Water vapor monitoring method and device, electronic equipment, storage medium and program product

Abstract

The application provides a water vapor monitoring method and device, electronic equipment, storage medium and program product, and belongs to the technical field of artificial intelligence. The method comprises the following steps: inputting preprocessed cloud layer track data into a cloud layer track prediction model to obtain a cloud layer track prediction result in a future period; determining a 5G-A base station information transmission chain in the future period based on the cloud layer track prediction result; issuing a cloud layer water vapor sensing instruction to a 5G-A base station in the 5G-A base station information transmission chain; and inputting cloud layer water vapor data differences returned by each 5G-A base station into a cloud layer water content change sensing model after receiving the cloud layer water vapor data differences to obtain a cloud layer water vapor anomaly result. The application fully utilizes the detection capability of the 5G-A base station information transmission chain for water vapor changes in the moving process of cloud systems, and realizes large-scale and continuous monitoring of water vapor dynamics in rapidly changing cloud systems.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a water vapor monitoring method and device, electronic equipment, storage medium and program product. BACKGROUND

[0002] The distribution and dynamic change of water vapor in the atmosphere are of great significance to weather forecasting, climate research and disaster warning. Cloud system, as an important part of the water cycle in the earth's atmospheric system, its formation, development and dissipation process is closely related to the transport and transformation of water vapor. However, despite the significant progress made in modern meteorology, the direct detection capability for the change of water vapor content in the moving process of cloud system is still very limited.

[0003] The traditional radio sonde can measure the atmospheric temperature and humidity conditions at each layer from the ground to the upper air, but its sampling frequency is low and the coverage is limited, which is difficult to meet the demand of large-scale continuous monitoring of the dynamic change of water vapor in the cloud system.

[0004] Therefore, how to large-scale and continuously monitor the change of water vapor in the cloud system has become a technical problem to be solved. SUMMARY

[0005] The present application provides a water vapor monitoring method, device, electronic equipment, storage medium and program product to solve the defect that the prior art cannot large-scale and continuously monitor the dynamic change of water vapor in the cloud system.

[0006] The present application provides a water vapor monitoring method, comprising the following steps: inputting the preprocessed cloud layer trajectory data into a cloud layer trajectory prediction model to obtain a cloud layer trajectory prediction result in a future period output by the cloud layer trajectory prediction model; the cloud layer trajectory prediction model is used to predict cloud layer trajectory data in a future period based on given historical cloud layer trajectory data; determining a 5G-A base station information transmission chain in the future period based on the cloud layer trajectory prediction result; the 5G-A base station information transmission chain includes a plurality of 5G-A base stations arranged in sequence; issuing a cloud layer water vapor sensing instruction to the 5G-A base stations in the 5G-A base station information transmission chain; the cloud layer water vapor sensing instruction is used to instruct an i-th 5G-A base station to transmit the sensed cloud layer water vapor data to an i+1-th 5G-A base station, and instruct the i+1-th 5G-A base station to calculate a cloud layer water vapor data difference value between the i-th 5G-A base station and the i+1-th 5G-A base station, the cloud layer water vapor data difference value is a difference value of cloud layer water vapor data, and i is a positive integer; After receiving the cloud water vapor data difference values returned by each of the 5G-A base stations, the cloud water vapor data difference values are input into a cloud water content change perception model to obtain cloud water vapor anomaly results output by the cloud water content change perception model; the cloud water content change perception model is used to detect cloud water vapor anomaly data based on changes in cloud water content change cloud water vapor data difference values.

[0007] According to the water vapor monitoring method provided by the application, the cloud trajectory prediction model comprises an input layer, a plurality of hidden layers and an output layer connected in sequence; The input layer comprises n neurons for receiving cloud trajectory data in the past n time periods; The plurality of hidden layers comprise eight long short-term memory networks connected in sequence, and each long short-term memory network comprises an LSTM layer and a dropout layer connected in sequence; The output layer comprises m neurons for outputting cloud trajectory data in the future m time periods.

[0008] According to the water vapor monitoring method provided by the application, the 5G-A base station information transmission chain in the future time period is determined based on the cloud trajectory prediction result, comprising: Based on the cloud trajectory prediction result, the 5G-A base station corresponding to the cloud trajectory in the future time period is determined; The 5G-A base station is input into a 5G-A base station information transmission chain generation model to obtain a 5G-A base station information transmission chain output by the 5G-A base station information transmission chain generation model; The 5G-A base station information transmission chain generation model comprises a 5G-A base station encoder and a 5G-A base station information transmission chain generator: The 5G-A base station encoder is used to map each 5G-A base station to a specified dimension to obtain a latent feature representation of each 5G-A base station; The 5G-A base station information transmission chain generator is used to predict the potential information transmission relationship of each 5G-A base station based on the latent feature representation of each 5G-A base station, and generate a 5G-A base station information transmission chain.

[0009] According to the water vapor monitoring method provided by the application, the 5G-A base station encoder comprises two convolutional networks connected in sequence, and each convolutional network comprises a relational graph convolutional layer and a random discard layer connected in sequence.

[0010] According to the water vapor monitoring method provided by the application, the 5G-A base station information transmission chain generator predicts the potential information transmission relationship of any two 5G-A base stations based on the following steps: For any two 5G-A base stations, a score of existence of a specified association relationship between the two 5G-A base stations is calculated using a DistMult scoring function; The score is converted into a probability value using an activation function; If the probability value is greater than a preset threshold, it is determined that the specified association relationship exists between the two 5G-A base stations.

[0011] According to the cloud water content change perception model, the cloud water vapor data difference value is compressed and dimensionally reduced to obtain a feature vector of the cloud water vapor data difference value. The cloud water vapor anomaly result is obtained based on the following steps: If the reconstruction error of a 5G-A base station is greater than a reconstruction error threshold, it is determined that the cloud water vapor data of the 5G-A base station is abnormal; the reconstruction error threshold is the error between the cloud water vapor data difference value of the 5G-A base station and its previous 5G-A base station and the reconstruction difference value corresponding to the cloud water vapor data difference value.

[0012] The application further provides a water vapor monitoring device, comprising the following modules: The trajectory prediction module is configured to input the preprocessed cloud layer trajectory data into a cloud layer trajectory prediction model to obtain a cloud layer trajectory prediction result in a future period output by the cloud layer trajectory prediction model; the cloud layer trajectory prediction model is configured to predict cloud layer trajectory data in a future period based on given historical cloud layer trajectory data; The transmission chain generation module is configured to determine a 5G-A base station information transmission chain in a future period based on the cloud layer trajectory prediction result; the 5G-A base station information transmission chain comprises a plurality of 5G-A base stations arranged in sequence; The instruction issuing module is configured to issue a cloud water vapor perception instruction to a 5G-A base station in the 5G-A base station information transmission chain; the cloud water vapor perception instruction is configured to instruct an i-th 5G-A base station to transmit perceived cloud water vapor data to an i+1-th 5G-A base station and instruct the i+1-th 5G-A base station to calculate a cloud water vapor data difference value between the i-th 5G-A base station and the i+1-th 5G-A base station, the cloud water vapor data difference value being a difference value of cloud water vapor data, i being a positive integer; An anomaly prediction module is configured to: after receiving the cloud water vapor data difference returned by each 5G-A base station, input each cloud water vapor data difference into a cloud water content change perception model to obtain a cloud water vapor anomaly result output by the cloud water content change perception model; and the cloud water content change perception model is configured to detect cloud water vapor anomaly data based on the change of the cloud water content change cloud water vapor data difference.

[0013] The application further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the water vapor monitoring method according to any one of the above when executing the computer program.

[0014] The application further provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the water vapor monitoring method according to any one of the above.

[0015] The application further provides a computer program product, which includes a computer program, and the computer program is executable on a processor to implement the water vapor monitoring method according to any one of the above.

[0016] The water vapor monitoring method, device, electronic equipment, storage medium and program product provided by the application input the preprocessed cloud layer track data into a cloud layer track prediction model to obtain cloud layer track prediction results in a future period output by the cloud layer track prediction model; the cloud layer track prediction model is used to predict cloud layer track data in a future period based on given historical cloud layer track data; based on the cloud layer track prediction results, a 5G-A base station information transmission chain in the future period is determined; the 5G-A base station information transmission chain includes a plurality of 5G-A base stations arranged in sequence; cloud layer water vapor sensing instructions are issued to the 5G-A base stations in the 5G-A base station information transmission chain; the cloud layer water vapor sensing instructions are used to instruct an i-th 5G-A base station to transmit sensed cloud layer water vapor data to an i+1-th 5G-A base station, and instruct the i+1-th 5G-A base station to calculate a cloud layer water vapor data difference value between the i-th 5G-A base station and the i+1-th 5G-A base station, the cloud layer water vapor data difference value is a difference value of cloud layer water vapor data, and i is a positive integer; after receiving the cloud layer water vapor data difference values returned by the 5G-A base stations, the cloud layer water vapor data difference values are input into a cloud layer water content change sensing model to obtain cloud layer water vapor anomaly results output by the cloud layer water content change sensing model; the cloud layer water content change sensing model is used to detect cloud layer water vapor anomaly data based on changes in cloud layer water content change cloud layer water vapor data difference values. The application uses 5G-A base station multi-hop relay cooperation sensing under cloud system moving track to detect water vapor changes in the moving process of the cloud system, fully utilizes the detection capability of the 5G-A base station information transmission chain for water vapor changes in the moving process of the cloud system, realizes large-scale and continuous monitoring of water vapor dynamics in the fast-changing cloud system, and provides reliable observation support for disaster prevention and reduction and emergency disposal. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0018] Figure 1 is one of the flowcharts of the water vapor monitoring method provided by the application; Figure 2 is the second flowchart of the water vapor monitoring method provided by the application; Figure 3 is a structural schematic diagram of the cloud layer track prediction model provided by the application; Figure 4 is a structural schematic diagram of the 5G-A base station information transmission chain generation model provided by the application; Figure 5is a structural schematic diagram of a cloud water content change sensing model provided by the present application. Figure 6 is a structural schematic diagram of a water vapor monitoring device provided by the present application.

[0019] Figure 7 is a structural schematic diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0020] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described below in connection with the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0021] It should be noted that, in the description of the embodiments of the present application, the terms "comprise", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitation, the element defined by the statement "comprising a" does not exclude the presence of another identical element in the process, method, article or device comprising the element. The terms "upper", "lower" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise specified and limited, the terms "mount", "connect", "connect" should be understood broadly, for example, it can be a fixed connection, or it can be a detachable connection, or an integral connection; it can be a mechanical connection, or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the internal communication of two elements. For those of ordinary skill in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0022] The terms "first", "second", and the like in the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally a class and do not limit the number of objects, for example, the first object can be one or more. In addition, "and / or" means at least one of the connected objects, and " / " generally means that the front and rear associated objects are in an "or" relationship.

[0023] Figure 1 is one of the flow diagrams of the water vapor monitoring method provided by the present application, as shown in Figure 1 the method comprises the following: S110, input the preprocessed cloud layer track data into a cloud layer track prediction model to obtain a cloud layer track prediction result in a future period output by the cloud layer track prediction model; the cloud layer track prediction model is used to predict cloud layer track data in a future period based on given historical cloud layer track data; S120, based on the cloud layer track prediction result, determine a 5G-A base station information transmission chain in the future period; the 5G-A base station information transmission chain comprises a plurality of 5G-A base stations arranged in sequence; S130, issue a cloud layer water vapor sensing instruction to the 5G-A base stations in the 5G-A base station information transmission chain; the cloud layer water vapor sensing instruction is used to instruct an i-th 5G-A base station to transmit sensed cloud layer water vapor data to an i+1-th 5G-A base station, and instruct the i+1-th 5G-A base station to calculate a cloud layer water vapor data difference value between the i-th 5G-A base station and the i+1-th 5G-A base station, the cloud layer water vapor data difference value is a difference value of cloud layer water vapor data, and i is a positive integer; S140, after receiving the cloud layer water vapor data difference values returned by each of the 5G-A base stations, input each of the cloud layer water vapor data difference values into a cloud layer water content change sensing model to obtain a cloud layer water vapor anomaly result output by the cloud layer water content change sensing model; the cloud layer water content change sensing model is used to detect cloud layer water vapor anomaly data based on the change of the cloud layer water content change cloud layer water vapor data difference value.

[0024] It should be noted that the execution subject of the water vapor monitoring method provided by the embodiment of the present application can be a water vapor monitoring device, which can include but is not limited to: a server, a computer device, such as a desktop computer, a notebook computer, a tablet computer and the like. The execution subject of the water vapor monitoring method can also be a water vapor monitoring system, which belongs to the water vapor monitoring device.

[0025] 5G-A base station, i.e. 5G advanced base station. The 5G-A base station contains two air interfaces, i.e. a Uu interface for establishing communication with a mobile phone and a GNSS (Global Navigation Satellite System) interface for establishing communication with a satellite. The ground-based GNSS has the comprehensive advantages of low cost, high precision and all-weather compared with the traditional water vapor content detection radar. The satellite transmits wireless signals of two frequency bands, and the water vapor content in the atmosphere is inversely calculated according to the satellite signal delay measured by the ground-based satellite receiver. The 5G-A base station already has the GNSS receiving capability, and is currently only used for time synchronization, without calculating the water vapor content.

[0026] In the embodiment of the application, the cloud layer trajectory data is the latitude and longitude data of the cloud layer movement trajectory.

[0027] Before S110, the cloud layer movement trajectory data in the historical n time periods is obtained from the passive Internet of Things management platform, and after preprocessing, the preprocessed cloud layer trajectory data is obtained. The passive Internet of Things management platform serves as a bridge connecting the bottom layer hardware and the upper layer business application, and is mainly responsible for the management of the passive Internet of Things devices and the aggregation processing of terminal data, including the registration, configuration, state monitoring and maintenance management of various devices; at the same time, the passive Internet of Things management platform is responsible for data cleaning, format unification and standardized processing, and provides a reliable data basis for the subsequent application platform data analysis.

[0028] Optionally, the cloud layer movement trajectory data in the historical n time periods is preprocessed based on the following steps: For the cloud layer movement trajectory data {x t-n+1 ,...x t-2 ,x t-1 ,x t} in the historical n time periods, normalization processing is performed; normalization refers to scaling the data in proportion so as to fall into a small specific interval, i.e. scaling the data between a given minimum value and a maximum value, usually between 0 and 1. Since the LSTM (Long Short-Term Memory) is relatively sensitive to the size of the input data, it is necessary to uniformly map the data to the range of [0, 1]. After normalization, the convergence speed of the model will be improved, and the precision of the model will be improved. The normalization formula is as follows: X norm =(X-X min ) / (X max -X min ); Secondly, the data set is divided: the data set is divided into a training data set and a test data set according to a certain proportion, the model is trained with the training data set, and the performance of the model is verified with the test data set. For example, 80% of the total data set is used as the training data set, and the remaining 20% is used as the test data set; Finally, shape conversion is performed on the data: since the LSTM neural network requires the shape of the input data to be a 3-dimensional array, the data needs to be converted from a 2-dimensional array [samples, features] to a 3-dimensional array [samples, timesteps, features] to predict the cloud layer trajectory data in the future m time periods according to the cloud layer trajectory data in the recent n time periods. timesteps = n, timesteps is the number of successive input data that the LSTM considers to be associated with each input data; features = 1, that is, only one feature attribute is involved in the embodiment of the present application.

[0029] During offline training, the recent time series of several lengths are used to predict the time series of future several lengths, and the size of the window can be adjusted according to the problem. For example, the input window = 3 and the output window = 2 are set, that is, the values of the current time point (t) and the previous two time points (t-1) and (t-2) are used to predict the values of the future two time points (t+1) and (t+2). For ease of understanding, the following embodiments take the input window = n and the output window = m as examples to describe the water vapor monitoring method provided by the present application.

[0030] In S110, the preprocessed cloud layer trajectory data in the recent n time periods is input into the cloud layer trajectory prediction model, and the cloud layer trajectory prediction model outputs the cloud layer trajectory prediction result, that is, the prediction result of the cloud layer trajectory in the future m time periods.

[0031] In S120, the predicted cloud layer trajectory in the future m time periods is input into the 5G-A passive Internet of Things management and control platform, and the 5G-A passive Internet of Things management and control platform outputs the 5G-A base station corresponding to the cloud layer trajectory in the future m time periods; the 5G-A base station corresponding to the cloud layer trajectory in the future m time periods is input into the 5G-A base station information transmission chain generation model, and the 5G-A base station information transmission chain generation model generates the 5G-A base station information transmission chain corresponding to the cloud layer trajectory in the future m time periods by using the relational graph convolutional neural network.

[0032] In S130, in the 5G-A base station information transmission chain, the base station i transmits the sensed cloud layer water vapor data to the next hop base station i+1, and the base station i+1 calculates the difference value between the sensed cloud layer water vapor data and the cloud layer water vapor data sensed by the previous hop base station i; each base station in the 5G-A base station information transmission chain reports the difference value result to the 5G-A passive Internet of Things management and control platform.

[0033] In S140, the 5G-A passive Internet of Things management and control platform inputs the cloud layer water vapor data difference perceived by each base station in the 5G-A base station information transmission chain into a cloud layer water content change perception model, the cloud layer water content change perception model learns the change of the cloud layer water vapor data difference by using a self-encoding neural network, and finally outputs a cloud layer water content change perception result, that is, cloud layer water vapor abnormal data, to provide observation data support for emergency disposal.

[0034] The water vapor monitoring method provided by the embodiment of the application inputs the preprocessed cloud layer trajectory data into a cloud layer trajectory prediction model to obtain cloud layer trajectory prediction results in a future period output by the cloud layer trajectory prediction model; the cloud layer trajectory prediction model is used to predict cloud layer trajectory data in a future period based on given historical cloud layer trajectory data; based on the cloud layer trajectory prediction results, a 5G-A base station information transmission chain in the future period is determined; the 5G-A base station information transmission chain includes a plurality of 5G-A base stations arranged in sequence; a cloud layer water vapor perception instruction is issued to the 5G-A base stations in the 5G-A base station information transmission chain; the cloud layer water vapor perception instruction is used to instruct an i-th 5G-A base station to transmit perceived cloud layer water vapor data to an i+1-th 5G-A base station, and instruct the i+1-th 5G-A base station to calculate a cloud layer water vapor data difference between the i-th 5G-A base station and the i+1-th 5G-A base station, the cloud layer water vapor data difference being a difference of cloud layer water vapor data, and i being a positive integer; after receiving the cloud layer water vapor data difference returned by each 5G-A base station, each cloud layer water vapor data difference is input into a cloud layer water content change perception model to obtain cloud layer water vapor abnormal results output by the cloud layer water content change perception model; the cloud layer water content change perception model is used to detect cloud layer water vapor abnormal data based on the change of the cloud layer water content change cloud layer water vapor data difference. The 5G-A base station multi-hop relay cooperative perception under the cloud system moving trajectory is used to detect the water vapor change in the cloud system moving process, the detection capability of the 5G-A base station information transmission chain for the water vapor change in the cloud system moving process is fully utilized, large-scale and continuous monitoring of the water vapor dynamics in the fast-changing cloud system is realized, and reliable observation support is provided for disaster prevention and reduction and emergency disposal.

[0035] In an optional embodiment, the cloud layer trajectory prediction model includes an input layer, a plurality of hidden layers and an output layer connected in sequence; The input layer includes n neurons for receiving cloud layer trajectory data in the past n periods; The plurality of hidden layers include eight long short-term memory networks connected in sequence, and each long short-term memory network includes an LSTM layer and a dropout layer connected in sequence; The output layer includes m neurons for outputting cloud layer trajectory data in the future m periods.

[0036] In the embodiment of the present application, the cloud layer trajectory prediction model is composed of a long short-term memory neural network. The long short-term memory neural network LSTM is a special type of recurrent neural network. Each neuron has four inputs and one output, and each neuron has a Cell inside to store the value of memory.

[0037] In the embodiment of the present application, the cloud layer trajectory prediction model comprises 1 input layer, 16 hidden layers (8 LSTM layers and 8 dropout layers), and 1 output layer (Dense layer). The input layer contains n neurons, and the input layer inputs the cloud layer trajectory in the latest n time periods. The output layer contains m neurons, and the output layer outputs the cloud layer trajectory in the future m time periods. The activation function used by each layer is set to the ReLu function. The first and second LSTM layers are set to 128 LSTM neurons, the third and fourth LSTM layers are set to 64 LSTM neurons, the fifth and sixth LSTM layers are set to 32 LSTM neurons, and the seventh and eighth LSTM layers are set to 16 LSTM neurons. After each LSTM layer, a dropout layer is introduced to discard neurons with a probability p and let other neurons remain with a probability q=1-p, so as to effectively avoid overfitting. Preferably, the discard probability is set to 0.2, that is, 20% of the neurons are randomly ignored to make them invalid.

[0038] Figure 3 is a structural diagram of the cloud layer trajectory prediction model provided by the present application, as shown in Figure 3 Each circle represents a neuron, and each line has a different weight. The neural network learns the weight value autonomously through training. The weight value in the model needs to be learned autonomously by the neural network, and the weight value does not need to be set manually.

[0039] Here, the sliding window mechanism is used for LSTM regression, which can use multiple recent historical time points to predict the next time point. The size parameter of the window can be adaptively set according to actual use requirements. During offline training, the latest n minutes of time series are used to predict the future m minutes of time series, and the input window is set to n and the output window is set to m.

[0040] Alternatively, the cloud layer trajectory prediction model is trained by the following method: training for 1000 epochs (epochs=1000), setting the batch size to 10 (batch_size=10), selecting the mean absolute error MSE (Mean Squared Error) as the loss function, that is, the objective function (loss='MSE'), and selecting the adam optimizer for gradient descent optimization algorithm to improve the learning speed of the traditional gradient descent (optimizer='adam').

[0041] The water vapor monitoring method provided by the embodiment of the application uses a long short-term memory network to predict a cloud layer track, effectively processes a long-term dependence problem existing in the cloud layer track, introduces a dropout layer after each LSTM layer to discard neurons at a certain probability, thereby avoiding model overfitting and improving the performance of the cloud layer track prediction model.

[0042] In an optional embodiment, the 5G-A base station information transmission chain in the future period is determined based on the cloud layer track prediction result, including: determining the 5G-A base station corresponding to the cloud layer track in the future period based on the cloud layer track prediction result; inputting the 5G-A base station into a 5G-A base station information transmission chain generation model to obtain the 5G-A base station information transmission chain output by the 5G-A base station information transmission chain generation model; the 5G-A base station information transmission chain generation model includes a 5G-A base station encoder and a 5G-A base station information transmission chain generator: the 5G-A base station encoder is configured to map each 5G-A base station to a specified dimension to obtain a latent feature representation of each 5G-A base station; the 5G-A base station information transmission chain generator is configured to predict the latent information transmission relationship of each 5G-A base station based on the latent feature representation of each 5G-A base station, and generate the 5G-A base station information transmission chain.

[0043] In the embodiment of the application, the 5G-A base station information transmission chain generation model is composed of a relational graph convolutional neural network, inputs the 5G-A base station corresponding to the cloud layer track in the future m period, and outputs the 5G-A base station information transmission chain in the future m period, that is, the 5G-A base station information transmission chain corresponding to the cloud layer track in the future m period.

[0044] Figure 4 is a structural diagram of the 5G-A base station information transmission chain generation model provided by the application, as Figure 4 shown, the 5G-A base station encoder corresponding to the cloud layer track, the input is the feature vector x i of each 5G-A base station node, the encoder maps each node to a latent feature representation z i with a dimension of d; the 5G-A base station information transmission chain generator scores the latent information transmission relationship between each 5G-A base station node by using a tensor factor decomposition operation to learn the latent feature representation of each 5G-A base station node output by the encoder, thereby determining the information transmission relationship between each 5G-A base station.

[0045] In the embodiment of the application, the 5G-A base station information transmission chain topology can be represented as a directed and weightless graph G=(V, E, R), E is the set of edges. V is the set of base station nodes V={V1, V2, V3, …, V N}. The data set includes: The adjacency matrix A is the information transmission relationship between nodes in the 5G-A base station information transmission chain topology, which is a feature description of the matrix form of the graph structure, e ij represents the information transmission relationship between the base station node V i and the base station node V j , -1 represents that the base station node j transmits information to the base station node i, 1 represents that the base station node i transmits information to the base station node j, and 0 represents that there is no information transmission relationship between the base station node i and the base station node j. The shape is N*N (N is the total number of base station nodes).

[0046] Here, the 5G-A base station node feature vector can be represented as {a1, a2, a3, …, a N}, which contains a feature sequence representing the longitude and latitude of the location of each 5G-A base station. All characters and punctuation marks are retained during text preprocessing. The encoding sequence length of each node feature text is defined as F, and the longest length F in the node feature set is taken as the encoding sequence length, and the length of each data is filled to F.

[0047] When training the 5G-A base station information transmission chain generation model, the error between the predicted potential information transmission relationship and the real potential information transmission relationship is calculated, and the training target is to minimize the error. The target function selects a 'categorical_crossentropy' multi-class logarithmic loss function. The number of training rounds is set to 2000 (epochs=2000), and the gradient descent optimization algorithm selects the adam optimizer to improve the learning speed of the traditional gradient descent (optimizer='adam').

[0048] The water vapor monitoring method provided by the embodiment of the application converts the 5G-A base station into a vector representation through an encoder, predicts the information transmission relationship between each 5G-A base station by using a generator, and thus accurately generates a 5G-A base station information transmission chain, thereby providing a basis for the transmission of the difference value of the cloud layer water vapor data.

[0049] In an optional embodiment, the 5G-A base station encoder includes two convolutional networks connected in sequence, and each convolutional network includes a relational graph convolutional layer and a random dropout layer connected in sequence.

[0050] In the embodiment of the application, the specific structure of the 5G-A base station encoder is as follows: The first layer is a relational graph convolutional layer (R-GCN): the number of convolutional kernels is 128 (i.e. the dimension of the output), and the activation function is ReLu; The second layer is a random dropout layer, and the dropout probability is preferably set to 0.2. During the training process, the input neurons are randomly disconnected with a certain probability (20%) each time the parameters are updated, which is used to prevent overfitting. The third layer is a relation graph convolutional layer (R-GCN), and the number of convolution kernels is 64, and the activation function is set to "lamda". The fourth layer is a random dropout layer, and the dropout probability is preferably set to 0.2.

[0051] For each 5G-A base station node, the expression of each layer of the relation graph convolutional layer is as follows: ; Where h i l is the hidden state of node v i in the lth layer of the neural network, r is the relation type, W r l is the parameter matrix of the lth neural network layer specific to the relation type, and the activation function is selected as ReLU(⋅) = max(0,⋅), ReLU is a rectified linear unit, and the input of the first layer is the node feature vector x i =h i 0 If there are L layers of stacking, the final output of the encoder is z i =h i L In a general GCN, D’ -1 / 2 A’ D’ -1 / 2 is the symmetric normalization of the adjacency matrix A, A’ = A + I, and D’ is the node degree diagonal matrix of A’. For a single node in this scenario, normalization means dividing it by the degree of the node. In this way, the value of each adjacent edge information transmission is normalized, and the influence of a node with more edges will not be greater than that of another node with fewer edges. Therefore, 1 / c i,r is equivalent to the normalization of the adjacency matrix in the GCN, c i,r is a regularization constant, and c i,r = |N i r |, N i r represents the neighbor set of node i under relation r.

[0052] In an optional embodiment, the 5G-A base station information transmission chain generator predicts the potential information transmission relationship between any two 5G-A base stations based on the following steps: For any two 5G-A base stations, a score of existence of a specified association relationship between the two 5G-A base stations is calculated using a DistMult scoring function; The score is converted into a probability value using an activation function; If the probability value is greater than a preset threshold, it is determined that the specified association relationship exists between the two 5G-A base stations.

[0053] As shown in Figure 4 The fifth layer is a tensor factorization layer (DistMult factorization): using the latent space vector representations z i ' and z j ' of 5G-A base station nodes i and j output by the encoder, the decoder predicts candidate edges (v i , r, v j ) through a factorization operation, and the activation function is set to "sigmoid". The decoder scores possible edges (v i , r, v j ) through a function g(v i , r, v j ) to determine the likelihood of these edges belonging to the set E, and the score g(v i , r, v j ) represents the likelihood of 5G-A base station nodes v i and v j being associated through a relationship r. Using the latent space vector representations z i ' and z j ' of 5G-A base station nodes i and j after merging, the decoder predicts candidate edges (v i , r, v j ) through DistMult factorization as a scoring function: ; where R r is a diagonal matrix of shape d*d, which represents the importance of each dimension in z i ' to the association relationship r. Finally, sigmoid(g(v i , r, v j )) represents the likelihood of edge (v i , r, v j ) belonging to a certain latent information transmission relationship.

[0054] The water vapor monitoring method provided by the embodiment of the application enables the model to effectively learn the complex relationship between nodes in the 5G-A base station network and accurately predict unknown relationships based on the learned representation, and the DistMult model is used to effectively utilize the simplicity and efficiency advantages of the DistMult model and is suitable for the link prediction task in the large-scale knowledge graph of cloud layer water vapor monitoring.

[0055] In an optional embodiment, the cloud layer water content change perception model comprises an input layer, eight fully connected layers and an output layer connected in sequence; the first four layers of the eight fully connected layers belong to an encoder, the encoder is used for compressing and reducing dimension of the cloud layer water vapor data difference value to obtain a feature vector of the cloud layer water vapor data difference value; the last four layers of the eight fully connected layers belong to a decoder, the decoder is used for restoring and reconstructing the feature vector to obtain a reconstructed difference value. The cloud layer water content change perception model obtains the cloud layer water vapor anomaly result based on the following steps: If the reconstructed error of a 5G-A base station is greater than a reconstructed error threshold, it is determined that the cloud layer water vapor data of the 5G-A base station is abnormal; the reconstructed error threshold is the error between the cloud layer water vapor data difference value of the 5G-A base station and its previous 5G-A base station and the reconstructed difference value corresponding to the cloud layer water vapor data difference value.

[0056] In the embodiment of the application, the cloud layer water content change perception model is composed of a long short-term memory neural network. The cloud layer water vapor data difference value perceived by each base station in the 5G-A base station information transmission chain is calculated, that is, the cloud layer water vapor data perceived by the base station i+1 is calculated from the cloud layer water vapor data perceived by the base station i.

[0057] Here, the cloud layer water content change perception model includes 1 input layer, 8 hidden layers (8 fully connected layers dense), and 1 output layer. The input layer is set to 18 neurons, and the output layer is set to 18 neurons; the first four layers of the eight hidden layers belong to an encoder, and the last four layers belong to a decoder.

[0058] In the embodiment of the application, the encoder is responsible for compressing and reducing dimension of the cloud layer water content difference value perceived by the 5G-A base station, and extracting a feature vector that can represent the cloud layer water content change difference value in the moving process. The first layer in the encoder is set to 128 neurons, the "tanh" function is selected as the activation function, the second layer is set to 64 neurons, the "relu" function is selected as the activation function, the third layer is set to 32 neurons, the "relu" function is selected as the activation function, and the fourth layer is set to 16 neurons, the "relu" function is selected as the activation function.

[0059] In the embodiment of the application, the decoder is responsible for restoring and reconstructing the compressed 5G-A base station awareness cloud layer water content difference feature vector. In the decoder, 16 neurons are arranged in the first layer, 'tanh' is selected as the activation function, 32 neurons are arranged in the second layer, 'tanh' is selected as the activation function, 64 neurons are arranged in the third layer, 'tanh' is selected as the activation function, and 128 neurons are arranged in the fourth layer, and'relu' is selected as the activation function.

[0060] Figure 5 is a structural diagram of the cloud layer water content change awareness model provided by the application, as shown in Figure 5 Each circle represents a neuron, and each hidden layer is a fully connected layer, that is, each neuron is connected to each other, and each line has different weights (weight). The auto-encoding neural network learns the weight value through training.

[0061] In the training process of the cloud layer water content change awareness model, the cloud layer water content type of each granularity time in the entire data set is labeled, and the normal state is labeled as 0 and the abnormal state is labeled as 1. In the anomaly detection problem, only the samples with normal labels in the training set are used, but a small number of abnormal samples are also needed to verify the model results in order to test the system performance; Secondly, the data set is standardized: (X-mean) / std. When calculating, each dimension is calculated separately, and the data is subtracted from the mean value according to the attribute (according to the column) and divided by the variance. After standardization, the convergence speed and accuracy of the model will be improved; Finally, the data set is divided: the total data set is divided into training data and test data, 80% of the entire data set is taken as training data, and the remaining 20% is taken as test data, and the data labeled as abnormal in the training data is removed, so that the cloud layer water content data in the training set is in the normal state, and the labels in the training set and the test set are also removed. Train the training set (only containing normal samples) to make the reconstructed data as close to the original data as possible, and use the test set (containing normal samples and abnormal samples) to evaluate and verify the model.

[0062] Specifically, 1000 epochs (epochs=1000) are trained, the batch size is set to 32 (batch_size=32), the mean squared error MSE (Mean Squared Error) is selected as the loss function, that is, the objective function (loss='mean_squared_error'), and the adam optimizer is selected to improve the learning speed of the traditional gradient descent (optimizer='adam').

[0063] The water vapor monitoring method provided by the embodiment of the application can effectively capture the mode of the change of the cloud water content over time and discover potential abnormal conditions through the powerful feature learning capability of the autoencoder by learning the change of the difference value of the water content during the movement of the cloud layer.

[0064] Figure 2 is a second flowchart of the water vapor monitoring method provided by the application, and the preferred embodiment of the application is described below. Figure 2 1) Obtain the cloud layer trajectory in a certain region in the last n time periods from the cloud layer observation system of the meteorological department; 2) Perform data standardization processing on the longitude and latitude data of the cloud layer trajectory in the last n time periods through a data preprocessing module; 3) Input the longitude and latitude of the cloud layer trajectory in the last n time periods after the data standardization processing into the cloud layer trajectory prediction model based on LSTM; 4) The cloud layer trajectory prediction model outputs the prediction result of the cloud layer trajectory in the future m time periods by using the advantage of LSTM in time series prediction; 5) Input the predicted cloud layer trajectory in the future m time periods into the 5G-A passive Internet of Things management and control platform; 6) The 5G-A passive Internet of Things management and control platform outputs the corresponding 5G-A base station under the cloud layer trajectory in the future m time periods; 7) Input the corresponding 5G-A base station under the cloud layer trajectory in the future m time periods into the 5G-A base station information transmission chain generation model; 8) The 5G-A base station information transmission chain generation model generates the 5G-A base station information transmission chain corresponding to the cloud layer trajectory in the future m time periods by using the relational graph convolutional neural network; 9) In the 5G-A base station information transmission chain, the base station i transmits the sensed cloud water vapor data to the next hop base station i+1, and the base station i+1 calculates the difference value of the sensed cloud water vapor data and the cloud water vapor data sensed by the last hop base station i; 10) Each base station in the 5G-A base station information transmission chain reports the difference value result to the 5G-A passive Internet of Things management and control platform; 11) The 5G-A passive Internet of Things management and control platform inputs the difference value of the cloud water vapor data sensed by each base station in the 5G-A base station information transmission chain into the cloud water content change sensing model; 12) The cloud water content change sensing model learns the change of the difference value of the cloud water vapor data by using the autoencoder neural network, and finally outputs the cloud water content change sensing result.

[0065] The water vapor monitoring device provided by the embodiment of the application is described below, and the water vapor monitoring device described below can be correspondingly referred to the water vapor monitoring method described above. ​

[0066] Figure 6 is a structural schematic diagram of the water vapor monitoring device provided by the present application, as shown in Figure 6 The water vapor monitoring device can include but is not limited to: The trajectory prediction module 610 is configured to input the preprocessed cloud layer trajectory data into a cloud layer trajectory prediction model to obtain a cloud layer trajectory prediction result in a future period output by the cloud layer trajectory prediction model; the cloud layer trajectory prediction model is configured to predict cloud layer trajectory data in a future period based on given historical cloud layer trajectory data. The transmission chain generation module 620 is configured to determine a 5G-A base station information transmission chain in a future period based on the cloud layer trajectory prediction result; the 5G-A base station information transmission chain includes a plurality of 5G-A base stations arranged in sequence. The instruction issuing module 630 is configured to issue a cloud layer water vapor sensing instruction to a 5G-A base station in the 5G-A base station information transmission chain; the cloud layer water vapor sensing instruction is configured to instruct an i-th 5G-A base station to transmit sensed cloud layer water vapor data to an (i+1)-th 5G-A base station, and instruct the (i+1)-th 5G-A base station to calculate a cloud layer water vapor data difference value between the i-th 5G-A base station and the (i+1)-th 5G-A base station, the cloud layer water vapor data difference value being a difference value of cloud layer water vapor data, i being a positive integer. The anomaly prediction module 640 is configured to input each cloud layer water vapor data difference value returned by each 5G-A base station into a cloud layer water content change sensing model to obtain a cloud layer water vapor anomaly result output by the cloud layer water content change sensing model after receiving the cloud layer water vapor data difference value; the cloud layer water content change sensing model is configured to detect cloud layer water vapor anomaly data based on changes in cloud layer water content change cloud layer water vapor data difference value.

[0067] It should be noted that the water vapor monitoring device provided by the embodiment of the present application can execute the water vapor monitoring method described in any of the above embodiments when actually running, and the embodiment will not be repeated here.

[0068] Figure 7 An example of an electronic device entity structure schematic diagram is shown in Figure 7 The electronic device can include a processor 710, a communications interface 720, a memory 730, and a communications bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other through the communications bus 740. The processor 710 can invoke the logic instructions in the memory 730 to execute a water vapor monitoring method, which includes: input the preprocessed cloud layer track data into a cloud layer track prediction model to obtain a cloud layer track prediction result in a future period output by the cloud layer track prediction model; the cloud layer track prediction model is used to predict cloud layer track data in a future period based on given historical cloud layer track data; based on the cloud layer track prediction result, determine a 5G-A base station information transmission chain in the future period; the 5G-A base station information transmission chain includes a plurality of 5G-A base stations arranged in sequence; issue a cloud layer water vapor sensing instruction to the 5G-A base stations in the 5G-A base station information transmission chain; the cloud layer water vapor sensing instruction is used to instruct an i-th 5G-A base station to transmit sensed cloud layer water vapor data to an i+1-th 5G-A base station, and instruct the i+1-th 5G-A base station to calculate a cloud layer water vapor data difference value between the i-th 5G-A base station and the i+1-th 5G-A base station, the cloud layer water vapor data difference value being a difference value of cloud layer water vapor data, i being a positive integer; after receiving the cloud layer water vapor data difference values returned by each of the 5G-A base stations, input each of the cloud layer water vapor data difference values into a cloud layer water content change sensing model to obtain a cloud layer water vapor anomaly result output by the cloud layer water content change sensing model; the cloud layer water content change sensing model is used to detect cloud layer water vapor anomaly data based on changes in cloud layer water content change cloud layer water vapor data difference values.

[0069] In addition, the logical instructions in the memory 730 described above can be implemented in the form of a software function unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the part of the prior art that contributes essentially or the part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various storage program codes.

[0070] On the other hand, the present application also provides a computer program product, which includes a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor, so that the computer can execute the water vapor monitoring method provided by the above-mentioned methods, and the method includes: input the preprocessed cloud layer trajectory data into a cloud layer trajectory prediction model to obtain a cloud layer trajectory prediction result in a future period output by the cloud layer trajectory prediction model, the cloud layer trajectory prediction model being configured to predict cloud layer trajectory data in a future period based on given historical cloud layer trajectory data; determine a 5G-A base station information transmission chain in the future period based on the cloud layer trajectory prediction result, the 5G-A base station information transmission chain including a plurality of 5G-A base stations arranged in sequence; issue a cloud layer water vapor sensing instruction to the 5G-A base stations in the 5G-A base station information transmission chain, the cloud layer water vapor sensing instruction being configured to instruct an i-th 5G-A base station to transmit sensed cloud layer water vapor data to an (i+1)-th 5G-A base station, and instruct the (i+1)-th 5G-A base station to calculate a cloud layer water vapor data difference between the i-th 5G-A base station and the (i+1)-th 5G-A base station, the cloud layer water vapor data difference being a difference in cloud layer water vapor data, i being a positive integer; after receiving the cloud layer water vapor data differences returned by the 5G-A base stations, input the cloud layer water vapor data differences into a cloud layer water content change sensing model to obtain cloud layer water vapor anomaly results output by the cloud layer water content change sensing model, the cloud layer water content change sensing model being configured to detect cloud layer water vapor anomaly data based on changes in cloud layer water content change cloud layer water vapor data differences.

[0071] In another aspect, the present application also provides a non-transitory computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement a water vapor monitoring method provided by the above method, the method comprising: input the preprocessed cloud layer trajectory data into a cloud layer trajectory prediction model to obtain a cloud layer trajectory prediction result in a future period output by the cloud layer trajectory prediction model, the cloud layer trajectory prediction model being configured to predict cloud layer trajectory data in a future period based on given historical cloud layer trajectory data; determine a 5G-A base station information transmission chain in the future period based on the cloud layer trajectory prediction result, the 5G-A base station information transmission chain including a plurality of 5G-A base stations arranged in sequence; issue a cloud layer water vapor sensing instruction to the 5G-A base stations in the 5G-A base station information transmission chain, the cloud layer water vapor sensing instruction being configured to instruct an i-th 5G-A base station to transmit sensed cloud layer water vapor data to an (i+1)-th 5G-A base station, and instruct the (i+1)-th 5G-A base station to calculate a cloud layer water vapor data difference between the i-th 5G-A base station and the (i+1)-th 5G-A base station, the cloud layer water vapor data difference being a difference in cloud layer water vapor data, i being a positive integer; After receiving the cloud water vapor data difference values returned by each of the 5G-A base stations, the cloud water vapor data difference values are input into a cloud water content change perception model to obtain cloud water vapor anomaly results output by the cloud water content change perception model; the cloud water content change perception model is used to detect cloud water vapor anomaly data based on changes in cloud water content change cloud water vapor data difference values.

[0072] The device embodiments described above are merely illustrative, wherein the units illustrated as separate components can or can not be physically separated, and the components illustrated as units can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement it without creative labor.

[0073] From the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and the necessary general hardware platform, and of course, it can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0074] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of water vapor monitoring, the method comprising: include: The preprocessed cloud trajectory data is input into the cloud trajectory prediction model to obtain the cloud trajectory prediction results for the future time period output by the cloud trajectory prediction model. The cloud trajectory prediction model is used to predict cloud trajectory data for future periods based on given historical cloud trajectory data. Based on the cloud trajectory prediction results, the 5G-A base station information transmission chain for the future period is determined; the 5G-A base station information transmission chain includes multiple 5G-A base stations arranged in sequence. The cloud moisture sensing command is sent to the 5G-A base station in the 5G-A base station information transmission chain; the cloud moisture sensing command is used to instruct the i-th 5G-A base station to transmit the sensed cloud moisture data to the (i+1)-th 5G-A base station, and instruct the (i+1)-th 5G-A base station to calculate the difference between the cloud moisture data of the i-th 5G-A base station and the (i+1)-th 5G-A base station, where the cloud moisture data difference is the difference between the cloud moisture data, and i is a positive integer; After receiving the cloud water vapor data difference returned by each of the 5G-A base stations, the cloud water vapor data difference is input into the cloud water content change sensing model to obtain the cloud water vapor anomaly result output by the cloud water content change sensing model; the cloud water content change sensing model is used to detect cloud water vapor anomaly data based on the change of cloud water vapor data difference in cloud water content change.

2. The water vapor monitoring method of claim 1, wherein, The cloud trajectory prediction model includes an input layer, multiple hidden layers, and an output layer connected in sequence. The input layer contains n neurons, used to receive cloud trajectory data over the past n time periods; The multiple hidden layers include eight sequentially connected long short-term memory networks, and each long short-term memory network includes a sequentially connected LSTM layer and a dropout layer. The output layer contains m neurons, which are used to output cloud trajectory data for the next m time periods.

3. The method of claim 1, wherein, The step of determining the 5G-A base station information transmission chain in the future time period based on the cloud trajectory prediction results includes: Based on the cloud trajectory prediction results, determine the 5G-A base station corresponding to the cloud trajectory in the future time period; The 5G-A base station is input into the 5G-A base station information transmission chain generation model to obtain the 5G-A base station information transmission chain output by the 5G-A base station information transmission chain generation model. The 5G-A base station information transmission chain generation model includes a 5G-A base station encoder and a 5G-A base station information transmission chain generator: The 5G-A base station encoder is used to map each 5G-A base station to a specified dimension to obtain the potential feature representation of each 5G-A base station; The 5G-A base station information transmission chain generator is used to predict the potential information transmission relationship of each 5G-A base station based on the potential feature representation of each 5G-A base station, and generate a 5G-A base station information transmission chain.

4. The method of claim 3, wherein, The 5G-A base station encoder includes two convolutional networks connected in sequence, each of which includes a relational graph convolutional layer and a random discard layer connected in sequence.

5. The method of claim 3, wherein, The 5G-A base station information transmission chain generator predicts the potential information transmission relationship between any two 5G-A base stations based on the following steps: For any two 5G-A base stations, the DistMult scoring function is used to calculate a score indicating a specified correlation between the two 5G-A base stations; The rating is converted into a probability value using an activation function; If the probability value is greater than a preset threshold, then it is determined that there is a specified association between the two 5G-A base stations.

6. The method of claim 1, wherein, The cloud moisture content change sensing model comprises an input layer, eight fully connected layers, and an output layer connected in sequence. The first four layers of the eight fully connected layers are encoders, which are used to compress and reduce the dimension of the cloud moisture data difference to obtain a feature vector of the cloud moisture data difference. The last four layers of the eight fully connected layers are decoders, which are used to restore and reconstruct the feature vector to obtain a reconstructed difference. The cloud moisture content change sensing model obtains cloud water vapor anomaly results based on the following steps: If the reconstruction error of a 5G-A base station is greater than the reconstruction error threshold, then the cloud water vapor data corresponding to the 5G-A base station is determined to be abnormal; the reconstruction error threshold is the difference between the cloud water vapor data of the 5G-A base station and the previous 5G-A base station, and the error between the reconstruction difference corresponding to the difference in cloud water vapor data.

7. A water vapor monitoring device, characterized by, include: The trajectory prediction module is used to: input preprocessed cloud trajectory data into the cloud trajectory prediction model to obtain the cloud trajectory prediction result for the future time period output by the cloud trajectory prediction model; the cloud trajectory prediction model is used to predict cloud trajectory data for the future time period based on given historical cloud trajectory data. The transmission chain generation module is used to: determine the 5G-A base station information transmission chain in the future time period based on the cloud trajectory prediction result; the 5G-A base station information transmission chain includes multiple 5G-A base stations arranged in sequence; The instruction issuing module is used to: issue cloud moisture sensing instructions to the 5G-A base stations in the 5G-A base station information transmission chain; the cloud moisture sensing instructions are used to instruct the i-th 5G-A base station to transmit the sensed cloud moisture data to the (i+1)-th 5G-A base station, and instruct the (i+1)-th 5G-A base station to calculate the difference between the cloud moisture data of the i-th 5G-A base station and the (i+1)-th 5G-A base station, wherein the cloud moisture data difference is the difference between the cloud moisture data, and i is a positive integer; The anomaly prediction module is used to: receive the cloud water vapor data difference returned by each of the 5G-A base stations, input the cloud water vapor data difference into the cloud water content change sensing model, and obtain the cloud water vapor anomaly result output by the cloud water content change sensing model; the cloud water content change sensing model is used to detect cloud water vapor anomaly data based on the change of cloud water vapor data difference in cloud water content change.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the water vapor monitoring method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the water vapor monitoring method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the water vapor monitoring method as described in any one of claims 1 to 6.

Images