RRU equipment detection method and device, equipment, storage medium and program product

Abstract

The invention provides an RRU equipment detection method and device, equipment, a storage medium and a program product, and the method comprises the steps: carrying out the dimension reduction and reconstruction processing of user measurement data of to-be-detected RRU equipment through employing a convolution variational auto-encoder, and obtaining the reconstructed user measurement data; determining a network statistical index of the RRU equipment, and performing enhancement processing on the network statistical index by adopting a Wasserstein generative adversarial network (WGAN) to obtain an enhanced network statistical index; associating the reconstructed user measurement data with the enhanced network statistical index based on a ridge regression algorithm to obtain time sequence data of the associated network statistical index; and inputting the time sequence data into a pre-constructed TCN prediction model to obtain predicted time sequence data corresponding to the associated network statistical indexes, and obtaining an anomaly detection result of the RRU equipment according to the time sequence data and the predicted time sequence data. According to the invention, the detection efficiency and accuracy of the RRU equipment are improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an RRU equipment detection method, device, equipment, storage medium and program product. Background Art

[0002] With the gradual popularization of 5G networks, multiple cells will be set up under one base station, and multiple remote radio units (RRU) will be set up under the indoor coverage cells to meet the growing terminal demand. In the operation and maintenance of 5G networks, it is usually necessary to perform abnormal detection on RRU devices. At present, the use of manual judgment to detect abnormalities in RRU devices is inefficient and the accuracy is difficult to guarantee. Summary of the invention

[0003] The present invention provides an RRU equipment detection method, device, equipment, storage medium and program product, which are used to solve the defects of the prior art that the abnormality detection of RRU equipment is performed by manual judgment, the efficiency is low and the accuracy is difficult to guarantee.

[0004] In a first aspect, the present invention provides an RRU device detection method, comprising: A convolutional variational autoencoder is used to perform dimension reduction and reconstruction processing on user measurement data of a remote radio unit (RRU) device to be detected, thereby obtaining reconstructed user measurement data; Determine the network statistical indicators of the RRU device, and enhance the network statistical indicators using a Wasserstein generative adversarial network (WGAN) to obtain enhanced network statistical indicators; Associating the reconstructed user measurement data with the enhanced network statistical indicators based on a ridge regression algorithm to obtain time series data of the associated network statistical indicators; Input the time series data into a pre-built time convolution network (TCN) prediction model to obtain predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and obtain an abnormality detection result of the RRU device according to the time series data and the predicted time series data; The TCN prediction model is trained based on the time series data of sample network statistical indicators associated with the reconstructed sample user measurement data.

[0005] In some embodiments, the associating the reconstructed user measurement data with the enhanced network statistical indicator based on the ridge regression algorithm to obtain the time series data of the associated network statistical indicator includes: Determining a regression function of the reconstructed user measurement data and the enhanced network statistical index based on a ridge regression algorithm; Calculating, according to the regression function, a correlation coefficient between the reconstructed user measurement data and the enhanced network statistical indicator; The associated network statistical index is determined according to the association coefficient, and the time series data of the associated network statistical index is obtained.

[0006] In some embodiments, obtaining the abnormality detection result of the RRU device according to the time series data and the predicted time series data includes: Determine the difference between the predicted time series data and the time series data, and obtain a corresponding anomaly score; An abnormality score threshold is determined, the abnormality score is compared with the abnormality score threshold, and it is determined whether the RRU device is abnormal, and if so, the abnormal time point of the RRU device is determined.

[0007] In some embodiments, the use of the Wasserstein generative adversarial network (WGAN) to enhance the network statistical indicators to obtain enhanced network statistical indicators includes: Inputting the network statistical indicator into a pre-built WGAN model to obtain a new network statistical indicator output by the WGAN model, wherein the new network statistical indicator is similar to the network statistical indicator; The new network statistical indicator is merged with the network statistical indicator to obtain the enhanced network statistical indicator; The WGAN model includes a generator and a discriminator, and the WGAN model is obtained by cross-iterative training of the generator and the discriminator based on sample network statistical indicators.

[0008] In some embodiments, the convolutional variational autoencoder is used to perform dimension reduction and reconstruction processing on the user measurement data of the remote radio unit RRU device to be detected to obtain the reconstructed user measurement data, including: Acquire user measurement data of a remote radio unit (RRU) device to be detected; The convolutional variational autoencoder is used to encode the user measurement data to obtain hidden features of the user measurement data, and the hidden features are decoded to obtain reconstructed user measurement data.

[0009] In some embodiments, the process of determining the TCN prediction model includes: Obtain sample user measurement data and sample network statistical indicators corresponding to the sample RRU device; Using a convolutional variational autoencoder, the sample user measurement data is subjected to dimensionality reduction and reconstruction processing to obtain reconstructed sample user measurement data; A Wasserstein generative adversarial network (WGAN) is used to enhance the statistical indicators of the sample network to obtain enhanced statistical indicators of the sample network; Associating the reconstructed sample user measurement data with the enhanced sample network statistical index based on a ridge regression algorithm to obtain time series data of the associated sample network statistical index in time period t and time series data of time period t+1; Input the time series data of the time period t into the pre-built initial TCN prediction model to obtain the predicted time series data of the time period t+1 output by the initial TCN prediction model; Calculate the loss function value based on the predicted time series data of the t+1 time period and the time series data of the t+1 time period; According to the loss function value, the parameters of the initial TCN prediction model are iteratively optimized to obtain the TCN prediction model.

[0010] In a second aspect, the present invention further provides an RRU device detection device, comprising: A reconstruction unit is used to use a convolutional variational autoencoder to perform dimension reduction and reconstruction processing on user measurement data of a remote radio unit RRU device to be detected, so as to obtain reconstructed user measurement data; An enhancement unit, used to determine a network statistical indicator of the RRU device, and enhance the network statistical indicator using a Wasserstein generative adversarial network (WGAN) to obtain an enhanced network statistical indicator; an associating unit, configured to associate the reconstructed user measurement data with the enhanced network statistical indicators based on a ridge regression algorithm to obtain time series data of the associated network statistical indicators; A prediction unit, configured to input the time series data into a pre-built time convolution network TCN prediction model, obtain predicted time series data corresponding to the associated network statistical indicator output by the TCN prediction model, and obtain an abnormality detection result of the RRU device according to the time series data and the predicted time series data; The TCN prediction model is trained based on the time series data of sample network statistical indicators associated with the reconstructed sample user measurement data.

[0011] In a third aspect, the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the RRU device detection method described above is implemented.

[0012] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described RRU device detection methods.

[0013] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned RRU device detection methods.

[0014] The RRU equipment detection method, apparatus, device, storage medium and program product provided by the present invention perform dimensionality reduction and reconstruction processing on user measurement data of the RRU equipment to be detected to obtain reconstructed user measurement data, determine network statistical indicators of the RRU equipment, enhance the network statistical indicators to obtain enhanced network statistical indicators, provide a comprehensive data basis for anomaly detection, associate the reconstructed user measurement data with the enhanced network statistical indicators based on the ridge regression algorithm to obtain time series data of the associated network statistical indicators, input the time series data into a pre-built TCN prediction model to obtain predicted time series data corresponding to the associated network statistical indicators, determine the difference according to the time series data and the predicted time series data, obtain the anomaly detection result of the RRU equipment, and improve the efficiency and accuracy of RRU equipment detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 It is a flowchart of the RRU device detection method provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of a process of performing dimensionality reduction and reconstruction processing on user measurement data using a convolutional variational autoencoder provided in an embodiment of the present invention; Figure 3 It is a structural schematic diagram of an RRU equipment detection device provided by an embodiment of the present invention; Figure 4 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0018] At present, the judgment of RRU device anomalies based on a single indicator or a single network-side statistical indicator is often not comprehensive enough, and the service effects and measurement data generated by different data such as registration, voice, and video are also quite different. In addition to statistical indicators at the device granularity, user-dimensional measurement perception data is also an important indicator for anomaly detection. How to combine the perception data measured by different users at different times and the statistical indicators of devices in different coverage areas to detect RRU devices is the focus of RRU equipment detection.

[0019] At present, due to the small number of abnormal data samples of RRU equipment, it is difficult to achieve reasonable evaluation of abnormalities by means of machine learning. Although the simple autoencoder (AE) can effectively obtain reconstructed sample data through dimensionality reduction, its overfitting problem is difficult to solve, and the abnormal samples generated by the adversarial network have poor accuracy and high errors.

[0020] To this end, the embodiments of the present invention provide an RRU equipment detection method, apparatus, device, storage medium and program product, which perform dimensionality reduction and reconstruction processing on user measurement data of the RRU equipment to be detected to obtain reconstructed user measurement data, determine network statistical indicators of the RRU equipment, enhance the network statistical indicators to obtain enhanced network statistical indicators, and provide a comprehensive data basis for anomaly detection. Based on the ridge regression algorithm, the reconstructed user measurement data is associated with the enhanced network statistical indicators to obtain time series data of the associated network statistical indicators, and the time series data is input into a pre-built TCN prediction model to obtain predicted time series data corresponding to the associated network statistical indicators. The difference is determined according to the time series data and the predicted time series data to obtain the anomaly detection result of the RRU equipment, thereby improving the efficiency and accuracy of RRU equipment detection.

[0021] Figure 1 The following is a flow chart of the RRU device detection method provided by the embodiment of the present invention. Figure 1 As shown, a RRU device detection method is provided, comprising the following steps: step 110, step 120, step 130 and step 140. The steps of the method flow are only a possible implementation of the present invention.

[0022] Step 110: Use a convolutional variational autoencoder to perform dimension reduction and reconstruction processing on the user measurement data of the remote radio unit RRU device to be detected to obtain reconstructed user measurement data.

[0023] Among them, the remote radio unit is a key device in the wireless communication network. It is usually located on or near the antenna tower of the base station and is connected to the base station control unit through optical fiber or other transmission media. The RRU device is used to perform modulation, amplification and transmission of radio frequency signals.

[0024] Optionally, the user measurement data includes but is not limited to: user registration cell ID, power of reference signal received by user equipment, amount of data that the user equipment needs to transmit, network resources occupied by the user equipment, modulation and coding scheme used by the user equipment in communication, user demodulation A / N result feedback, user rate and user delay.

[0025] Among them, the user demodulation A / N result feedback indicates the user equipment's confirmation or denial of the received data frame, which is the key information used to determine whether the data transmission is successful or not; the user rate refers to the rate at which the user equipment transmits or receives data within a specific time; the user delay refers to the time delay for the user equipment to send and / or receive data.

[0026] Among them, Convolutional Variational Autoencoder (CV-AE) is a deep learning model that combines Convolutional Neural Network (CNN) and Variational Autoencoder (VAE). CV-AE can map high-dimensional data to latent space and reconstruct the original data from the latent space.

[0027] Figure 2 A schematic diagram of a process for performing dimension reduction and reconstruction processing on user measurement data using a convolutional variational autoencoder provided in an embodiment of the present invention. Figure 2 As shown, in some embodiments, a convolutional variational autoencoder is used to perform dimension reduction and reconstruction processing on user measurement data of a remote radio unit RRU device to be detected to obtain reconstructed user measurement data, including: Acquire user measurement data of a remote radio unit (RRU) device to be detected; A convolutional variational autoencoder is used to encode the user measurement data to obtain hidden features of the user measurement data, and the hidden features are decoded to obtain reconstructed user measurement data.

[0028] It should be noted that if Figure 2As shown in the figure, the convolutional variational autoencoder includes an input layer (input), an encoder (Encoder), a hidden layer (latent), a decoder (Decoder) and an output layer (output), x represents user measurement data, q(z|x) represents the possibility of generating z from x, z represents a hidden feature, p(x|z) represents the possibility of generating x from z, and x' represents the reconstructed user measurement data.

[0029] Optionally, the encoder maps the user measurement data to a latent space, and the decoder converts the vectors of the latent space into reconstructed data similar to the user measurement data; the encoder encodes the user measurement data into a probability distribution for describing the position of each user measurement data in the latent space.

[0030] Optionally, an initial VC-AE model is constructed, and the initial VC-AE model is trained with the goal of minimizing the reconstruction error and the regularization error of the latent space, and the encoder parameters ϕ and the decoder parameters θ are optimized to obtain the VC-AE model.

[0031] Among them, the reconstruction error is used to describe the similarity between the reconstructed data and the original data; and the regularization error is used to constrain the distribution of the reconstructed data in the latent space, making it closer to the Gaussian distribution.

[0032] Optionally, the user measurement data x is input into the encoder, the mean m and variance v are calculated, and x is sampled according to the normal distribution N(μ(x),σ(x)), such as N(0,1), to obtain the sampled data s; the hidden feature z=m+vs is calculated and input into the decoder to obtain the reconstructed sample x', and the reconstruction error and regularization error are calculated.

[0033] It is understandable that VCAE uses a variational encoder to compress user measurement data into latent variables, and then solves the problem that the gradient cannot be derived through reparameterization.

[0034] Step 120: determine the network statistical indicators of the RRU device, and use the Wasserstein generative adversarial network (WGAN) to enhance the network statistical indicators to obtain enhanced network statistical indicators.

[0035] Among them, the network statistical indicators are used to evaluate the network performance of the RRU equipment.

[0036] Optionally, the network statistical indicators include, but are not limited to: the number of Radio Resource Control (RRC) connections, the number of occupied Physical Resource Blocks (PRBs), and traffic.

[0037] Among them, the Wasserstein Generative Adversarial Network (WGAN) introduced the Wasserstein distance as an indicator to measure the difference between the generator and the discriminator to solve the problems of training instability and mode collapse in traditional GAN.

[0038] Among them, Wasserstein distance is a more stable and optimizable distance metric. Wasserstein distance can more accurately measure the distance between two probability distributions, especially in high-dimensional space. By optimizing Wasserstein distance, WGAN can generate higher quality and more realistic data samples.

[0039] In some embodiments, the Wasserstein generative adversarial network (WGAN) is used to enhance the network statistical indicators to obtain enhanced network statistical indicators, including: Inputting the network statistical indicators into the pre-built WGAN model to obtain new network statistical indicators output by the WGAN model, the new network statistical indicators being similar to the network statistical indicators; The new network statistical indicators are integrated with the network statistical indicators to obtain enhanced network statistical indicators; Among them, the WGAN model includes a generator and a discriminator. The WGAN model is obtained by cross-iterative training of the generator and the discriminator based on sample network statistical indicators.

[0040] It should be noted that the generator generates false data based on random noise by learning the intrinsic distribution of real data, and the discriminator is used to judge whether the input data (real data and false data) is false data. The purpose of generator training is to make the generated data more realistic and make it difficult for the discriminator to predict whether the input data is false data, while the purpose of discriminator training is to make its own judgment more accurate.

[0041] Step 130: Correlate the reconstructed user measurement data with the enhanced network statistical indicators based on the ridge regression algorithm to obtain time series data of the correlated network statistical indicators.

[0042] It can be understood that by associating the reconstructed user measurement data with the enhanced network statistical indicators to obtain the time series data of the associated network statistical indicators, key data can be extracted and the processing of redundant data can be avoided, thereby improving the efficiency and accuracy of RRU equipment detection.

[0043] In some embodiments, the reconstructed user measurement data is associated with the enhanced network statistical indicators based on the ridge regression algorithm to obtain the time series data of the associated network statistical indicators, including: Based on the ridge regression algorithm, the regression function of the reconstructed user measurement data and the enhanced network statistical indicators is determined; According to the regression function, the correlation coefficient between the reconstructed user measurement data and the enhanced network statistical indicators is calculated; The associated network statistical indicators are determined according to the association coefficient, and the time series data of the associated network statistical indicators are obtained.

[0044] Among them, the ridge regression algorithm solves the problem of unstable parameter estimation in ordinary linear regression when there is multicollinearity by introducing a regularization parameter (i.e., ridge regression parameter).

[0045] Optionally, based on a ridge regression algorithm, a regression function of the reconstructed user measurement data and the enhanced network statistical index is determined to determine an optimal regression parameter.

[0046] Optionally, the regression function is calculated as follows: ; in, represents the regression function, represents the reconstructed user measurement data, Represents the enhanced network statistical indicators, represents the correlation coefficient between the reconstructed user measurement data and the enhanced network statistical indicators, represents the regression parameters.

[0047] Step 140: input the time series data into a pre-built time convolution network (TCN) prediction model to obtain predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and obtain the abnormality detection result of the RRU device based on the time series data and the predicted time series data.

[0048] Among them, the TCN prediction model is trained based on the time series data of sample network statistical indicators associated with the reconstructed sample user measurement data.

[0049] It can be understood that by using the TCN prediction model to detect RRU equipment anomalies, the accuracy, real-time and efficiency of the prediction can be improved, and automated and intelligent equipment status monitoring can be achieved, thereby significantly improving network operation and maintenance effects and equipment reliability.

[0050] In some embodiments, obtaining an abnormality detection result of the RRU device according to the time series data and the predicted time series data includes: Determine the difference between the predicted time series data and the time series data, and obtain the corresponding anomaly score; Determine an abnormal score threshold, compare the abnormal score with the abnormal score threshold, determine whether the RRU device is abnormal, and if so, determine the abnormal time point of the RRU device.

[0051] It is understandable that by analyzing the differences between time series data and predicted time series data and combining it with the anomaly score threshold for judgment, the anomaly detection capability and operation and maintenance efficiency of RRU equipment can be significantly improved, effectively ensuring the stable operation and service quality of network equipment.

[0052] In the embodiment of the present invention, by performing dimensionality reduction and reconstruction processing on the user measurement data of the RRU device to be detected, reconstructed user measurement data is obtained, the network statistical indicators of the RRU device are determined, and the network statistical indicators are enhanced to obtain enhanced network statistical indicators, thereby providing a comprehensive data basis for anomaly detection. Based on the ridge regression algorithm, the reconstructed user measurement data is associated with the enhanced network statistical indicators to obtain time series data of the associated network statistical indicators, and the time series data is input into a pre-built TCN prediction model to obtain predicted time series data corresponding to the associated network statistical indicators. The difference is determined based on the time series data and the predicted time series data to obtain the anomaly detection result of the RRU device, thereby improving the efficiency and accuracy of RRU device detection.

[0053] In some embodiments, the process of determining the TCN prediction model includes: Obtain sample user measurement data and sample network statistical indicators corresponding to the sample RRU device; A convolutional variational autoencoder is used to reduce the dimension and reconstruct the sample user measurement data to obtain the reconstructed sample user measurement data; The Wasserstein generative adversarial network (WGAN) is used to enhance the statistical indicators of the sample network to obtain the enhanced statistical indicators of the sample network. Based on the ridge regression algorithm, the reconstructed sample user measurement data is associated with the enhanced sample network statistical indicators to obtain the time series data of the associated sample network statistical indicators in the time period t and the time series data of the time period t+1; Input the time series data of time period t into the pre-built initial TCN prediction model to obtain the predicted time series data of time period t+1 output by the initial TCN prediction model; Calculate the loss function value based on the predicted time series data of the t+1 period and the time series data of the t+1 period; According to the loss function value, the parameters of the initial TCN prediction model are iteratively optimized to obtain the TCN prediction model.

[0054] Optionally, a dynamic time warping (DTW) algorithm is used to calculate the distance between the predicted time series data in the t+1 time period and the time series data in the t+1 time period. The calculation process is as follows: 1) Initialization: Define two sequences to be matched, A (prediction time series data of time period t+1) and B (time series data of time period t+1), and initialize two pointers to point to the starting positions of sequences A and B respectively; 2) Calculate the distance: Calculate the Euclidean distance between the elements pointed to by two pointers; 3) Move the pointer: According to the calculated distance result, move the pointer so that the elements of the two sequences are gradually aligned; 4) Repeat the calculation: Repeat steps 2) and 3) until both pointers reach the end of the sequence; 5) Calculate the total distance: Calculate the total distance between the two sequences, which is the loss function value.

[0055] The following describes an RRU device detection apparatus provided by an embodiment of the present invention. The apparatus described below and the RRU device detection method described above can refer to each other.

[0056] Figure 3 A schematic diagram of the structure of an RRU device detection device provided by an embodiment of the present invention is shown in FIG. Figure 3 As shown, the RRU equipment detection device 300 includes: The reconstruction unit 310 is used to use a convolutional variational autoencoder to perform dimension reduction and reconstruction processing on the user measurement data of the remote radio unit RRU device to be detected to obtain reconstructed user measurement data; An enhancement unit 320 is used to determine the network statistical indicators of the RRU device, and use the Wasserstein generative adversarial network WGAN to enhance the network statistical indicators to obtain enhanced network statistical indicators; An associating unit 330 is used to associate the reconstructed user measurement data with the enhanced network statistical indicators based on a ridge regression algorithm to obtain time series data of the associated network statistical indicators; The prediction unit 340 is used to input the time series data into a pre-built time convolution network TCN prediction model, obtain the predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and obtain the abnormality detection result of the RRU device according to the time series data and the predicted time series data; Among them, the TCN prediction model is trained based on the time series data of sample network statistical indicators associated with the reconstructed sample user measurement data.

[0057] Optionally, the reconstructed user measurement data is associated with the enhanced network statistical indicators based on a ridge regression algorithm to obtain time series data of the associated network statistical indicators, including: Based on the ridge regression algorithm, the regression function of the reconstructed user measurement data and the enhanced network statistical indicators is determined; According to the regression function, the correlation coefficient between the reconstructed user measurement data and the enhanced network statistical indicators is calculated; The associated network statistical indicators are determined according to the association coefficient, and the time series data of the associated network statistical indicators are obtained.

[0058] Optionally, obtaining an abnormality detection result of the RRU device according to the time series data and the predicted time series data includes: Determine the difference between the predicted time series data and the time series data, and obtain the corresponding anomaly score; Determine an abnormal score threshold, compare the abnormal score with the abnormal score threshold, determine whether the RRU device is abnormal, and if so, determine the abnormal time point of the RRU device.

[0059] Optionally, a Wasserstein generative adversarial network (WGAN) is used to enhance the network statistical indicators to obtain enhanced network statistical indicators, including: Inputting the network statistical indicators into the pre-built WGAN model to obtain new network statistical indicators output by the WGAN model, the new network statistical indicators being similar to the network statistical indicators; The new network statistical indicators are integrated with the network statistical indicators to obtain enhanced network statistical indicators; Among them, the WGAN model includes a generator and a discriminator. The WGAN model is obtained by cross-iterative training of the generator and the discriminator based on sample network statistical indicators.

[0060] Optionally, a convolutional variational autoencoder is used to perform dimension reduction and reconstruction processing on user measurement data of a remote radio unit RRU device to be detected, to obtain reconstructed user measurement data, including: Acquire user measurement data of a remote radio unit (RRU) device to be detected; A convolutional variational autoencoder is used to encode the user measurement data to obtain hidden features of the user measurement data, and the hidden features are decoded to obtain reconstructed user measurement data.

[0061] Optionally, the process of determining the TCN prediction model includes: Obtain sample user measurement data and sample network statistical indicators corresponding to the sample RRU device; A convolutional variational autoencoder is used to reduce the dimension and reconstruct the sample user measurement data to obtain the reconstructed sample user measurement data; The Wasserstein generative adversarial network (WGAN) is used to enhance the statistical indicators of the sample network to obtain the enhanced statistical indicators of the sample network. Based on the ridge regression algorithm, the reconstructed sample user measurement data is associated with the enhanced sample network statistical indicators to obtain the time series data of the associated sample network statistical indicators in the time period t and the time series data of the time period t+1; Input the time series data of time period t into the pre-built initial TCN prediction model to obtain the predicted time series data of time period t+1 output by the initial TCN prediction model; Calculate the loss function value based on the predicted time series data of the t+1 period and the time series data of the t+1 period; According to the loss function value, the parameters of the initial TCN prediction model are iteratively optimized to obtain the TCN prediction model.

[0062] It should be noted here that the RRU equipment detection device provided in the embodiment of the present invention can implement all the method steps implemented in the above-mentioned RRU equipment detection method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those of the method embodiment will not be described in detail here.

[0063] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 4 As shown, the electronic device may include: a processor (processor) 410 , a communication interface (Communications Interface) 420 , a memory (memory) 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other through the communication bus 440 . The processor 410 can call the logic instructions in the memory 430 to execute the RRU device detection method, which includes: using a convolutional variational autoencoder to reduce the dimension and reconstruct the user measurement data of the remote radio unit RRU device to be detected to obtain the reconstructed user measurement data; determining the network statistical indicators of the RRU device, and using the Wasserstein generative adversarial network WGAN to enhance the network statistical indicators to obtain the enhanced network statistical indicators; based on the ridge regression algorithm, the reconstructed user measurement data is associated with the enhanced network statistical indicators to obtain the time series data of the associated network statistical indicators; the time series data is input into a pre-built time convolution network TCN prediction model to obtain the predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and the abnormal detection result of the RRU device is obtained according to the time series data and the predicted time series data; wherein the TCN prediction model is trained based on the time series data of the sample network statistical indicators associated with the reconstructed sample user measurement data.

[0064] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the 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 for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. 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, etc. Various media that can store program codes.

[0065] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the RRU device detection method provided by the above methods, which includes: using a convolutional variational autoencoder to reduce the dimension and reconstruct the user measurement data of the remote radio frequency unit RRU device to be detected, so as to obtain the reconstructed user measurement data; determining the network statistical indicators of the RRU device, and using the Wasserstein generative adversarial network WGAN to enhance the network statistical indicators to obtain the enhanced network statistical indicators; based on the ridge regression algorithm, the reconstructed user measurement data is associated with the enhanced network statistical indicators to obtain the time series data of the associated network statistical indicators; the time series data is input into a pre-constructed time convolution network TCN prediction model to obtain the predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and the abnormal detection result of the RRU device is obtained according to the time series data and the predicted time series data; wherein the TCN prediction model is trained based on the time series data of the sample network statistical indicators associated with the reconstructed sample user measurement data.

[0066] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the processor executes the RRU device detection method provided by the above methods, the method comprising: using a convolutional variational autoencoder to reduce the dimension and reconstruct the user measurement data of the remote radio frequency unit RRU device to be detected, so as to obtain the reconstructed user measurement data; determining the network statistical indicators of the RRU device, and using the Wasserstein generative adversarial network WGAN to enhance the network statistical indicators, so as to obtain the enhanced network statistical indicators; associating the reconstructed user measurement data with the enhanced network statistical indicators based on the ridge regression algorithm, so as to obtain the time series data of the associated network statistical indicators; inputting the time series data into a pre-constructed time convolution network TCN prediction model, so as to obtain the predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and obtaining the abnormal detection result of the RRU device according to the time series data and the predicted time series data; wherein the TCN prediction model is trained based on the time series data of the sample network statistical indicators associated with the reconstructed sample user measurement data.

[0067] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0068] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A RRU equipment detection method, characterized in that: include: A convolutional variational autoencoder is used to perform dimension reduction and reconstruction processing on user measurement data of a remote radio unit (RRU) device to be detected, thereby obtaining reconstructed user measurement data; Determine the network statistical indicators of the RRU device, and use a Wasserstein generative adversarial network (WGAN) to enhance the network statistical indicators to obtain enhanced network statistical indicators; Associating the reconstructed user measurement data with the enhanced network statistical indicators based on a ridge regression algorithm to obtain time series data of the associated network statistical indicators; Input the time series data into a pre-built time convolution network (TCN) prediction model to obtain predicted time series data corresponding to the associated network statistical indicators output by the TCN prediction model, and obtain an abnormality detection result of the RRU device according to the time series data and the predicted time series data; The TCN prediction model is trained based on the time series data of sample network statistical indicators associated with the reconstructed sample user measurement data.

2. The RRU device detection method according to claim 1, characterized in that: The method of associating the reconstructed user measurement data with the enhanced network statistical indicators based on the ridge regression algorithm to obtain time series data of the associated network statistical indicators includes: Determining a regression function of the reconstructed user measurement data and the enhanced network statistical index based on a ridge regression algorithm; Calculating, according to the regression function, a correlation coefficient between the reconstructed user measurement data and the enhanced network statistical indicator; The associated network statistical index is determined according to the association coefficient, and the time series data of the associated network statistical index is obtained.

3. The RRU device detection method according to claim 1, characterized in that: The obtaining, according to the time series data and the predicted time series data, an abnormality detection result of the RRU device includes: Determine the difference between the predicted time series data and the time series data, and obtain a corresponding anomaly score; An abnormality score threshold is determined, the abnormality score is compared with the abnormality score threshold, and it is determined whether the RRU device is abnormal, and if so, the abnormal time point of the RRU device is determined.

4. The RRU device detection method according to claim 1, characterized in that: The Wasserstein generative adversarial network WGAN is used to enhance the network statistical indicators to obtain enhanced network statistical indicators, including: Inputting the network statistical indicator into a pre-built WGAN model to obtain a new network statistical indicator output by the WGAN model, wherein the new network statistical indicator is similar to the network statistical indicator; The new network statistical indicator is merged with the network statistical indicator to obtain the enhanced network statistical indicator; The WGAN model includes a generator and a discriminator, and the WGAN model is obtained by cross-iterative training of the generator and the discriminator based on sample network statistical indicators.

5. The RRU device detection method according to claim 1, characterized in that: The convolutional variational autoencoder is used to perform dimension reduction and reconstruction processing on the user measurement data of the remote radio unit RRU device to be detected to obtain the reconstructed user measurement data, including: Acquire user measurement data of a remote radio unit (RRU) device to be detected; The convolutional variational autoencoder is used to encode the user measurement data to obtain hidden features of the user measurement data, and the hidden features are decoded to obtain reconstructed user measurement data.

6. The RRU device detection method according to claim 1, characterized in that: The determination process of the TCN prediction model includes: Obtain sample user measurement data and sample network statistical indicators corresponding to the sample RRU device; Using a convolutional variational autoencoder, the sample user measurement data is subjected to dimensionality reduction and reconstruction processing to obtain reconstructed sample user measurement data; A Wasserstein generative adversarial network (WGAN) is used to enhance the statistical indicators of the sample network to obtain enhanced statistical indicators of the sample network; Associating the reconstructed sample user measurement data with the enhanced sample network statistical index based on a ridge regression algorithm to obtain time series data of the associated sample network statistical index in time period t and time series data of time period t+1; Input the time series data of the time period t into the pre-built initial TCN prediction model to obtain the predicted time series data of the time period t+1 output by the initial TCN prediction model; Calculate the loss function value based on the predicted time series data of the t+1 time period and the time series data of the t+1 time period; According to the loss function value, the parameters of the initial TCN prediction model are iteratively optimized to obtain the TCN prediction model.

7. An RRU equipment detection device, characterized in that: include: A reconstruction unit is used to use a convolutional variational autoencoder to perform dimension reduction and reconstruction processing on user measurement data of a remote radio unit RRU device to be detected, so as to obtain reconstructed user measurement data; An enhancement unit, used to determine a network statistical indicator of the RRU device, and enhance the network statistical indicator using a Wasserstein generative adversarial network (WGAN) to obtain an enhanced network statistical indicator; an associating unit, configured to associate the reconstructed user measurement data with the enhanced network statistical indicators based on a ridge regression algorithm to obtain time series data of the associated network statistical indicators; A prediction unit, configured to input the time series data into a pre-built time convolution network (TCN) prediction model, obtain predicted time series data corresponding to the associated network statistical indicator output by the TCN prediction model, and obtain an abnormality detection result of the RRU device according to the time series data and the predicted time series data; The TCN prediction model is trained based on the time series data of sample network statistical indicators associated with the reconstructed sample user measurement data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the RRU device detection method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the RRU device detection method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the RRU device detection method according to any one of claims 1 to 6 is implemented.

Images