Early warning information processing method and device, computer program product and electronic equipment

Abstract

The invention discloses an early warning information processing method and device, a computer program product and electronic equipment. The method relates to the field of big data or other related fields, and comprises the following steps: acquiring M pieces of monitoring and early warning information generated by a departure system, and processing each piece of monitoring and early warning information to obtain M groups of early warning text vectors; obtaining a target classification model, inputting the M groups of early warning text vectors into the target classification model, and outputting N groups of early warning information sets; dividing the N groups of early warning information sets according to an early warning grade threshold to obtain X groups of early warning information sets and Z groups of early warning information sets; monitoring and early warning information contained in the X groups of early warning information sets is sent to the client, the monitoring and early warning information contained in the Z groups of early warning information sets is aggregated to obtain Z groups of aggregated early warning information sets, and the Z groups of aggregated early warning information sets are sent to the client. According to the method and the device, the problems of low processing efficiency and low processing accuracy when the early warning information is processed in the prior art are solved.

Description

Technical Field

[0001] The present application relates to the field of big data or other related fields, and specifically, to a method, device, computer program product and electronic device for processing early warning information. Background Art

[0002] In the modern aviation industry, the airport departure software monitoring system is a key technology to ensure the safe and efficient operation of flights. The system is responsible for monitoring and managing all aspects of the airport, including but not limited to aircraft takeoff and landing, ground services, passenger processing, etc. With the rapid development of the aviation industry, the challenges faced by the airport departure monitoring system are also increasing, especially in dealing with a large amount of warning information.

[0003] In the monitoring process, traditional systems may generate repeated warning information due to sensor failure, communication errors, etc.; due to system misjudgment or external interference, the system may issue unnecessary warnings or issue warnings for some non-critical issues, causing operation and maintenance personnel to spend extra time to deal with them, thereby increasing processing pressure and reducing efficiency. In other words, in existing technologies, most monitoring systems use more traditional warning processing methods, which often lack efficient data processing capabilities and cannot accurately filter and classify warning information, resulting in excessive reliance on operation and maintenance personnel and delayed response. In addition, it may not be possible to accurately identify the importance and urgency of the warning, resulting in the need for operation and maintenance personnel to manually screen information.

[0004] In order to solve the above problems, related technologies usually only optimize a specific aspect of the early warning system, such as enhancing data processing capabilities or improving the user interface. However, these solutions fail to fully consider the comprehensive needs of the early warning system, such as real-time data processing, automatic classification, user feedback integration, etc., and still cannot effectively reduce the workload of operation and maintenance personnel and improve the efficiency and accuracy of processing early warning information.

[0005] There are problems of low processing efficiency and low processing accuracy when processing early warning information in related technologies, and no effective solution has been proposed yet. Summary of the invention

[0006] The main purpose of this application is to provide a method, device, computer program product and electronic device for processing early warning information to solve the problems of low processing efficiency and low processing accuracy when processing early warning information in related technologies.

[0007] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for processing warning information is provided. The method includes: obtaining M monitoring warning information generated by the departure system, processing each monitoring warning information, and obtaining M groups of warning text vectors, wherein M is a positive integer; obtaining a target classification model, inputting the M groups of warning text vectors into the target classification model, and outputting N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer; obtaining a warning level threshold, dividing the N groups of warning information sets according to the warning level threshold, and obtaining X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z, X are positive integers; sending the monitoring warning information contained in the X group of warning information sets to the client, aggregating the monitoring warning information contained in the Z group of warning information sets to obtain the Z group of aggregated warning information sets, and sending the Z group of aggregated warning information sets to the client, wherein the client is used to process the monitoring warning information contained in each group of warning information sets.

[0008] Furthermore, each monitoring warning information is processed to obtain M groups of warning text vectors, including: for one monitoring warning information, segmenting the monitoring warning information to obtain a group of segmented text data; converting the group of segmented text data into integer identifiers to obtain a group of processed sequence data; performing sequence completion processing on the group of processed sequence data to obtain a group of text sequence data; inputting a group of text sequences into a word embedding model to obtain a group of warning text vectors associated with the monitoring warning information.

[0009] Furthermore, the target classification model is trained through the following steps: obtaining Y historical warning information and warning levels associated with Y historical warning information, preprocessing the Y historical warning information to obtain Y groups of historical warning vectors, and the Y groups of historical warning vectors and the warning levels associated with Y historical warning information constitute a model training set, wherein Y is a positive integer; obtaining a preset classification model, initializing the preset classification model using a target optimizer to obtain an initialized classification model; training the initialized classification model with the model training set to obtain a trained classification model, and obtaining the output result of the trained classification model, optimizing the trained classification model based on the output result to obtain a target classification model.

[0010] Furthermore, the trained classification model is optimized based on the output result to obtain a target classification model, including: obtaining a preset classification result, calculating the cross entropy function between the output result and the preset classification result, and obtaining cross entropy data; calculating the parameter gradient of the cross entropy data based on the back propagation algorithm, wherein the parameter gradient refers to the gradient of the loss function associated with the trained classification model with respect to the model parameters; obtaining a first-order matrix vector, calculating a second-order matrix vector using the parameter gradient and the first-order matrix vector, and updating the second-order matrix vector using the target optimizer to obtain an updated matrix vector, wherein the first-order matrix vector is determined by the model parameters; determining an updated loss function using the updated matrix vector, and when the updated loss function is less than a preset threshold, determining an updated classification model according to the updated loss function until the updated loss function is greater than or equal to the preset threshold, and determining the updated classification model as the target classification model.

[0011] Furthermore, the monitoring warning information contained in the Z group of warning information sets is aggregated to obtain the Z group of aggregated warning information sets, including: for a group of warning information sets, obtaining K monitoring warning information associated with a group of warning information sets, and adding the K monitoring warning information to the pending warning list, where K is a positive integer; obtaining a time window threshold, filtering the K monitoring warning information according to the time window threshold, and obtaining P monitoring warning information, where P is a positive integer; calculating the similarity of every two monitoring warning information in the P monitoring warning information to obtain L similarity data, where L is a positive integer; aggregating the P monitoring warning information according to the L similarity data to obtain T aggregated warning information, and the T aggregated warning information constitute a group of aggregated warning information sets, where T is a positive integer.

[0012] Furthermore, P monitoring warning information are aggregated according to L similarity data to obtain T aggregated warning information, including: obtaining a similarity threshold, judging whether the L similarity data are greater than the similarity threshold; in the case that there are similarity data greater than the similarity threshold in the L similarity data, obtaining T target similarity data, wherein each target similarity data refers to similarity data greater than the similarity threshold; merging a group of monitoring warning information associated with each target similarity data to obtain T aggregated warning information.

[0013] Furthermore, before the monitoring warning information included in the X group warning information set is sent to the client, the method also includes: obtaining a preset processing rule, and determining whether the time period associated with the M monitoring warning information is a preset time period, wherein the preset processing rule is used to adjust the response sensitivity of the client; when the time period associated with the M monitoring warning information is the preset time period, sending a first prompt message to the client according to the preset processing rule; when the time period associated with the M monitoring warning information is not the preset time period, sending a second prompt message to the client according to the preset processing rule, wherein the emergency level of the first prompt message is higher than the emergency level of the second prompt message.

[0014] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a device for processing warning information is provided. The device includes: a first acquisition unit, which is used to acquire M monitoring warning information generated by the departure system, process each monitoring warning information, and obtain M groups of warning text vectors, wherein M is a positive integer; a second acquisition unit, which is used to acquire a target classification model, input the M groups of warning text vectors into the target classification model, and output N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer; a third acquisition unit, which is used to acquire a warning level threshold, divide the N groups of warning information sets according to the warning level threshold, and obtain X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z, X are positive integers; a first sending unit, which is used to send the monitoring warning information contained in the X group of warning information sets to the client, aggregate the monitoring warning information contained in the Z group of warning information sets to obtain the Z group of aggregated warning information sets, and send the Z group of aggregated warning information sets to the client, wherein the client is used to process the monitoring warning information contained in each group of warning information sets.

[0015] According to another aspect of an embodiment of the present invention, a computer storage medium is provided. The computer storage medium is used to store a program. When the program is executed, the device where the computer storage medium is located is controlled to execute a method for processing early warning information.

[0016] According to another aspect of an embodiment of the present invention, an electronic device is provided, comprising one or more processors and a memory; the memory stores computer-readable instructions, and the processor is used to run the computer-readable instructions, wherein a method for processing early warning information is executed when the computer-readable instructions are run.

[0017] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, a method for processing early warning information is executed.

[0018] Through this application, the following steps are adopted: obtain M monitoring warning information generated by the departure system, process each monitoring warning information, and obtain M groups of warning text vectors, where M is a positive integer; obtain a target classification model, input the M groups of warning text vectors into the target classification model, and output N groups of warning information sets, where each group of warning information sets corresponds to a warning level, and N is a positive integer; obtain the warning level threshold, divide the N groups of warning information sets according to the warning level threshold, and obtain X groups of warning information sets and Z groups of warning information sets, where the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z, X are positive integers; the monitoring warning information contained in the X group of warning information sets The monitoring warning information contained in the Z group of warning information sets is aggregated and processed to obtain the Z group of aggregated warning information sets, and the Z group of aggregated warning information sets are sent to the client, wherein the client is used to process the monitoring warning information contained in each group of warning information sets, which solves the problems of low processing efficiency and low processing accuracy when processing warning information in related technologies. The monitoring warning information is obtained, the information is processed to obtain a warning text vector, and the warning text vector is processed using a target classification model to obtain a warning information set. After the warning information set is divided into warning levels, the warning information sets of different warning levels are processed in a targeted manner, thereby achieving the effect of improving the accuracy and efficiency of processing warning information. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a flowchart of a method for processing warning information provided in an embodiment of the present application;

[0021] Figure 2 is a schematic diagram of a classification method for early warning information provided according to an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of a preset classification model provided according to an embodiment of the present application;

[0023] Figure 4 It is a schematic diagram of a method for aggregating and processing monitoring warning information provided in an embodiment of the present application;

[0024] Figure 5 is a schematic diagram of a method for executing a preset processing rule provided in an embodiment of the present application;

[0025] Figure 6 is a schematic diagram of an optional method for processing early warning information provided according to an embodiment of the present application;

[0026] Figure 7 is a schematic diagram of an optional target classification model optimization method provided according to an embodiment of the present application;

[0027] Figure 8 is a schematic diagram of a device for processing early warning information provided in an embodiment of the present application;

[0028] Fig. 9 is a schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0030] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties. For example, an interface is set between this system and the relevant user or organization. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0033] It should be noted that the collected information used in this application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse to use it.

[0034] The present invention is described below in conjunction with preferred implementation steps. Figure 1 is a flowchart of a method for processing warning information provided in an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0035] Step S101, obtaining M monitoring warning information generated by the departure system, processing each monitoring warning information, and obtaining M groups of warning text vectors, wherein M is a positive integer.

[0036] Specifically, in order to improve the processing efficiency and accuracy of warning information, we can first obtain the monitoring warning information generated by the departure system in real time, and then perform standardization operations on each monitoring warning information, clean unstructured data, identify and eliminate noise information, etc., so as to obtain the warning text vector, ensure the quality of the data input into the target classification model, and lay the foundation for subsequent classification processing.

[0037] Step S102, obtaining a target classification model, inputting M groups of warning text vectors into the target classification model, and outputting N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer.

[0038] Specifically, the target classification model can be a long short-term memory network model, which can understand the context of monitoring warning information, identify and filter out repeated, false alarms or non-critical information, and then classify the monitoring warning information according to different warning levels, where the warning levels can include four levels: "urgent", "important", "normal" and "minor" for different processing and responses.

[0039] It should be noted that in order to improve the integration of the system, the target classification model can be integrated into the monitoring system of the departure system. Figure 2 is a schematic diagram of a classification method for early warning information provided in an embodiment of the present application, such as Figure 2 As shown, when the probe collects monitoring warning information, it sends the monitoring warning information to the monitoring system. At this time, the monitoring system uses the target classification model to identify the content and context of the warning text vector corresponding to the monitoring warning information, and then divides the monitoring warning information into four warning level warning information sets: "urgent", "important", "normal" and "minor".

[0040] Step S103, obtaining a warning level threshold, dividing N groups of warning information sets according to the warning level threshold, obtaining X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z and X are positive integers.

[0041] Specifically, since the warning level of each monitoring warning information is different, the monitoring warning information can be processed in different ways according to the warning level, thereby achieving the effect of improving the flexibility and practicality of the system. For example, if the warning level includes "urgent", "important", "ordinary" and "minor", the "urgent" and "important" monitoring warning information can be immediately notified to the operation and maintenance personnel, and the "ordinary" and "minor" monitoring warning information can be further processed using the aggregation technology based on the time window and the similarity threshold. For example, the similarity of different monitoring warning information is calculated, and then the monitoring warning information is judged whether there is similar information between the monitoring warning information according to the similarity threshold, and then the similar monitoring warning information is aggregated. When classifying the monitoring warning information of different warning levels, the warning level threshold can be first obtained, and the warning information set containing the monitoring warning information can be classified according to the threshold, and multiple warning information sets with warning levels of "urgent" or "important" and multiple warning information sets with warning levels of "ordinary" or "minor" are obtained.

[0042] Step S104: send the monitoring warning information contained in the X group warning information set to the client, aggregate the monitoring warning information contained in the Z group warning information set to obtain the Z group aggregated warning information set, and send the Z group aggregated warning information set to the client, wherein the client is used to process the monitoring warning information contained in each group warning information set.

[0043] Specifically, after classifying the warning information sets, multiple warning information sets of "urgent" or "important" can be sent directly to the client for timely processing by operation and maintenance personnel. Then, multiple warning information sets of "ordinary" or "minor" can be aggregated by calculating similarity data and using an aggregation algorithm to aggregate similar monitoring warning information. The aggregated warning information set is then sent to the client, thereby reducing redundant warnings and reducing the processing burden of operation and maintenance personnel.

[0044] The method for processing warning information provided in the embodiment of the present application obtains M monitoring warning information generated by the departure system, processes each monitoring warning information, and obtains M groups of warning text vectors, wherein M is a positive integer; obtains a target classification model, inputs the M groups of warning text vectors into the target classification model, and outputs N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer; obtains a warning level threshold, divides the N groups of warning information sets according to the warning level threshold, and obtains X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z, X are positive integers; the monitoring information sets contained in the X group of warning information sets are divided into the following groups: The warning information is sent to the client, the monitoring warning information contained in the Z group of warning information sets is aggregated and processed to obtain the Z group of aggregated warning information sets, and the Z group of aggregated warning information sets are sent to the client, wherein the client is used to process the monitoring warning information contained in each group of warning information sets, which solves the problems of low processing efficiency and low processing accuracy when processing warning information in related technologies. By obtaining the monitoring warning information, the information is processed to obtain a warning text vector, and the warning text vector is processed using a target classification model to obtain a warning information set. After the warning information set is divided into warning levels, the warning information sets of different warning levels are processed in a targeted manner, thereby achieving the effect of improving the accuracy and efficiency of processing warning information.

[0045] In order to ensure the quality of data input into the target classification model so that the target classification model can quickly classify the monitoring warning information, it is necessary to process the monitoring warning information. Optionally, in the warning information processing method provided in the embodiment of the present application, each monitoring warning information is processed to obtain M groups of warning text vectors, including: for one monitoring warning information, segmenting the monitoring warning information to obtain a group of segmented text data; converting a group of segmented text data into integer identifiers to obtain a group of processed sequence data; performing sequence completion processing on the group of processed sequence data to obtain a group of text sequence data; inputting a group of text sequences into a word embedding model to obtain a group of warning text vectors associated with the monitoring warning information.

[0046] Specifically, the monitoring and warning information can first be converted into a form that can be understood by the machine, and then segmented, the text content is segmented into simple words or phrases, and then a plurality of segmented text data are obtained, and then each unique word or phrase is converted into an integer identifier to obtain a set of processed sequence data for subsequent processing.

[0047] Since the sequence lengths of the converted integer identifiers are inconsistent, in order to ensure that the sequence lengths of each data input to the model are the same, these sequence data can be sequence-completed to ensure that they have the same length, thereby obtaining text sequence data. For example, the PADDING completion method or the over-length truncation method can be used for completion. For shorter sequence data, it can be completed, and the completion method can be filled with 0 values, that is, the shorter sequence is completed with 0 values; when the sequence data is too long and exceeds a fixed length, the sequence can be fixedly truncated to ensure that the input sequence length is consistent. Since the word vector can better capture the semantic relationship between words, so that the model can better understand the input sequence, these text sequence data can be input into the word embedding model, and the text sequence data can be converted into word vectors, thereby obtaining a more accurate set of warning text vectors associated with the monitoring warning information, wherein the word embedding model can be a Word2Vec model or a GloVe model. This embodiment processes the monitoring warning information, thereby helping the target classification model to better understand and analyze the monitoring warning information, and providing more accurate information support for processing the warning information.

[0048] In order to improve the accuracy of the classification model, optionally, in the warning information processing method provided in the embodiment of the present application, the target classification model is trained through the following steps: obtaining Y historical warning information and warning levels associated with Y historical warning information, preprocessing the Y historical warning information to obtain Y groups of historical warning vectors, and the model training set is composed of Y groups of historical warning vectors and warning levels associated with Y historical warning information, where Y is a positive integer; obtaining a preset classification model, and initializing the preset classification model using a target optimizer to obtain an initialized classification model; training the initialized classification model with the model training set to obtain a trained classification model, and obtaining the output result of the trained classification model, optimizing the trained classification model based on the output result to obtain a target classification model.

[0049] Specifically, in order to achieve accurate classification of monitoring warning information, it is necessary to train the classification model, wherein the preset classification model can be a long short-term memory network (LSTM) model. Before training the preset classification model, it is first necessary to obtain multiple historical warning information and their corresponding warning levels from the historical warning information database, pre-process these historical warning information, convert them into corresponding multiple groups of historical warning vectors, and finally use these historical warning vectors and their corresponding warning levels as model training sets for training the preset classification model.

[0050] Then you need to initialize the preset classification model: Figure 3is a schematic diagram of a preset classification model provided according to an embodiment of the present application, such as Figure 3 As shown, first, you can set the number of LSTM layers and the activation function ReLU, pass in the time series vector sequence data, specify the fully connected layer as the output layer, specify the number of its output neurons as 4 (that is, "urgent", "important", "normal" and "minor"), and use the softmax function as its activation function to map the output to the probability distribution. Since the output result of this model belongs to a multi-classification problem, cross entropy can be used as the loss function of this model. Assuming that the model prediction is p and the actual output classification result is q, the loss function is defined as: E = -q*log(p), that is, the loss function is the negative log-likelihood of the two label vectors. Finally, when compiling the model, the loss function is specified as the cross entropy loss, the target optimizer is the Adam optimizer, and the evaluation indicator is the accuracy.

[0051] After completing the construction and compilation of the preset classification model, it is also necessary to initialize the model parameters of the compiled model, that is, it is necessary to set the model weights and biases, etc., and use the Adam optimizer to initialize the gradient vector and the first-order matrix vector to obtain the initialized classification model. Furthermore, the above-mentioned model training set can be used for model training, that is, the historical warning vector in the model training set is input into the initialized classification model to obtain the output result, and then the trained classification model is optimized according to the output result, and finally the target classification model is obtained. In this embodiment, the initialized classification model is trained by using the historical warning information as the model training set, and finally the target classification model is constructed, and then the model is used to realize the classification and prediction of monitoring warning information, which reduces the labor cost while improving the processing efficiency of warning information.

[0052] Optionally, in the method for processing warning information provided in the embodiment of the present application, the trained classification model is optimized based on the output result to obtain the target classification model, including: obtaining a preset classification result, calculating the cross entropy function between the output result and the preset classification result, and obtaining cross entropy data; calculating the parameter gradient of the cross entropy data based on the back propagation algorithm, wherein the parameter gradient refers to the gradient of the loss function associated with the trained classification model with respect to the model parameters; obtaining a first-order matrix vector, calculating a second-order matrix vector using the parameter gradient and the first-order matrix vector, and updating the second-order matrix vector using the target optimizer to obtain an updated matrix vector, wherein the first-order matrix vector is determined by the model parameters; determining an updated loss function using the updated matrix vector, and when the updated loss function is less than a preset threshold, determining an updated classification model according to the updated loss function until the updated loss function is greater than or equal to the preset threshold, and determining the updated classification model as the target classification model.

[0053] Specifically, when optimizing the trained classification model, we first obtain the preset classification results and calculate the cross entropy function between the output results obtained by the model and the preset classification results to obtain the cross entropy data. Then, we use the back propagation algorithm to calculate the parameter gradient of the cross entropy data. The parameter gradient can help us adjust the model parameters so that the model can better fit the training data. As a gradient calculation method, the back propagation algorithm can reversely transfer the error signal from the output layer to the input layer.

[0054] Furthermore, the first-order matrix vector representing the gradient of the loss function to the model parameters can be obtained, and the parameter gradient and the first-order matrix vector can be used to update the second-order matrix vector, where the second-order matrix vector involves the curvature information of the parameters, that is, the gradient of the gradient. Then, the target optimizer is used to update the parameters of the model according to the calculated gradient and matrix vector to obtain the updated matrix vector. It should be noted that the above optimizer has an adaptive characteristic and can adjust the learning rate during the training process to speed up the convergence speed or avoid falling into the local minimum.

[0055] Finally, the updated loss function is determined using the updated matrix vector, and compared with the preset threshold. When the updated loss function is less than the preset threshold, it can be considered that the performance of the model has reached the requirement. The updated classification model is determined according to the updated loss function. If the loss function is greater than or equal to the preset threshold, iterative training is continued until the updated loss function is greater than or equal to the preset threshold. In each iteration, the parameters of the model are updated according to the gradient of the loss function in the hope of finding the global minimum, so that the model has the strongest generalization ability on new data. At this time, the updated classification model is determined as the target classification model. It should be noted that during the training process of the model, the model can be evaluated by calculating the accuracy of the model, which helps to train a more accurate classification model and improve the generalization ability of the model. This embodiment optimizes the classification model by using the output results, and then obtains a classification model with good classification ability, which lays the foundation for the subsequent classification of monitoring and early warning information.

[0056] In order to improve the efficiency of processing monitoring warning information, optionally, in the warning information processing method provided in the embodiment of the present application, the monitoring warning information contained in Z groups of warning information sets is aggregated to obtain Z groups of aggregated warning information sets, including: for a group of warning information sets, obtaining K monitoring warning information associated with a group of warning information sets, and adding the K monitoring warning information to the pending warning list, where K is a positive integer; obtaining a time window threshold, and filtering the K monitoring warning information according to the time window threshold to obtain P monitoring warning information, where P is a positive integer; calculating the similarity between every two monitoring warning information in the P monitoring warning information to obtain L similarity data, where L is a positive integer; aggregating the P monitoring warning information according to the L similarity data to obtain T aggregated warning information, and the T aggregated warning information constitute a group of aggregated warning information sets, where T is a positive integer.

[0057] Specifically, since the amount of monitoring and warning information collected by the departure system is huge, the same fault may be reported by different nodes at the same time. The surge in repeated warnings will increase the burden on operation and maintenance personnel, and also occupy the departure bandwidth, affecting the normal departure business processing. The algorithm based on time window and similarity aggregation can ensure timely and effective identification and response to real faults, greatly improving the accuracy of the airport's warnings. Therefore, the monitoring warning information with a lower warning level can be aggregated.

[0058] Before the monitoring warning information is aggregated, the length of the time window threshold is first set and a pending warning list is created, wherein the time window can be obtained by tracing back a given length with the current time as the center. Then all the monitoring warning information in the warning information set is added to the pending warning list for caching. The monitoring warning information in the above list is checked, and the monitoring warning information that is less than the time window is removed according to the time window threshold to obtain multiple monitoring warning information. Further, the similarity data between the removed monitoring warning information can be calculated, and the monitoring warning information can be aggregated according to the similarity data to obtain the aggregated warning information and the aggregated warning information set composed of the aggregated warning information.

[0059] It should be noted that Figure 4 is a schematic diagram of a method for aggregating monitoring and warning information provided in accordance with an embodiment of the present application, such as Figure 4As shown, when calculating similarity data, first, you can extract keywords after performing word segmentation, stop word removal, stem extraction, word frequency statistics and other technologies, and identify important information in the text for further similarity calculation. Then calculate the vector corresponding to the keyword. For example, the monitoring warning information A is: airport departure process delay, and the monitoring warning information B is: the delay in the aircraft takeoff leads to a delay in departure. These two sentences are converted into word vectors and a digital vector is assigned to each keyword. For example, the vector of "airport" is represented as [1, 0, 0, 0], the vector of "departure" is represented as [0, 1, 0, 0], the vector of "delay" is represented as [0, 0, 1, 0], and the vector of "airplane" is represented as [0, 0, 0, 1]. Then, the word vectors of monitoring warning information A and monitoring warning information B are calculated, the word vector of monitoring warning information A is: [1, 1, 1, 0]; the word vector of monitoring warning information B is: [0, 1, 1, 1], and then the word vector is used to calculate the cosine similarity between the monitoring warning information, and then the similarity data is obtained. Finally, different monitoring warning information is marked as similar according to the similarity threshold, and the monitoring warning information greater than the similarity threshold is aggregated, and the aggregated warning information is added to the aggregated warning list, and the monitoring warning information less than the similarity threshold is added to the empty aggregated warning list. The calculation method of the similarity data can be: first calculate the dot product of the two word vectors: A*B=∑(A i *B i ), and then calculate the norm of the two word vectors, which can be calculated by the following formula: Among them, n represents the nth element in the word vector. Finally, use the above data to calculate the cosine similarity: Where θ represents the angle between the two word vectors. If the cosine similarity is equal to 1, it means that monitoring warning information A and monitoring warning information B are exactly the same; if the cosine similarity is close to 0, it means that there is no similarity between monitoring warning information A and monitoring warning information B; if the cosine similarity is equal to -1, it means that monitoring warning information A and monitoring warning information B are completely opposite. Among them, the cosine similarity of monitoring warning information A and monitoring warning information B can be calculated by the following formula:

[0060]

[0061] This embodiment can effectively reduce redundant warning information and improve the processing efficiency of the system by aggregating the monitoring warning information.

[0062] Aggregation processing of monitoring and warning information is achieved by judging similarity data. Optionally, in the warning information processing method provided in the embodiment of the present application, P monitoring and warning information are aggregated according to L similarity data to obtain T aggregated warning information, including: obtaining a similarity threshold, and judging whether the L similarity data are greater than the similarity threshold; in the case that there are similarity data greater than the similarity threshold in the L similarity data, obtaining T target similarity data, wherein each target similarity data refers to similarity data greater than the similarity threshold; merging a group of monitoring and warning information associated with each target similarity data to obtain T aggregated warning information.

[0063] Specifically, when using similarity data to aggregate monitoring warning information, first determine the similarity threshold to compare whether the calculated similarity data is greater than the threshold. If there is similarity data greater than the similarity threshold, obtain target similarity data greater than the similarity threshold, and merge the monitoring warning information associated with these target similarity data to obtain the corresponding aggregated warning information. This embodiment can effectively reduce the amount of monitoring warning information, reduce the workload of operation and maintenance personnel, improve the accuracy and operability of warning information, and improve the management efficiency of the system by using the similarity threshold to judge the similarity data.

[0064] According to the needs of a specific time period of the airport, the warning sensitivity can be dynamically adjusted. Optionally, in the method for processing warning information provided in an embodiment of the present application, before sending the monitoring warning information contained in the X group warning information set to the client, the method also includes: obtaining a preset processing rule, and determining whether the time period associated with the M monitoring warning information is a preset time period, wherein the preset processing rule is used to adjust the response sensitivity of the client; when the time period associated with the M monitoring warning information is the preset time period, sending a first prompt message to the client according to the preset processing rule; when the time period associated with the M monitoring warning information is not the preset time period, sending a second prompt message to the client according to the preset processing rule, wherein the emergency level of the first prompt message is higher than the emergency level of the second prompt message.

[0065] Specifically, in actual operation, the operation and maintenance personnel may have different sensitivities to warning information of different levels in different time periods, and the equipment may also be restarted and maintained on a daily basis, which may lead to inappropriate monitoring warning information. Therefore, before sending the corresponding monitoring warning information to the client, you can first obtain the customized preset processing rules, and then use the rules to dynamically adjust the warning sensitivity, and then determine whether the time period associated with the monitoring warning information is the preset time period. For example, the preset time period may be a peak time period. If the time period associated with the monitoring warning information is a peak time period, then according to the preset processing rules, you can send the first prompt information to the client indicating that the warning information needs to be processed in real time. Conversely, if the time period associated with the monitoring warning information is a non-peak time period, then you can send the second prompt information to the client indicating that only the warning information needs to be processed periodically. For example, Figure 5 is a schematic diagram of a method for executing a preset processing rule according to an embodiment of the present application, such as Figure 5 As shown, if the departure system collects new monitoring warning information, it can use the preset processing rules to parse it and determine whether the time period associated with the monitoring warning information is the preset time period. If it is in the peak period, the client can be directly notified to process the monitoring warning information with the warning level of "urgent". On the contrary, if it is not in the peak period, the client can be notified to record the monitoring warning information with the warning level of "important", "normal" and "minor". This embodiment uses the time period associated with the monitoring warning information to send different prompt information to the client, thereby flexibly adjusting the working hours of the operation and maintenance personnel and reducing labor costs.

[0066] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0067] This embodiment also provides a method for processing early warning information. Figure 6 is a schematic diagram of an optional method for processing early warning information provided in an embodiment of the present application, such as Figure 6 As shown, the method includes:

[0068] In order to solve the problem of early warning information processing in modern airport departure software monitoring systems and improve information processing efficiency and accuracy, we can first collect historical early warning information, standardize the data, clean unstructured data, identify and remove noise, and ensure the quality of input data, laying the foundation for subsequent intelligent processing. Then, we use the processed historical early warning information to train the target classification model, and integrate the trained target classification model into the intelligent classification system, which enables the system to intelligently classify early warning information in a real-time environment.

[0069] When the probe collects new warning information, the warning text vector corresponding to the warning information is input into the intelligent classification system. The target classification model in the system divides the warning information into four levels: "urgent", "important", "ordinary" and "minor" for different processing and responses.

[0070] Furthermore, after the warning information is classified, similar or low-priority warning information can be aggregated by setting time windows and similarity thresholds, thereby reducing redundant warnings and reducing the processing burden of operation and maintenance personnel. At the same time, user rules can be set according to the airport's operating hours and the needs of specific time periods, and then the warning information can be filtered based on the user rules, and the warning information notification can be sent to the client used by the operation and maintenance personnel to dynamically adjust the warning sensitivity. For example, during non-peak hours, unless it is an "emergency" level warning, the response sensitivity to other levels of warnings is reduced.

[0071] It should be noted that for the warning issued after the target classification model performs classification, the operation and maintenance personnel can provide feedback based on the accuracy of the warning results. For example, through the warning control panel, the operation and maintenance personnel can query the current warning, historical warning, warning level, and label the warning feedback. The warning information can be marked as "important", "unimportant" or "false alarm". The labeled warning information will be used for subsequent training and optimization of the target classification model in the future to make it more in line with the operation and maintenance personnel's definition of the warning level, thereby improving the accuracy of the warning.

[0072] Figure 7 is a schematic diagram of an optional target classification model optimization method provided in an embodiment of the present application, such as Figure 7 As shown, the self-assessment scheduling timer can evaluate the accuracy of the warning results, that is, the accuracy of the target classification model according to the feedback from the operation and maintenance personnel. When the accuracy of the target classification model is lower than 90%, the operation and maintenance personnel can be notified to conduct intensive training on the target classification model and make appropriate adjustments to the model parameters. Otherwise, wait for the next feedback from the operation and maintenance personnel to continuously improve the model accuracy.

[0073] This embodiment obtains warning information, processes the information to obtain a warning text vector, uses a target classification model to process the warning text vector to obtain a warning information set, divides the warning information set into warning levels, and then processes the warning information sets of different warning levels in a targeted manner, thereby achieving the effect of improving the accuracy and efficiency of processing warning information.

[0074] The embodiment of the present application also provides a device for processing early warning information. It should be noted that the device for processing early warning information in the embodiment of the present application can be used to execute the method for processing early warning information provided in the embodiment of the present application. The device for processing early warning information provided in the embodiment of the present application is introduced below.

[0075] Figure 8 is a schematic diagram of a device for processing early warning information provided in an embodiment of the present application, such as Figure 8 As shown, the device includes: a first acquiring unit 81, a second acquiring unit 82, a third acquiring unit 83, and a first sending unit 84.

[0076] The first acquisition unit 81 is used to acquire M monitoring warning information generated by the departure system, and process each monitoring warning information to obtain M groups of warning text vectors, where M is a positive integer;

[0077] The second acquisition unit 82 is used to acquire a target classification model, input M groups of warning text vectors into the target classification model, and output N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer;

[0078] The third acquisition unit 83 is used to obtain a warning level threshold, and divide the N groups of warning information sets according to the warning level threshold to obtain X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z and X are positive integers;

[0079] The first sending unit 84 is used to send the monitoring warning information contained in the X group warning information set to the client, aggregate the monitoring warning information contained in the Z group warning information set to obtain the Z group aggregated warning information set, and send the Z group aggregated warning information set to the client, wherein the client is used to process the monitoring warning information contained in each group warning information set.

[0080] The warning information processing device provided in the embodiment of the present application obtains M monitoring warning information generated by the departure system through the first acquisition unit 81, processes each monitoring warning information, and obtains M groups of warning text vectors, wherein M is a positive integer; the second acquisition unit 82 obtains the target classification model, inputs the M groups of warning text vectors into the target classification model, and outputs N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer; the third acquisition unit 83 obtains the warning level threshold, divides the N groups of warning information sets according to the warning level threshold, and obtains X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z, X are positive integers; the first sending unit Element 84 sends the monitoring warning information contained in the X group of warning information sets to the client, aggregates the monitoring warning information contained in the Z group of warning information sets to obtain the Z group of aggregated warning information sets, and sends the Z group of aggregated warning information sets to the client, wherein the client is used to process the monitoring warning information contained in each group of warning information sets, thereby solving the problems of low processing efficiency and low processing accuracy when processing warning information in related technologies, by acquiring monitoring warning information, processing the information to obtain a warning text vector, using a target classification model to process the warning text vector to obtain a warning information set, dividing the warning information set into warning levels, and then performing targeted processing on warning information sets of different warning levels, thereby achieving the effect of improving the accuracy and efficiency of processing warning information.

[0081] Optionally, in the warning information processing device provided in the embodiment of the present application, the first acquisition unit 81 includes: a first processing module, which is used to segment the monitoring warning information to obtain a group of segmented text data; a second processing module, which is used to convert the group of segmented text data into integer identifiers to obtain a group of processed sequence data, and perform sequence completion processing on the group of processed sequence data to obtain a group of text sequence data; an input module, which is used to input a group of text sequences into a word embedding model to obtain a group of warning text vectors associated with the monitoring warning information.

[0082] Optionally, in the warning information processing device provided in the embodiment of the present application, the second acquisition unit 82 includes: the target classification model is trained through the following steps: a first acquisition module, used to obtain Y historical warning information and warning levels associated with Y historical warning information, pre-process the Y historical warning information to obtain Y groups of historical warning vectors, and the model training set is composed of Y groups of historical warning vectors and warning levels associated with Y historical warning information, where Y is a positive integer; a second acquisition module, used to obtain a preset classification model, and use a target optimizer to initialize the preset classification model to obtain an initialized classification model; a training module, used to train the initialized classification model with the model training set to obtain a trained classification model, and obtain the output result of the trained classification model, and optimize the trained classification model based on the output result to obtain a target classification model.

[0083] Optionally, in the warning information processing device provided in the embodiment of the present application, the second acquisition unit 82 includes: a third acquisition module, used to obtain a preset classification result, calculate the cross entropy function between the output result and the preset classification result, and obtain cross entropy data; a first calculation module, used to calculate the parameter gradient of the cross entropy data based on the back propagation algorithm, wherein the parameter gradient refers to the gradient of the loss function associated with the trained classification model with respect to the model parameters; a fourth acquisition module, used to obtain a first-order matrix vector, calculate a second-order matrix vector using the parameter gradient and the first-order matrix vector, and update the second-order matrix vector using the target optimizer to obtain an updated matrix vector, wherein the first-order matrix vector is determined by the model parameters; a determination module, used to determine an updated loss function using the updated matrix vector, and when the updated loss function is less than a preset threshold, determine an updated classification model according to the updated loss function until the updated loss function is greater than or equal to the preset threshold, and determine the updated classification model as the target classification model.

[0084] Optionally, in the warning information processing device provided in the embodiment of the present application, the first sending unit 84 includes: a fifth acquisition module, which is used to obtain K monitoring warning information associated with a set of warning information sets, and add the K monitoring warning information to the pending warning list, where K is a positive integer; a sixth acquisition module, which is used to obtain a time window threshold, and filter the K monitoring warning information according to the time window threshold to obtain P monitoring warning information, where P is a positive integer; a second calculation module, which is used to calculate the similarity between every two monitoring warning information in the P monitoring warning information, and obtain L similarity data, where L is a positive integer; a third processing module, which is used to aggregate the P monitoring warning information according to the L similarity data to obtain T aggregated warning information, and the T aggregated warning information constitute a set of aggregated warning information sets, where T is a positive integer.

[0085] Optionally, in the warning information processing device provided in the embodiment of the present application, the first sending unit 84 includes: a seventh acquisition module, used to obtain a similarity threshold and determine whether L similarity data are greater than the similarity threshold; an eighth acquisition module, used to obtain T target similarity data when there is similarity data greater than the similarity threshold in the L similarity data, wherein each target similarity data refers to similarity data greater than the similarity threshold; a fourth processing module, used to merge a group of monitoring warning information associated with each target similarity data to obtain T aggregated warning information.

[0086] Optionally, in the warning information processing device provided in the embodiment of the present application, the device also includes: a fourth acquisition unit, used to obtain a preset processing rule before sending the monitoring warning information included in the X group warning information set to the client, and determine whether the time period associated with the M monitoring warning information is a preset time period, wherein the preset processing rule is used to adjust the response sensitivity of the client; a second sending unit, used to send a first prompt message to the client according to the preset processing rule when the time period associated with the M monitoring warning information is the preset time period; a third sending unit, used to send a second prompt message to the client according to the preset processing rule when the time period associated with the M monitoring warning information is not the preset time period, wherein the emergency level of the first prompt message is higher than the emergency level of the second prompt message.

[0087] The above-mentioned warning information processing device includes a processor and a memory. The above-mentioned first acquisition unit 81, second acquisition unit 82, third acquisition unit 83, first sending unit 84, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0088] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the problems of low processing efficiency and low processing accuracy in the related technology when processing warning information are solved by adjusting kernel parameters.

[0089] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0090] An embodiment of the present invention provides a computer storage medium, which is used to store a program. When the program is running, the device where the computer storage medium is located is controlled to execute a method for processing early warning information.

[0091] Fig. 9is a schematic diagram of an electronic device provided according to an embodiment of the present application, such as Fig. 9 As shown, an embodiment of the present invention provides an electronic device, the electronic device 90 includes a processor, a memory, and a program stored in the memory and executable on the processor, the processor is used to execute a computer-readable instruction, wherein the computer-readable instruction executes a method for processing early warning information when it is executed. The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0092] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of a method for processing early warning information in each embodiment of the present application.

[0093] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0094] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0095] 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 comprising 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.

[0096] 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.

[0097] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0098] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0099] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0100] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0101] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for processing early warning information, characterized in that: include: Obtain M monitoring warning information generated by the departure system, process each monitoring warning information, and obtain M groups of warning text vectors, where M is a positive integer; Obtain a target classification model, input the M groups of warning text vectors into the target classification model, and output N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer; Obtaining a warning level threshold, dividing the N groups of warning information sets according to the warning level threshold to obtain X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z and X are positive integers; The monitoring warning information contained in the X group warning information set is sent to the client, the monitoring warning information contained in the Z group warning information set is aggregated to obtain Z group aggregated warning information sets, and the Z group aggregated warning information sets are sent to the client, wherein the client is used to process the monitoring warning information contained in each group warning information set.

2. The method according to claim 1, characterized in that Each monitoring warning information is processed to obtain M groups of warning text vectors including: For a monitoring warning information, segmenting the monitoring warning information to obtain a group of segmented text data; The group of segmented text data is converted into integer identifiers to obtain a group of processed sequence data, and the group of processed sequence data is completed to obtain a group of text sequence data; The set of text sequences is input into a word embedding model to obtain a set of warning text vectors associated with the monitoring warning information.

3. The method according to claim 1, characterized in that The target classification model is trained by the following steps: Obtain Y historical warning information and warning levels associated with the Y historical warning information, pre-process the Y historical warning information to obtain Y groups of historical warning vectors, and the Y groups of historical warning vectors and the warning levels associated with the Y historical warning information constitute a model training set, where Y is a positive integer; Obtaining a preset classification model, and initializing the preset classification model using a target optimizer to obtain an initialized classification model; The initialized classification model is trained by the model training set to obtain a trained classification model, and an output result of the trained classification model is obtained. The trained classification model is optimized based on the output result to obtain the target classification model.

4. The method according to claim 3, characterized in that Optimizing the trained classification model based on the output result to obtain the target classification model includes: Obtaining a preset classification result, calculating a cross entropy function between the output result and the preset classification result, and obtaining cross entropy data; Calculating a parameter gradient of the cross entropy data based on a back propagation algorithm, wherein the parameter gradient refers to a gradient of a loss function associated with the trained classification model with respect to a model parameter; Obtaining a first-order matrix vector, calculating a second-order matrix vector using the parameter gradient and the first-order matrix vector, and updating the second-order matrix vector using the target optimizer to obtain an updated matrix vector, wherein the first-order matrix vector is determined by the model parameters; The updated loss function is determined using the updated matrix vector. When the updated loss function is less than a preset threshold, an updated classification model is determined based on the updated loss function until the updated loss function is greater than or equal to the preset threshold, and the updated classification model is determined as the target classification model.

5. The method according to claim 1, characterized in that Aggregate the monitoring warning information contained in the Z group warning information set to obtain the Z group aggregated warning information set including: For a set of warning information sets, obtain K monitoring warning information associated with the set of warning information sets, and add the K monitoring warning information to the pending warning list, where K is a positive integer; Obtaining a time window threshold, and filtering the K monitoring warning information according to the time window threshold to obtain P monitoring warning information, where P is a positive integer; Calculate the similarity between every two pieces of monitoring and warning information in the P pieces of monitoring and warning information to obtain L pieces of similarity data, where L is a positive integer; The P monitoring warning information are aggregated according to the L similarity data to obtain T aggregated warning information, and the T aggregated warning information constitute a group of aggregated warning information sets, where T is a positive integer.

6. The method according to claim 5, characterized in that The P monitoring warning information are aggregated according to the L similarity data to obtain T aggregated warning information including: Obtain a similarity threshold, and determine whether the L similarity data are greater than the similarity threshold; In the case that there is similarity data greater than the similarity threshold value among the L similarity data, T target similarity data are acquired, wherein each target similarity data refers to similarity data greater than the similarity threshold value; A set of monitoring warning information associated with each target similarity data is merged to obtain the T aggregated warning information.

7. The method according to claim 1, characterized in that Before sending the monitoring warning information included in the X group warning information set to the client, the method further includes: Obtaining a preset processing rule, and determining whether the time period associated with the M monitoring warning information is a preset time period, wherein the preset processing rule is used to adjust the response sensitivity of the client; When the time period associated with the M monitoring warning information is the preset time period, sending a first prompt message to the client according to the preset processing rule; When the time period associated with the M monitoring warning information is not the preset time period, a second prompt message is sent to the client according to the preset processing rule, wherein the emergency level of the first prompt message is higher than the emergency level of the second prompt message.

8. A device for processing early warning information, characterized in that: include: The first acquisition unit is used to acquire M monitoring warning information generated by the departure system, and process each monitoring warning information to obtain M groups of warning text vectors, where M is a positive integer; A second acquisition unit is used to acquire a target classification model, input the M groups of warning text vectors into the target classification model, and output N groups of warning information sets, wherein each group of warning information sets corresponds to a warning level, and N is a positive integer; A third acquisition unit is used to obtain a warning level threshold, and divide the N groups of warning information sets according to the warning level threshold to obtain X groups of warning information sets and Z groups of warning information sets, wherein the warning level corresponding to the X group of warning information sets is higher than the warning level corresponding to the Z group of warning information sets, Z=NX, Z and X are positive integers; The first sending unit is used to send the monitoring warning information included in the X group warning information set to the client, aggregate the monitoring warning information included in the Z group warning information set to obtain Z group aggregated warning information set, and send the Z group aggregated warning information set to the client, wherein the client is used to process the monitoring warning information included in each group warning information set.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for processing warning information described in any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the warning information processing method described in any one of claims 1 to 7.

Images