A channel state identification method and system for a heterogeneous network of an electricity consumption information collection system

Abstract

The present invention belongs to the technical field of power consumption information collection and communication, and provides a channel state identification method and system for a heterogeneous network of a power consumption information collection system, including: obtaining historical channel state data of the heterogeneous network of the power consumption information collection system; based on the obtained historical channel state data, analyzing channel characteristics, and constructing a multi-dimensional channel state evaluation index; according to the constructed multi-dimensional channel state evaluation index, obtaining real-time channel state data of the heterogeneous network of the power consumption information collection system under different channel scenarios; using a channel state identification model based on a density clustering algorithm-long short-term memory network to identify the acquired real-time channel state data, obtain a channel state category, and complete the identification of the channel state of the heterogeneous network of the power consumption information collection system.

Description

Technical Field

[0001] The present invention belongs to the technical field of power consumption information collection and communication, and in particular relates to a channel state identification method and system for a heterogeneous network of a power consumption information collection system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The electricity consumption information collection system mainly works together through high-speed power line carrier and high-speed wireless dual-mode network to realize the aggregation, transmission and interaction of electricity consumption information of low-voltage power users and energy consumption information of other energy-consuming customers. High-speed power line carrier uses existing power lines as signal transmission media, without the need to lay additional communication lines, and has a low construction cost. High-speed wireless provides greater flexibility, without the limitation of physical lines, and is easy to install and expand. However, due to communication channel congestion, communication channel attenuation and other reasons, the communication network is often not smooth, affecting the transmission of electricity consumption information data. In particular, with the large-scale promotion of high-speed power line carrier and high-speed wireless in the power grid, the monitoring and identification of the channel status of the heterogeneous network of the electricity consumption information collection system is becoming more and more important.

[0004] The monitoring and identification of the channel status of the heterogeneous network of the electricity consumption information collection system is divided into three stages: data collection, model construction and status identification; among them, in the data collection stage, it is necessary to collect sufficient network channel feature data for subsequent analysis; in the model construction stage, it is necessary to use various models to analyze the collected data in order to find the representation of the network channel status; in the status identification stage, it is necessary to classify and process the network channel data in order to realize the monitoring of abnormal conditions in the network channel.

[0005] However, the channel state recognition of the heterogeneous network of the existing power consumption information collection system has the following problems:

[0006] (1) Channel state representation has a single dimension

[0007] In the traditional channel state characterization system, the channel state is usually only characterized from the channel level itself (such as signal strength, signal-to-noise ratio, etc.); however, the dynamically changing network topology structure (such as network hierarchy) of complex networks will also have a great impact on the channel quality; therefore, the traditional method cannot fully reflect the complexity of the channel and the dynamic changes of the topology.

[0008] (2) The generalization ability of the channel state recognition model is poor

[0009] The current channel state recognition model is built based on a certain communication network. However, the current electricity consumption information collection system uses a dual-mode communication network (high-speed power line carrier and high-speed wireless), and its communication environment is complex. Summary of the invention

[0010] To solve the above problems, the present invention proposes a channel state identification method and system for a heterogeneous network of an electricity consumption information collection system, designs a multi-dimensional channel state evaluation index, constructs a channel state identification model based on a density clustering algorithm-long short-term memory network D-LSTM, identifies the collected channel data of the heterogeneous network of the electricity consumption information collection system, identifies the channel state of the heterogeneous network in real time, and completes efficient and stable identification of the heterogeneous network channels of the electricity consumption information collection system.

[0011] According to some embodiments, a first solution of the present invention provides a channel state identification method for a heterogeneous network of an electricity consumption information collection system, using the following technical solution:

[0012] A method for identifying channel status of a heterogeneous network of an electricity consumption information collection system, comprising:

[0013] Obtain historical channel status data of the heterogeneous network of the power consumption information collection system;

[0014] Based on the acquired historical channel status data, the channel characteristics are analyzed and multi-dimensional channel status evaluation indicators are constructed;

[0015] According to the constructed multi-dimensional channel status evaluation index, the real-time channel status data of the power consumption information collection system in different channel scenarios of the heterogeneous network is obtained;

[0016] The channel state recognition model based on density clustering algorithm-long short-term memory network is used to identify the real-time channel state data, obtain the channel state category, and complete the recognition of the heterogeneous network channel state of the power consumption information collection system.

[0017] As a further technical limitation, the constructed multi-dimensional channel state evaluation index at least considers channel characteristic indicators and channel characterization indicators; the channel characteristic indicators are the results of channel characteristic analysis, including normal channels, high interference channels, high attenuation channels and high congestion channels; the channel characterization indicators include uplink success rate, downlink success rate, network level, signal-to-noise ratio, channel attenuation, transmission delay and delay jitter.

[0018] Furthermore, in the process of obtaining real-time channel status data of the heterogeneous network of the electricity consumption information collection system in different channel scenarios according to the constructed multi-dimensional channel status evaluation indicators, the real-time channel status data acquisition schemes for different scenarios are determined according to the channel characteristic indicators, and the acquisition of real-time channel status data in different channel scenarios is completed based on the determined acquisition schemes.

[0019] As a further technical limitation, in the process of identifying the acquired real-time channel state data, a density clustering algorithm is used to cluster the channel state data to generate a channel state sample library including a channel state training set and a channel state verification set.

[0020] Furthermore, the process of clustering the channel state data using the density clustering algorithm is as follows: determining the optimal hyperparameters of the density clustering algorithm based on normalized mutual information, clustering the channel state data using the optimal hyperparameters, and evaluating the density clustering algorithm using the adjusted Rand index as a clustering effect measurement indicator.

[0021] Furthermore, the clustering results of the channel state data according to the density clustering algorithm are trained and verified by the long short-term memory network, and the clustering results are classified and recognized considering the time series changes according to the trained long short-term memory network to obtain the channel state category.

[0022] According to some embodiments, a second solution of the present invention provides a channel state identification system for a heterogeneous network of an electricity consumption information collection system, which adopts the following technical solution:

[0023] A channel state identification system for a heterogeneous network of an electricity consumption information collection system, comprising:

[0024] A first acquisition module configured to acquire historical channel status data of a heterogeneous network of an electricity consumption information collection system;

[0025] An analysis and construction module, which is configured to analyze channel characteristics based on the acquired historical channel state data and construct a multi-dimensional channel state evaluation index;

[0026] A second acquisition module is configured to acquire real-time channel status data in different channel scenarios of a heterogeneous network of a power consumption information collection system according to the constructed multi-dimensional channel status evaluation index;

[0027] The state recognition module is configured to use a channel state recognition model based on a density clustering algorithm-long short-term memory network to identify the acquired real-time channel state data, obtain the channel state category, and complete the recognition of the heterogeneous network channel state of the power consumption information collection system.

[0028] According to some embodiments, a third solution of the present invention provides a computer-readable storage medium, which adopts the following technical solution:

[0029] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps in a channel state identification method for a heterogeneous network of an electricity consumption information collection system as described in the first solution of the present invention.

[0030] According to some embodiments, a fourth solution of the present invention provides an electronic device, which adopts the following technical solution:

[0031] An electronic device comprises a memory, a processor and a program stored in the memory and running on the processor, wherein when the processor executes the program, the steps in a method for identifying the channel state of a heterogeneous network of an electricity consumption information collection system as described in the first solution of the present invention are implemented.

[0032] According to some embodiments, a fifth solution of the present invention provides a computer program product, which adopts the following technical solution:

[0033] A computer program product includes software codes, wherein the program in the software codes executes the steps in the channel state identification method of a heterogeneous network of an electricity consumption information collection system as described in the first solution of the present invention.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] The present invention constructs a multi-dimensional channel status evaluation index for comprehensively and accurately reflecting the channel status. By comprehensively considering multiple indicators such as the channel characteristics (such as signal attenuation, signal-to-noise ratio, etc.) and network topology characteristics (such as network level), it can not only evaluate the instantaneous performance of the channel, but also consider the influence of factors such as delay and network topology volatility, identify potential problems and bottlenecks in the channel, and more comprehensively and accurately reflect the channel status, thereby providing a scientific basis for network optimization and fault location.

[0036] The present invention constructs a channel state recognition model based on density clustering-long short-term memory network D-LSTM, which improves the accuracy and robustness of channel state recognition of heterogeneous networks of power consumption collection systems; after preprocessing the channel data, density clustering (Density-Based Spatial Clustering of Applications with Noise, referred to as DBSCAN) is used to remove noise and abnormal points through clustering, thereby reducing the interference of abnormal data on the training of long short-term memory network (Long Short-Term Memory, referred to as LSTM) classification model, ensuring that the LSTM classification model focuses on learning channel state characteristics and their change laws; the clustered data not only helps to reduce confusion, but also can effectively divide different types of channel states, so that the LSTM classification model can perform classification and recognition more accurately; the DBSCAN clustering model can adaptively identify complex channel change patterns, and the LSTM classification model reveals the long-term dependency behind the identified channel change patterns through time series analysis, thereby improving the accuracy of heterogeneous network channel monitoring and recognition of the power consumption information collection system and the generalization ability of the recognition model.

[0037] This embodiment designs a multi-dimensional channel state evaluation indicator, constructs a channel state recognition model based on the density clustering algorithm-long short-term memory network, identifies the collected channel data of the heterogeneous network of the power consumption information collection system, identifies the channel state of the heterogeneous network in real time, and completes the efficient and stable recognition of the heterogeneous network channels of the power consumption information collection system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings in the specification that constitute a part of this embodiment are used to provide a further understanding of this embodiment. The schematic embodiments of this embodiment and their descriptions are used to explain this embodiment and do not constitute improper limitations on this embodiment.

[0039] Figure 1 This is a flow chart of a method for identifying the channel state of a heterogeneous network of an electricity consumption information collection system in Embodiment 1 of the present invention;

[0040] Figure 2 This is an architecture diagram of the D-LSTM channel state recognition model in the first embodiment of the present invention;

[0041] Figure 3 This is a schematic diagram of the high-speed wireless channel state clustering effect in the first embodiment of the present invention;

[0042] Figure 4 This is a schematic diagram of the high-speed wireless channel state training set recognition result in the first embodiment of the present invention;

[0043] Figure 5 This is a schematic diagram of the identification result of the high-speed wireless channel state test set in the first embodiment of the present invention;

[0044] Figure 6 This is a schematic diagram of the high-speed power line carrier channel state clustering effect in the first embodiment of the present invention;

[0045] Figure 7 This is a schematic diagram of the classification prediction results of the high-speed power line carrier training set in the first embodiment of the present invention;

[0046] Figure 8 This is a schematic diagram of the classification prediction results of the high-speed power line carrier test set in the first embodiment of the present invention;

[0047] Fig. 9 This is a structural block diagram of a channel state identification system for a heterogeneous network of an electricity consumption information collection system in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0048] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0049] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0050] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0051] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0052] Embodiment 1

[0053] Embodiment 1 of the present invention introduces a channel state identification method for a heterogeneous network of an electricity consumption information collection system.

[0054] This embodiment designs multi-dimensional and multi-indicator channel state evaluation indicators from two levels: the channel itself and the network topology structure, aiming to comprehensively and accurately characterize the channel characteristics of the dual-mode communication network in the power consumption information collection system; based on the above-mentioned multi-dimensional and multi-indicator channel state evaluation indicators, the channel state data under different environments and conditions in the dual-mode communication network of the power consumption information collection system are collected to provide key data support for channel state identification; using the respective advantages of DBSCAN and LSTM, a D-LSTM channel state identification model is built to perform heterogeneous network channel state identification, realize real-time monitoring and identification of channel states, locate the causes of channel abnormalities, so as to achieve rapid recovery of abnormal channels, thereby providing a more intelligent and accurate channel optimization strategy.

[0055] like Figure 1 A method for identifying channel status of a heterogeneous network of an electricity consumption information collection system is shown, comprising:

[0056] Obtain historical channel status data of the heterogeneous network of the power consumption information collection system;

[0057] Based on the acquired historical channel status data, the channel characteristics are analyzed and multi-dimensional channel status evaluation indicators are constructed;

[0058] According to the constructed multi-dimensional channel status evaluation index, the real-time channel status data of the power consumption information collection system in different channel scenarios of the heterogeneous network is obtained;

[0059] The channel state recognition model based on density clustering algorithm-long short-term memory network is used to identify the real-time channel state data, obtain the channel state category, and complete the recognition of the heterogeneous network channel state of the power consumption information collection system.

[0060] As one or more implementations, a small amount of collected heterogeneous network channel state data is analyzed to obtain channels with different characteristics; the characteristics of the channels generally involve normal channels, high interference channels, high attenuation channels, and high congestion channels; specifically as follows:

[0061] (1) Normal channel

[0062] A channel in which signals can be transmitted with expected quality and rate in a normal channel, usually without significant attenuation, interference or congestion. The characteristics of a normal channel are that the signal can be transmitted stably and clearly from the sender to the receiver with a low bit error rate.

[0063] (2) High Interference Channel

[0064] In a high-interference channel, in addition to the useful signal, other signals or noise affect the transmitted signal. These interferences may come from other communication systems on the same or adjacent frequency, external electromagnetic interference sources, or even the noise of the device itself. Interference signals can cause distortion or attenuation of the received signal, thereby reducing the communication quality and signal-to-noise ratio.

[0065] (3) High attenuation channel

[0066] In a high attenuation channel, due to the influence of factors such as distance and medium, as the signal transmission distance increases or encounters obstacles, the signal power will be greatly attenuated, causing the received signal to be weaker than the original signal, thus affecting the communication quality.

[0067] (4) Highly congested channels

[0068] In highly congested channels, excessive data traffic or insufficient channel resources can lead to increased signal transmission delays, increased packet loss rates, and decreased communication quality. When multiple communication requests compete for limited bandwidth or channel capacity at the same time, the system cannot process all requests in a timely manner, causing congestion in the network or communication channel.

[0069] In order to fully characterize the performance of the channel, key indicators are selected to characterize the channel characteristics; indicators are designed from four types of channels: normal channels, high interference channels, high attenuation channels, and high congestion channels. In the power consumption information collection system, the preferred channel characterization indicators include: uplink success rate, downlink success rate, network level, signal-to-noise ratio, channel attenuation, transmission delay, and delay jitter.

[0070] (1) Uplink success rate

[0071] The uplink success rate refers to the ratio of successful data transmission from the terminal device (such as a meter or sensor) to the concentrator. The performance in different channel categories is:

[0072] Normal channel: high uplink success rate and stable data transmission.

[0073] High-interference channel: The uplink success rate is reduced, interference causes data loss or errors, and multiple retransmissions are required.

[0074] High attenuation channel: Signal attenuation causes a decrease in the uplink success rate, especially in long distances and due to obstacles.

[0075] Highly congested channels: When multiple devices share bandwidth, the uplink success rate decreases, the transmission delay increases, and the probability of successful transmission decreases.

[0076] (2) Downlink success rate

[0077] The downlink success rate refers to the ratio of successful data transmission from the concentrator to the terminal device, reflecting the reliability of the concentrator sending instructions or data to the terminal device.

[0078] The performance in different channel categories is as follows:

[0079] Normal channel: The downlink success rate is high and data can be transmitted quickly and reliably.

[0080] High-interference channel: Interference causes downlink data packets to be lost, reducing the success rate.

[0081] High attenuation channel: Signal attenuation affects the quality of downlink data, resulting in a decrease in the downlink success rate.

[0082] Highly congested channels: Congestion will increase the delay of downlink data transmission, which may cause packet loss or transmission failure, and reduce the downlink success rate.

[0083] (3) Network level

[0084] The network level refers to the number of relay stations that the signal passes through during transmission, and reflects the indicators at the network topology level. Especially in the high-speed power line carrier and high-speed wireless dual-mode network used in the power consumption information collection system, the network level will have a greater impact on the channel quality. The smaller the network level, the better the channel quality is generally.

[0085] Performance in different channel categories:

[0086] Normal channel: The network layer is small, the signal is transmitted directly or passes through fewer relay stations, and the channel quality is good.

[0087] High-interference channels: Interference may require more retransmissions and relays, increase network layers, and degrade channel quality.

[0088] High attenuation channels: Signal attenuation may require more relays to compensate for the attenuation, increase network layers, and reduce channel quality.

[0089] Highly congested channels: Channel congestion leads to more data queuing and relay transmission, increasing the network layer and reducing channel quality.

[0090] (4) Signal-to-noise ratio

[0091] The signal-to-noise ratio is an indicator that measures the ratio of signal to noise strength. The higher the signal-to-noise ratio, the better the signal quality and the higher the data transmission reliability.

[0092] Performance in different channel categories:

[0093] Normal channel: high signal-to-noise ratio, stable transmission, and low bit error rate.

[0094] High-interference channel: Interference increases noise, causing a decrease in the signal-to-noise ratio and impaired transmission quality.

[0095] High attenuation channel: As the signal attenuates, the signal-to-noise ratio decreases, especially at long distances or in environments with many obstacles.

[0096] Highly congested channels: The signal is shared by multiple devices, and the signal-to-noise ratio may decrease, resulting in increased interference and bit errors.

[0097] (5) Channel attenuation

[0098] Channel attenuation refers to the decrease in signal strength due to various factors (such as distance, obstacles, weather, etc.) during signal propagation.

[0099] Performance in different channel categories:

[0100] Normal channel: The signal attenuation is small and the signal can be transmitted stably.

[0101] High-interference channels: Interference may increase the signal attenuation effect, but mainly affects the transmission quality.

[0102] High attenuation channel: The signal attenuation is obvious, and the attenuation is serious in long-distance transmission or complex environments.

[0103] Highly congested channels: Congestion itself does not directly affect attenuation, but the signal contention caused by congestion may indirectly affect the stability and effective transmission of the signal.

[0104] (6) Transmission delay

[0105] Transmission delay is the time required for data to travel from the source node to the destination node, including factors such as propagation delay, processing delay, and queuing delay.

[0106] Performance in different channel categories:

[0107] Normal channel: low transmission delay and fast data transmission.

[0108] High-interference channel: Interference causes increased transmission delay, and errors and retransmissions increase the overall delay.

[0109] High attenuation channel: Signal attenuation increases transmission delay, especially over long distances and in complex environments.

[0110] Highly congested channels: Congestion increases queuing delays, and multiple devices occupying the channel at the same time causes longer data transmission delays.

[0111] (7) Delay jitter

[0112] Delay jitter is the standard deviation of data transmission delay.

[0113] Performance in different channel categories:

[0114] Normal channel: small delay jitter and stable data transmission.

[0115] High-interference channel: Interference causes unstable transmission and fluctuating transmission delay.

[0116] High attenuation channel: Attenuation increases transmission delay and delay jitter.

[0117] Highly congested channel: Congestion increases the queuing time in the channel, increases the transmission delay, and increases the delay jitter.

[0118] According to the channel category and the performance of each channel feature, the influence of the characteristics of different channel types on the index performance is obtained as shown in Table 1:

[0119] Table 1 Correspondence between channel types and channel characterization indicators

[0120] Channel type / characterization index Normal channel High Interference Channel High attenuation channel Highly congested channels Uplink success rate high Low Low Low Downlink success rate high Low Low Low Network level Low Moderation high Moderation Signal-to-Noise Ratio high Low Low Moderation Channel attenuation Low Moderation high Moderation Transmission delay Low Moderation Moderation high Delay Jitter Low Moderation Moderation high

[0121] According to Table 1, different types of channels can be distinguished based on channel characterization indicators, and channel state evaluation indicators as shown in Table 2 can be constructed:

[0122] Table 2 Channel state evaluation indicators

[0123]

[0124]

[0125] In order to fully obtain the characteristics of high-speed power line carrier and high-speed wireless in different channel states of the power consumption collection system, it is necessary to design data collection solutions under different channel state scenarios:

[0126] (1) Normal channel scenario data collection solution

[0127] Based on the normal working scenario of the power consumption collection system, the channel status indicators of high-speed power line carrier and high-speed wireless are collected respectively.

[0128] (2) Data collection solution for high-interference channel scenarios

[0129] In the normal operation scenario of the power consumption collection system, by adding a co-frequency interference source (preferably using a signal generator to generate a co-frequency interference signal), the high-speed power line carrier and high-speed wireless high-interference channel scenarios are simulated respectively, and then the channel status indicators of the high-speed power line carrier and high-speed wireless in the high-interference scenario are collected respectively.

[0130] (3) Data collection solution for high attenuation channel scenarios

[0131] In the normal working scenario of the power consumption collection system, the attenuator is coupled in the channel to simulate the high-speed power line carrier and high-speed wireless high channel attenuation scenarios respectively, and then the channel status indicators of the high-speed power line carrier and high-speed wireless in the high attenuation scenario are collected respectively.

[0132] (4) Data collection solution for high-congestion channel scenarios

[0133] In the normal working scenario of the electricity consumption collection system, by increasing the number of concurrency or reducing the channel bandwidth, the high-speed power line carrier and high-speed wireless high-congestion channel scenarios are simulated respectively, and then the channel status indicators of the high-speed power line carrier and high-speed wireless in the high-congestion scenario are collected respectively.

[0134] Normalization of collected data can help improve model training results, reduce calculation errors, and accelerate convergence. Available normalization methods include Min-Max normalization, Z-Score normalization, maximum absolute value normalization, and Log normalization. Since Min-Max normalization is simple to operate and can effectively avoid the impact of differences in feature scales on analysis results, the Min-Max normalization method is preferred in data preprocessing to ensure that all features are processed at the same scale to improve model training results.

[0135] The minimum-maximum normalization formula is as follows:

[0136]

[0137] Among them, x max , x min are the minimum and maximum values ​​in the data respectively, and x′ is the normalized data.

[0138] Considering that the DBSCAN model has the advantages of processing noise points, not requiring the number of clusters to be specified, and adapting to complex data structures, and the LSTM model can deeply reveal the long-term dependence between channel status and channel indicators through time series analysis. This embodiment combines the advantages of the DBSCAN model and the LSTM model to build Figure 2 The D-LSTM channel state recognition model shown.

[0139] The D-LSTM channel state recognition model includes the DBSCAN clustering model, the channel state sample library, the channel state training set and verification set, and the LSTM classification model. Finally, the optimized D-LSTM channel state recognition model is built. Specifically:

[0140] (1) DBSCAN clustering model

[0141] Since the DBSCAN model has the advantages of being able to handle noise points, not requiring the specification of the number of clusters, and adapting to complex data structures, DBSCAN is preferred as the model for high-speed power line carrier and high-speed wireless channel state clustering in order to better reveal the intrinsic structure of the data.

[0142] The DBSCAN clustering process is as follows:

[0143] ① Select part of the collected channel state data for DBSCAN clustering.

[0144] ②Use NMI as a measurement indicator to select the optimal hyperparameters of the DBSCAN clustering algorithm.

[0145] NMI (Normalized Mutual Information) is a commonly used evaluation metric in cluster analysis, which is used to measure the similarity between two data sets (usually the clustering results and the true labels or the results of different clustering algorithms). It is based on the concept of mutual information in information theory and aims to quantify the amount of information shared between two data sets.

[0146] The value of NMI is between 0 and 1. NMI = 1 means that the two clustering results are exactly the same, which means that the clustering results are highly consistent with the true labels or two different clustering results. NMI = 0 means that the two clustering results are completely independent, which means that there is no shared information between them.

[0147] ③ According to the ratio of the collected full channel state data to the partial channel state data, scale the DSBCAN hyperparameters proportionally.

[0148] ④ Perform DBSCAN clustering on the collected full channel state data, and use ARI as the clustering effect measurement indicator.

[0149] ARI (Adjusted Rand Index) is an indicator used to evaluate the quality of clustering, especially to measure the similarity between clustering results and true labels. Unlike the traditional Rand Index (RI), ARI adjusts the expected value of random clustering to eliminate the accidental consistency of clustering results, so it is more reliable.

[0150] The calculation formula of Rand Index is: in:

[0151] a means that two data points are in the same cluster and are also in the same class in the true label (True Positive);

[0152] b means that two data points are in the same cluster but belong to different categories in the true label (false positive, FalsePositive);

[0153] c means that two data points are in different clusters but belong to the same class in the true label (false negative, FalseNegative);

[0154] d means that the two data points are in different clusters and belong to different categories in the true label (True Negative);

[0155] E[RI] is the expected Rand index of random clustering;

[0156] max(RI) is the maximum Rand Index.

[0157] ARI=1 means that the clustering result is completely consistent with the true label, that is, the clustering is perfect. ARI=0 means that there is no consistency between the clustering result and the true label, or the clustering result and the true label are randomly consistent. ARI is a negative value, which means that the similarity between the clustering result and the true label is worse than random clustering, that is, the clustering result is completely inappropriate.

[0158] (2) Channel state sample library

[0159] According to the clustering results of the DBSCAN clustering model, the cluster labels of the collected full channel state data are marked. Then, the characteristics of the channel in terms of uplink success rate, downlink success rate, network layer, signal-to-noise ratio, channel attenuation, delay jitter and transmission delay are extracted to generate a channel state sample library. Preferably, the sample library records the case ID, cluster ID, channel state type, channel state characteristics, intra-cluster statistical information, etc. of each channel state sample.

[0160] (3) Channel state training set and validation set

[0161] The channel state sample library is divided into a training set and a test set. Preferably, 70% of the samples in the channel state sample library are used as the training set, and the remaining 30% of the samples are used as the test set.

[0162] (4) LSTM classification model

[0163] Long short-term memory network (LSTM) is a special recurrent neural network (RNN) that is particularly suitable for processing and predicting time series-based data. LSTM controls the flow of information through its unique gating mechanism (including input gate, forget gate, and output gate), thereby effectively solving the shortcomings of traditional RNN in the problem of long-term dependence. In the channel classification task, LSTM can use channel features (such as channel impulse response, amplitude and phase information, etc.) for classification. By learning the time series patterns of these features, LSTM can accurately identify and classify different channel states. Therefore, this proposal prefers to build an LSTM model for channel state recognition. It is mainly to build a hierarchical structure of the LSTM network, including:

[0164] (a) Input layer: The dimension of the input data, that is, the data features at each time step.

[0165] (b) LSTM layer: The core part that processes the temporal dependency of time series data. You can choose to use a single or multi-layer LSTM layer and add a dropout layer to prevent overfitting.

[0166] (c) Dense layer: maps the output of the LSTM layer to the target value (regression or classification problem).

[0167] (d) Output layer: According to the task type (regression or classification), set the number of output neurons and activation function.

[0168] (5) Optimized D-LSTM channel state recognition model

[0169] The LSTM model is trained and validated using the channel state training set and validation set, and cross-validation and hyperparameter tuning are used to optimize model performance. The main process is as follows:

[0170] ①Model training

[0171] Use the training set data to train the model, and use backpropagation to update the model parameters, such as batch size, learning rate, training rounds (epochs) and other hyperparameters. During the training process, LSTM will gradually adjust the weights to minimize the loss function.

[0172] ②Evaluation and adjustment

[0173] Use the test set data to evaluate the performance of the optimized D-LSTM channel state recognition model and view the prediction accuracy or error. For the channel state recognition task, you can use evaluation indicators such as accuracy and F1 score.

[0174] According to the evaluation results, adjust the model structure or hyperparameters, such as adjusting the number of neurons in the D-LSTM layer, adding a dropout layer, adjusting the learning rate, etc., until the model achieves better generalization performance.

[0175] This embodiment uses the trained D-LSTM model to classify and identify the newly collected high-speed power line carrier and high-speed wireless channel status data. The output can intuitively reflect the performance status of the channel, such as "normal channel", "high interference channel", "high attenuation channel", and "high congestion channel". Through the channel status output, network maintenance personnel can promptly identify and locate the cause of abnormal channels, take corresponding optimization measures or adjust resource allocation to ensure the stability and efficient operation of the dual-mode network of the power consumption information collection system.

[0176] Case Analysis - High-Speed ​​Wireless Channel State Identification

[0177] Based on the analysis of the small amount of high-speed wireless channel status data that has been collected, the channels are divided into four main categories:

[0178] Normal channel: good signal quality, high signal-to-noise ratio, stable communication, no obvious interference or attenuation, suitable for efficient data transmission.

[0179] High-interference channel: Affected by external interference (such as other wireless devices, noise, etc.), the signal quality decreases, resulting in poor communication performance.

[0180] High attenuation channel: The signal strength is attenuated due to distance or obstacles (such as buildings, walls, etc.), resulting in reduced signal reception quality, which may cause increased transmission delays or an increase in network layers.

[0181] Highly congested channel: Due to excessive network traffic or insufficient resources, the channel bandwidth utilization is low and the transmission delay increases.

[0182] Using the channel state evaluation indicators described in Table 2, uplink success rate, downlink success rate, network layer, signal-to-noise ratio, channel attenuation and transmission delay are the key features to describe the performance of high-speed wireless channels. They can comprehensively characterize normal channels, high-interference channels, high-attenuation channels and high-congestion channels.

[0183] We collected 4,000 pieces of high-speed wireless channel status data, including uplink success rate, downlink success rate, signal-to-noise ratio, channel attenuation, transmission delay, and network level. We used the Min-Max normalization method to preprocess the data to ensure that each feature contributes evenly to the model, thereby improving the training effect and convergence speed of the clustering and classification algorithms.

[0184] To ensure the best effect of DBSAN clustering algorithm, the optimal hyperparameter combination is selected by normalized mutual information (NMI).

[0185] (1) Select the hyperparameter space: For example, set different combinations of epsilon and MinPts values ​​and perform grid search or random search. Epsilon can be set to 0.1, 0.5, 1.0, 2.0, 5.0, etc., and MinPts can be set to 3, 5, 7, 10, etc.

[0186] (2) Clustering and calculation of NMI value: For each set of parameter combinations, DBSAN is used to perform clustering and calculate the NMI value between the clustering result and the true label (if known). If there is no true label, you can choose to indirectly evaluate the clustering effect through cross-validation or unsupervised methods.

[0187] (3) Select the optimal hyperparameters: Select the hyperparameter combination with the highest NMI value as the optimal hyperparameters. When the optimal hyperparameters are epsilon = 0.2 and minPts = 5, the maximum NMI value is obtained, NMI = 0.8956.

[0188] According to the ratio of partial data to all data, minPts is proportionally enlarged to obtain the optimal hyperparameters for all data.

[0189] The DBSAN algorithm is run with the optimal hyperparameters to perform cluster analysis on the 4000 high-speed wireless channel status data collected. Each data point (channel feature) will be assigned to a cluster. Each cluster represents a type of channel data, which may have similar characteristics (such as similar downlink success rate, signal-to-noise ratio, etc.). Under this hyperparameter, ARI = 0.92876, indicating that the clustering effect is good.

[0190] The clustering effect is visualized using the silhouette coefficient diagram. Figure 3 shown.

[0191] According to the results of DBSAN clustering, 4000 channel data are assigned to different clusters, and each data is marked with a corresponding classification label, which represents the different types of channel data; the 4000 channel state sample library generated by the DBSAN clustering model is used as a training set with 70% and 30% as a validation set.

[0192] Construct the hierarchical structure of the LSTM network, including:

[0193] Input layer: The dimension of the input data, that is, the data features at each time step.

[0194] LSTM layer: The core part, which processes the time dependency of time series data. You can choose to use a single or multi-layer LSTM, and add a dropout layer to prevent overfitting.

[0195] Dense layer: maps the output of the LSTM layer to the target value (regression or classification problem).

[0196] Output layer: According to the task type (regression or classification), set the number of output neurons and activation function.

[0197] The channel state training set and validation set data are used to train and validate the D-LSTM channel state recognition model respectively.

[0198] like Figure 4 and Figure 5 As shown in the figure, the D-LSTM channel state recognition model has an accuracy rate of 98.17% and 97.9% for high-speed wireless channel state recognition results in the training set and test set, respectively, and can successfully classify the channel state feature data into the corresponding clusters accurately. This shows that the model can effectively capture the complex relationship between the potential patterns and features in different channel state data, and at the same time proves that D-LSTM has good generalization ability and stability in prediction and classification tasks, and can adapt to different types of channel state data.

[0199] Use the trained D-LSTM channel state recognition model to classify and output the newly collected high-speed wireless channel state data.

[0200] Example Analysis 2 High-speed Power Line Carrier Channel State Identification

[0201] Based on the analysis of a small amount of high-speed power line carrier channel status data that has been collected, the channels are divided into three main categories:

[0202] Normal channel: good signal quality, high signal-to-noise ratio, stable communication, no obvious interference or attenuation, suitable for efficient data transmission.

[0203] High attenuation channel: The signal strength is attenuated due to distance or obstacles (such as buildings, walls, etc.), resulting in reduced signal reception quality, which may cause increased transmission delays or data loss.

[0204] Highly congested channels: Due to excessive network traffic or insufficient resources, the channel bandwidth utilization is low, transmission delay increases, and problems such as data packet loss or connection disconnection may occur.

[0205] Using the channel status evaluation indicators described in Table 2, uplink success rate, downlink success rate, delay jitter, signal-to-noise ratio, channel attenuation and transmission delay are the key features to describe the performance of high-speed power line carrier channels. They can fully characterize normal channels, high attenuation channels and high congestion channels.

[0206] The high-speed power line carrier channel status data is collected to obtain 3000 high-speed power line carrier channel status data. The channel status information includes: uplink success rate, downlink success rate, delay jitter, signal-to-noise ratio, channel attenuation, and transmission delay.

[0207] The Min-Max normalization method is used for data preprocessing to ensure that each feature contributes to the model in a balanced manner, thereby improving the training effect and convergence speed of the clustering and classification algorithms.

[0208] To ensure the best effect of DBSAN clustering algorithm, the optimal hyperparameter combination is selected by normalized mutual information (NMI).

[0209] (1) Select the hyperparameter space: For example, set different combinations of epsilon and MinPts values ​​and perform grid search or random search. Epsilon can be set to 0.1, 0.5, 1.0, 2.0, 5.0, etc., and MinPts can be set to 3, 5, 7, 10, etc.

[0210] (2) Clustering and calculation of NMI value: For each set of parameter combinations, DBSAN is used to perform clustering and calculate the NMI value between the clustering result and the true label (if known). If there is no true label, you can choose to indirectly evaluate the clustering effect through cross-validation or unsupervised methods.

[0211] (3) Select the optimal hyperparameters: Select the hyperparameter combination with the highest NMI value as the optimal hyperparameters. When the optimal hyperparameters are epsilon = 0.25 and minPts = 18, the maximum NMI value is obtained, NMI = 0.8829.

[0212] According to the ratio of partial data to all data, minPts is proportionally enlarged to obtain the optimal hyperparameters for all data. The DBSAN algorithm is run with the optimal hyperparameters to perform cluster analysis on the 3,000 high-speed power line carrier channel state data collected. Each data point (channel feature) will be assigned to a cluster. Each cluster represents a type of channel data and may have similar characteristics (such as similar downlink success rate, signal-to-noise ratio, etc.). Clustering under this hyperparameter, ARI = 0.93482, indicating that the clustering effect is good.

[0213] The clustering effect is visualized using the silhouette coefficient diagram. Figure 6 shown.

[0214] According to the results of DBSAN clustering, 3000 channel data are assigned to different clusters, and each data is labeled with a corresponding classification label, which represents the different types of channel data. The 3000 channel state sample library generated by the DBSAN clustering model is used as a training set with 70% and a validation set with 30%.

[0215] Construct the hierarchical structure of the LSTM network, including:

[0216] Input layer: The dimension of the input data, that is, the data features at each time step.

[0217] LSTM layer: The core part, which processes the time dependency of time series data. You can choose to use a single or multi-layer LSTM, and add a dropout layer to prevent overfitting.

[0218] Dense layer: maps the output of the LSTM layer to the target value (regression or classification problem).

[0219] Output layer: According to the task type (regression or classification), set the number of output neurons and activation function.

[0220] The channel state training set and validation set data are used to train and validate the D-LSTM channel state recognition model respectively.

[0221] like Figure 7 and Figure 8 As shown in the figure, the D-LSTM channel state recognition model has an accuracy rate of 98.95% and 98.7% for the high-speed power line carrier channel state recognition results in the training set and the test set, respectively, and successfully classifies the channel state feature data into the corresponding clusters. This shows that the model can effectively capture the complex relationship between the potential patterns and features in different channel state data, and at the same time proves that D-LSTM has good generalization ability and stability in prediction and classification tasks, and can adapt to different types of channel state data.

[0222] The trained D-LSTM channel state recognition model is used to classify and identify the newly collected high-speed power line carrier channel state data.

[0223] This embodiment constructs a multi-dimensional and multi-indicator channel status evaluation index based on the complexity of the heterogeneous network of the power consumption information collection system and the characteristics of the channel status; this evaluation index comprehensively and in real time reflects the channel's own status and network topology status. This evaluation index divides the channel status into: normal channel, high interference channel, high attenuation channel and high congestion channel. Based on the above channel type classification, the evaluation index uses seven key indicators, including uplink success rate, downlink success rate, network level, signal-to-noise ratio, channel attenuation, delay jitter and transmission delay, to characterize the channel status.

[0224] This embodiment makes full use of the advantages of DBSCAN and LSTM models, and effectively improves the accuracy and robustness of channel state recognition by integrating density clustering and time series modeling. Build a DBSCAN clustering model, cluster the channel state data, generate a channel state sample library, and then generate a channel state training set and verification set. Build an LSTM classification model, and use the channel state training set and verification set to train and verify the LSTM classification model, and build a D-LSTM channel state recognition model. The DBSCAN clustering model provides LSTM with clearer channel state distribution information, while the LSTM model further improves the classification ability of time series changes. The D-LSTM model combining the above two models will make channel state recognition more accurate and efficient.

[0225] In order to effectively identify and evaluate the channel state, this embodiment constructs a channel state clustering model based on DBSCAN. The model automatically identifies different channel state categories by analyzing multi-dimensional indicators in the channel state data (such as signal-to-noise ratio, transmission delay, channel attenuation, network level, etc.) and using the density clustering characteristics of DBSCAN. DBSCAN does not need to preset the number of clusters, and can automatically determine the cluster center according to the distribution of the data, and effectively identify different channel states.

[0226] This embodiment constructs a channel state classification model based on LSTM to accurately identify the channel state; by processing the time series data of the channel state, it can effectively capture the long-term dependence of channel performance and identify different channel states. The LSTM network solves the gradient vanishing problem of traditional neural networks when processing long sequence data through its unique gating mechanism, making it more stable and accurate when processing dynamically changing channel characteristics such as signal fluctuations and delays. This model can not only monitor and identify changes in channel states in real time, but also predict future channel state change trends, thereby providing reliable support for network optimization and fault warning.

[0227] Embodiment 2

[0228] Embodiment 2 of the present invention introduces a channel state identification system for a heterogeneous network of an electricity consumption information collection system.

[0229] like Fig. 9 A channel state identification system for a heterogeneous network of an electricity consumption information collection system is shown, comprising:

[0230] A first acquisition module configured to acquire historical channel status data of a heterogeneous network of an electricity consumption information collection system;

[0231] An analysis and construction module, which is configured to analyze channel characteristics based on the acquired historical channel state data and construct a multi-dimensional channel state evaluation index;

[0232] A second acquisition module is configured to acquire real-time channel status data in different channel scenarios of a heterogeneous network of a power consumption information collection system according to the constructed multi-dimensional channel status evaluation index;

[0233] The state recognition module is configured to use a channel state recognition model based on a density clustering algorithm-long short-term memory network to identify the acquired real-time channel state data, obtain the channel state category, and complete the recognition of the heterogeneous network channel state of the power consumption information collection system.

[0234] The detailed steps are the same as those of the channel state identification method for a heterogeneous network of a power consumption information collection system provided in Example 1, and will not be repeated here.

[0235] Embodiment 3

[0236] Embodiment 3 of the present invention provides a computer-readable storage medium.

[0237] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps in a method for identifying a channel state of a heterogeneous network of an electricity consumption information collection system as described in Embodiment 1 of the present invention.

[0238] The detailed steps are the same as those of the channel state identification method for a heterogeneous network of a power consumption information collection system provided in Example 1, and will not be repeated here.

[0239] Embodiment 4

[0240] A fourth embodiment of the present invention provides an electronic device.

[0241] An electronic device comprises a memory, a processor and a program stored in the memory and running on the processor, wherein when the processor executes the program, the steps in a method for identifying the channel state of a heterogeneous network of an electricity consumption information collection system as described in Embodiment 1 of the present invention are implemented.

[0242] The detailed steps are the same as those of the channel state identification method for a heterogeneous network of a power consumption information collection system provided in Example 1, and will not be repeated here.

[0243] Embodiment 5

[0244] Embodiment 5 of the present invention provides a computer program product.

[0245] A computer program product includes software codes, wherein the program in the software codes executes the steps in the channel state identification method of a heterogeneous network of a power consumption information collection system as described in the first embodiment of the present invention.

[0246] The detailed steps are the same as those of the channel state identification method for a heterogeneous network of a power consumption information collection system provided in Example 1, and will not be repeated here.

[0247] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0248] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0249] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0250] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0251] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0252] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0253] The above description is only a preferred embodiment of the present embodiment and is not intended to limit the present embodiment. For those skilled in the art, the present embodiment may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present embodiment shall be included in the protection scope of the present embodiment.

Claims

1. A method for identifying the channel state of a heterogeneous network of an electricity consumption information collection system, characterized in that: include: Obtain historical channel status data of the heterogeneous network of the power consumption information collection system; Based on the acquired historical channel status data, the channel characteristics are analyzed and multi-dimensional channel status evaluation indicators are constructed; According to the constructed multi-dimensional channel status evaluation index, the real-time channel status data of the power consumption information collection system in different channel scenarios of the heterogeneous network is obtained; The channel state recognition model based on density clustering algorithm-long short-term memory network is used to identify the real-time channel state data, obtain the channel state category, and complete the recognition of the heterogeneous network channel state of the power consumption information collection system.

2. A method for identifying channel status of a heterogeneous network of an electricity consumption information collection system as described in claim 1, characterized in that: The constructed multi-dimensional channel state evaluation index at least considers channel characteristic index and channel characterization index; the channel characteristic index is the result of channel characteristic analysis, including normal channel, high interference channel, high attenuation channel and high congestion channel; the channel characterization index includes uplink success rate, downlink success rate, network level, signal-to-noise ratio, channel attenuation, transmission delay and delay jitter.

3. A method for identifying channel status of a heterogeneous network of an electricity consumption information collection system as described in claim 2, characterized in that: In the process of obtaining real-time channel status data of the heterogeneous network of the electricity consumption information collection system in different channel scenarios according to the constructed multi-dimensional channel status evaluation indicators, the real-time channel status data acquisition schemes for different scenarios are determined according to the channel characteristic indicators, and the acquisition of real-time channel status data in different channel scenarios is completed based on the determined acquisition schemes.

4. The method for identifying the channel status of a heterogeneous network of an electricity consumption information collection system as claimed in claim 1, characterized in that: In the process of identifying the acquired real-time channel state data, a density clustering algorithm is used to cluster the channel state data to generate a channel state sample library including a channel state training set and a channel state verification set.

5. A method for identifying channel status of a heterogeneous network of an electricity consumption information collection system as described in claim 4, characterized in that: The process of clustering channel state data using the density clustering algorithm is as follows: determining the optimal hyperparameters of the density clustering algorithm based on normalized mutual information, clustering the channel state data using the optimal hyperparameters, and evaluating the density clustering algorithm using the adjusted Rand index as a clustering effect measurement indicator.

6. A method for identifying channel status of a heterogeneous network of an electricity consumption information collection system as claimed in claim 4, characterized in that: The clustering results of the channel state data according to the density clustering algorithm are trained and verified by the long short-term memory network. The clustering results are classified and recognized considering the time series changes according to the trained long short-term memory network to obtain the channel state category.

7. A channel state identification system for a heterogeneous network of an electricity consumption information collection system, characterized in that: include: A first acquisition module configured to acquire historical channel status data of a heterogeneous network of an electricity consumption information collection system; An analysis and construction module, which is configured to analyze channel characteristics based on the acquired historical channel state data and construct a multi-dimensional channel state evaluation index; A second acquisition module is configured to acquire real-time channel status data in different channel scenarios of a heterogeneous network of a power consumption information collection system according to the constructed multi-dimensional channel status evaluation index; The state recognition module is configured to use a channel state recognition model based on a density clustering algorithm-long short-term memory network to identify the acquired real-time channel state data, obtain the channel state category, and complete the recognition of the heterogeneous network channel state of the power consumption information collection system.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of a channel state identification method for a heterogeneous network of an electricity consumption information collection system as described in any one of claims 1 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the steps of a channel state identification method for a heterogeneous network of an electricity consumption information collection system as described in any one of claims 1-6 are implemented.

10. A computer program product comprising software code, characterized in that The program in the software code executes the steps of a channel state identification method for a heterogeneous network of an electricity consumption information collection system as described in any one of claims 1-6.

Images