and terminating the training early when the loss function value no longer decreases within several rounds.

[0037] Preferably, it also includes: using root mean square error (RMSE), mean absolute error (MAE) or R-squared to evaluate model performance.

[0038] S5) LSTM model application:

[0039] New time series data is input into the LSTM model to predict the remaining life of the equipment and provide a decision-making basis for equipment maintenance and replacement.

[0040] Beneficial effects of the present invention:

[0041] 1. Improve data quality through data preprocessing, and use the Pearson correlation coefficient to select key features with large correlation as model input; improve the prediction performance of the LSTM model, thereby outputting accurate and reliable remaining life prediction values.

[0042] 2. By building a multi-layer LSTM architecture, stacking multiple layers of LSTM can increase the complexity of the model, capture more complex timing relationships, and enhance the applicability to various types of broadband and fixed network equipment and complex usage scenarios; providing a decision-making basis for equipment maintenance and replacement. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0044] Figure 1 The present invention is a flow chart of a method for predicting the remaining life analysis of broadband and solid-state network equipment based on an artificial intelligence algorithm.

[0045] Figure 2 This is a specific flow chart of step S4) LSTM model construction and training in a method for predicting the remaining life of broadband and fixed network equipment based on an artificial intelligence algorithm in the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] See also Figure 1-Figure 2 As shown, a method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm includes the following steps:

[0048] S1) Data collection: Real-time collection of equipment performance indicator data, environmental stress data, and hardware aging data through system monitoring or the equipment’s own sensors;

[0049] The performance indicator data specifically includes throughput, latency or packet loss rate; the environmental stress data specifically includes temperature or vibration intensity; and the hardware aging data specifically includes capacitor ESR value.

[0050] Specifically, the system monitoring tools or the built-in sensors of the equipment are used to collect various data of the equipment to ensure that the collected data is representative and covers the key performance and operating environment of the equipment. The quality and integrity of the data directly affect the accuracy of the prediction results. Among them, the performance indicator data includes:

[0051] Throughput: A measure of the amount of data a device can process per unit time, usually expressed in bits per second (bps) or packets per second (pps).

[0052] Latency: A measure of the time it takes for a data packet to travel from the sender to the receiver, usually measured in milliseconds (ms).

[0053] Packet loss rate: Measures the proportion of data packets lost during transmission, usually expressed as a percentage.

[0054] Environmental stress data:

[0055] Temperature: The temperature range in which the device can work normally, for example -20℃-70℃.

[0056] Vibration strength: The ability of the equipment to resist vibration and shock, usually needs to comply with relevant standards such as MIL-STD-810G.

[0057] Humidity: The humidity range in which the device can work normally, such as 5%-95% RH (relative humidity).

[0058] Hardware aging data:

[0059] Capacitor ESR value: The equivalent series resistance value of a capacitor. As the capacitor ages, the ESR value will gradually increase. You can use a special ESR tester to measure the ESR value of a capacitor.

[0060] Electromagnetic compatibility: The ability of the device to work stably in an electromagnetic environment must comply with relevant standards, such as FCC and CE.

[0061] As an optional implementation, different features may be combined to construct a new feature example, such as combining temperature and capacitor ESR value to reflect the impact of the comprehensive environment on the life of the device.

[0062] S2) Data preprocessing: Data cleaning of the collected time series data, including filling missing values ​​and abnormal data processing, to ensure data quality;

[0063] Furthermore, the method for filling missing data values ​​is to use linear interpolation or polynomial interpolation to fill missing values, thereby improving the continuity of data.

[0064] Furthermore, the method flow of the abnormal data processing includes:

[0065] Use statistical methods (such as z-score or ) to detect outliers in time series data;

[0066] Use smoothing methods to correct outliers or remove obviously erroneous outliers.

[0067] Furthermore, it also includes: standardizing the time series data, scaling the feature data of different dimensions to the same range, and preventing features with large values ​​from having too much impact on model training.

[0068] Specifically, missing value filling is to ensure data integrity, and abnormal data processing is to remove noise and erroneous data.

[0069] The z-score method converts the data into a distribution with a mean of 0 and a standard deviation of 1, which is applicable to the case where the data obeys a normal distribution. The remaining life of the broadband and fixed network equipment of the present invention and the various data obviously conform to a normal distribution.

[0070] S3) Feature selection: Select key features related to equipment life from the collected data types; compose the time series data set of key features;

[0071] The key features include at least three of throughput, latency, packet loss rate, temperature, vibration intensity, and capacitor ESR value.

[0072] Further, in step S3), the Pearson correlation coefficient is used to select key features related to the equipment life from the collected data types; the specific calculation steps include:

[0073] Calculate the remaining life of the equipment based on known data, denoted as Y;

[0074] Using the preprocessed time series data, the key features of each time point are paired with the corresponding remaining life value to form a vector data pair (X i , Y);

[0075] For each feature X i , where i = 1, 2, ..., n; calculate the Pearson correlation coefficient ρ between it and the remaining life Y; the calculation formula is:

[0076]

[0077] Where, X ij is the i-th eigenvalue of the j-th key feature; is the mean of the i-th feature; is Y i The remaining life value of the jth key feature; is the mean of the remaining life; j is the type of key feature;

[0078] Sort the Pearson correlation coefficients of all key features from large to small, and take the absolute value of the Pearson correlation coefficient;

[0079] A threshold is set and key features whose Pearson correlation coefficient is greater than the threshold are selected as the input of the model.

[0080] Specifically, the Pearson correlation coefficient can measure the linear correlation between features and remaining life. Its value range is between -1 and 1: 1 indicates a perfect positive correlation; -1 indicates a perfect negative correlation; 0 indicates no linear correlation. By calculating the Pearson correlation coefficient between each feature and the remaining life, and selecting key features with a large absolute value of the correlation coefficient, the features most relevant to the remaining life can be selected as training data for the model, thereby improving the efficiency and accuracy of the model.

[0081] Furthermore, the method for calculating the remaining life of the equipment based on known data is:

[0082] Based on the correlation between the performance index of the equipment and its life span, a performance threshold is defined. When the performance of the equipment is lower than the threshold, the equipment is considered to be failed.

[0083] Based on the known degradation time curve of the performance indicators of the equipment, the covariance value is used to calculate the remaining life of the equipment at a certain time point.

[0084] Furthermore, the threshold is set to 0.6-0.7.

[0085] S4) LSTM model construction and training: A long short-term memory network (LSTM) model is selected and a multi-layer LSTM architecture is constructed: the input layer is a time series data set with key features.

[0086] The LSTM layer contains multiple layers of LSTM units to capture long-term dependencies in time series data. The activation function of the LSTM unit is tanh or relu function. The Dropout rate is set to prevent overfitting.

[0087] The fully connected layer is used to map the output of the LSTM layer to the predicted value of the remaining life of the equipment; the activation function of the fully connected layer unit selects a linear function;

[0088] The output layer outputs the remaining life prediction results of the equipment;

[0089] Model training: Divide the time series data set into training set, validation set and test set, and use the training set data to train the LSTM model; optimize by adjusting the model's hyperparameters; use the mean square error (MSE) loss function and Adam optimizer to update the model parameters;

[0090] Specifically, the LSTM model is used to model the selected key features to predict the remaining life of the equipment. The LSTM network is good at processing time series data and can capture long-term dependencies in the data. Since a single-layer LSTM may encounter the problem of gradient vanishing or gradient exploding when processing longer sequence data, it is difficult to capture long-term dependencies.

[0091] Different types of broadband network equipment (such as routers, switches, firewalls, etc.) and different usage scenarios (such as high temperature, low temperature, high humidity, strong vibration, etc.) will cause the performance degradation mode and life characteristics of the equipment to vary greatly. The flexibility of multi-layer LSTM enables it to adapt to these different equipment types and usage scenarios, and build a more generalized model by learning the characteristics of different data sets.

[0092] The present invention uses a multi-layer LSTM architecture and stacks multiple LSTM layers to better learn and remember information with a longer time span, thereby more effectively capturing long-term dependencies in the data. This is crucial for predicting the remaining life of a device, because the life of a device is usually affected by various factors accumulated over a long period of time. In a multi-layer LSTM architecture, each layer of LSTM can learn feature representations at different levels. The bottom-level LSTM tends to learn local and short-term temporal patterns, while the high-level LSTM can learn more global and long-term features. This hierarchical feature extraction mechanism enables the multi-layer LSTM to capture information in the data more comprehensively, thereby improving prediction accuracy.

[0093] In the specific implementation process, when adding the first layer of LSTM units, you need to specify the key features ('input_shape'); when adding the intermediate LSTM units, you need to set 'return_sequences = True'; so that the sequence output is passed to the next layer of LSTM. Only the last layer of LSTM units needs to set 'return_sequences = False'; to output a single dimension (remaining life prediction value).

[0094] The multi-layer LSTM architecture can adapt to different equipment types and usage scenarios and improve the robustness of the model by increasing the depth and complexity of the model. This makes the multi-layer LSTM have a significant advantage in the task of predicting the remaining life of broadband and solid network equipment, and can provide more accurate and reliable prediction results, providing more effective support for equipment maintenance and management.

[0095] Furthermore, in the division of the time series data set, the time of the training set data is earlier than that of the validation set data, and the time of the validation set data is earlier than that of the test set data, so as to ensure the continuity in time.

[0096] Specifically, time series data is sample data with a fixed window step size. The data at consecutive time points is divided into a window as a sample. The window step size needs to be adjusted according to the actual situation, and is usually selected to capture the performance trend of the device.

[0097] The training set is used to train model parameters, the validation set is used to adjust model hyperparameters and prevent overfitting, and the test set is used to evaluate model performance.

[0098] The division ratio can be adjusted according to the amount of data, usually 70%:15%:15% or 80%:10%:10% can be adopted.

[0099] Furthermore, it also includes: using the validation set data to monitor the loss function value of the model, and terminating the training early when the loss function value no longer decreases within several rounds.

[0100] In the specific implementation process, the performance of the model on the validation set is monitored during model training, and training is stopped when the performance of the model no longer improves. This can avoid overfitting of the model on the training data, thereby improving the generalization ability of the model on the test data.

[0101] Furthermore, it also includes: using root mean square error (RMSE), mean absolute error (MAE) or R-squared to evaluate model performance.

[0102] Specifically, the root mean square error (RMSE): measures the square root of the deviation between the predicted value and the true value. RMSE is more sensitive to larger errors, so it better reflects the stability of the model prediction. The smaller the RMSE value, the better the model performance.

[0103] Mean Absolute Error (MAE): A measure of the average absolute value of the deviation between the predicted value and the true value. MAE is less sensitive to outliers and better reflects the average level of model predictions. The smaller the MAE value, the better the model performance.

[0104] squared(R 2 ): Measures the degree to which the model explains the variance of the data. 2 The value range of is [0,1]. The closer the value is to 1, the better the model fits the data. It should be noted that R 2 May be affected by the number of features, so R should be used with caution when comparing different models 2 Negative R 2 A value of 0 indicates that the model's predictive ability is very poor, even worse than a simple average prediction.

[0105] In the specific implementation process, multiple indicators can be combined to comprehensively evaluate the performance of the model, achieve a comprehensive evaluation of the performance of the LSTM model, and select appropriate model parameters for practical applications.

[0106] S5) LSTM model application:

[0107] New time series data is input into the LSTM model to predict the remaining life of the equipment and provide a decision-making basis for equipment maintenance and replacement.

[0108] Specifically, if the Pearson correlation coefficient feature selection method is used during the training phase, the same Pearson correlation coefficient feature selection method needs to be applied to the new data to ensure that the features of the input model are consistent.

[0109] Convert the selected key features’ time series data into the input format required by the LSTM model, usually a three-dimensional tensor (samples, timesteps, features). Samples represents the number of samples, timesteps represents the time step, and features represents the number of features.

[0110] The output of the LSTM model is usually a standardized value. It is necessary to perform an inverse transformation based on the standardization method in the training data to convert the predicted results into actual remaining life values.

[0111] In the specific implementation of maintenance and replacement decisions, maintenance and replacement strategies are formulated based on the predicted remaining life value combined with pre-defined thresholds. For example:

[0112] Maintenance: When the predicted remaining life is less than 50%, perform preventive maintenance such as replacing wearing parts, cleaning the radiator, etc.

[0113] Replacement: When the predicted remaining life is less than 10%, plan to replace the entire equipment to avoid production interruptions or safety accidents caused by equipment failure.

[0114] The present invention proposes a method for predicting the remaining life of Kuanggu network equipment based on artificial intelligence algorithm. By collecting the performance, environment and hardware aging data of the equipment in real time, the correlation between the data and the remaining life is established; by building a multi-layer LSTM architecture model, the remaining life of Kuanggu network equipment is predicted, and basic data is provided for formulating maintenance and replacement strategies. This improves equipment reliability and reduces operating costs.

[0115] In the specific implementation process provided in this application, it should be understood that the disclosed methods, processes and steps can be implemented in other ways. For example, the calculation method described above is only illustrative.

[0116] If the specific process is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk and other media that can store program code.

[0117] In the description of the specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0118] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. A method for predicting the remaining life of broadband and solid network equipment based on artificial intelligence algorithm, characterized by: The following steps are involved: S1) Data collection: collect the performance index data, environmental stress data and hardware aging data of the equipment in real time through system monitoring or the sensors built into the equipment; S2) Data preprocessing: Data cleaning of the collected time series data, including filling missing values ​​and abnormal data processing; S3) Feature selection: Select key features related to equipment life from the collected data types; A time series dataset that constitutes the key features; The key features include at least three of throughput, latency, packet loss rate, temperature, vibration intensity, and capacitor ESR value; S4) LSTM model construction and training: Select the long short-term memory network model and build a multi-layer LSTM architecture; S5) LSTM model application: Input new time series data into the LSTM model to predict the remaining life of the equipment.

2. According to claim 1, a method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm is characterized by: The performance indicator data specifically includes throughput, latency or packet loss rate; the environmental stress data specifically includes temperature or vibration intensity; and the hardware aging data specifically includes capacitor ESR value.

3. According to claim 1, a method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm is characterized by: The method for filling missing values ​​in the data is to use linear interpolation or polynomial interpolation to fill missing values.

4. According to claim 1, a method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm is characterized by: The method flow of abnormal data processing includes: Use the z-score method to detect outliers in time series data; Use smoothing methods to correct outliers or remove obviously erroneous outliers.

5. According to claim 1, a method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm is characterized by: In step S3), the key features related to the equipment life are selected from the collected data types using the Pearson correlation coefficient; The specific calculation steps include: Calculate the remaining life of the equipment based on known data, denoted as Y; Using the preprocessed time series data, the key features of each time point are paired with the corresponding remaining life value to form a vector data pair (X i , Y); For each feature X i , calculate the Pearson correlation coefficient ρ between it and the remaining life Y; the calculation formula is: Where, X ij is the feature value of the jth key feature at the i-th time point, where i = 1, 2, ..., n; is the mean of the feature at the i-th time point; is Y i The remaining life value of the jth key feature; is the mean of the remaining life; j is the type of key feature; Sort the Pearson correlation coefficients of all key features from large to small, and take the absolute value of the Pearson correlation coefficient; A threshold is set and key features whose Pearson correlation coefficient is greater than the threshold are selected as the input of the model.

6. The method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm according to claim 1 is characterized by: In step S4), the specific process of LSTM model construction and training is as follows: Build a multi-layer LSTM architecture: the input layer is a time series dataset of key features; The LSTM layer contains multiple layers of LSTM units to capture long-term dependencies in time series data. The activation function of the LSTM unit is tanh or relu function. The Dropout rate is set to prevent overfitting. The fully connected layer is used to map the output of the LSTM layer to the predicted value of the remaining life of the equipment; the activation function of the fully connected layer unit selects a linear function; The output layer outputs the remaining life prediction results of the equipment; Model training: Divide the time series data set into training set, validation set and test set, use the training set data to train the LSTM model; optimize by adjusting the model's hyperparameters; The mean square error loss function and Adam optimizer are used to update the model parameters.

7. The method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm according to claim 6 is characterized by: In the division of the time series data, the time of the training set data is earlier than that of the validation set data, and the time of the validation set data is earlier than that of the test set data.

8. The method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm according to claim 6 is characterized by: Also includes: Use the validation set data to monitor the loss function value of the model. If the loss function value no longer decreases within several rounds, terminate the training early.

9. The method for predicting the remaining life of broadband network equipment based on artificial intelligence algorithm according to claim 6 is characterized by: Also includes: Root mean square error, mean absolute error or R-squared is used to evaluate model performance.

Images