Parameter prediction method, prediction server, prediction system and electronic equipment

Abstract

The present application provides a parameter prediction method, a prediction server, a prediction system and an electronic device. The embodiments of the present application relate to the remote control driving and high-speed fields in the vehicle networking technology. The method includes: obtaining input parameters for predicting multiple parameters; determining a target prediction model for predicting the multiple parameters based on a first prediction error corresponding to multiple prediction models running in parallel and a second prediction error corresponding to a single prediction model running alone; wherein the multiple prediction models correspond to the multiple parameters one-to-one, each prediction model in the multiple prediction models is used to predict the parameters corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time; based on the input parameters and the target prediction model, the multiple parameters are predicted. The solution provided by the present application can not only realize the simultaneous prediction of multiple parameters, but also realize the diversified prediction of the multiple parameters.

Description

Technical Field

[0001] The embodiments of the present application relate to the remote control driving and high-speed fields in vehicle networking technology, and more specifically, to a parameter prediction method, a prediction server, a prediction system and an electronic device. Background Art

[0002] Quality of service (QoS) parameters include parameters in dimensions such as throughput, bandwidth, transmission rate, and latency. For example, in a remote control scenario, since the uplink throughput and latency of the terminal device are both sensitive variables, it is necessary to predict both throughput and latency to improve the remote control effect.

[0003] However, related techniques can only predict a single QoS parameter, such as throughput or latency.

[0004] In addition, different scenarios have different prediction requirements for QoS parameters. For example, when both remote control scenarios and high-speed scenarios need to predict latency, the high-speed scenario has higher requirements for latency prediction accuracy compared to the remote control scenario. However, the existing prediction schemes require the accuracy of QoS parameters to be as high as possible, and there are no relevant schemes for multiple accuracy predictions. Based on this, for some scenarios with lower requirements for parameter prediction accuracy, using prediction algorithms with higher prediction accuracy for prediction may lead to excess computing power and increase prediction costs. Summary of the invention

[0005] The embodiments of the present application provide a parameter prediction method, a prediction server, a prediction system and an electronic device, which can not only realize the simultaneous prediction of multiple parameters, but also realize the diversified prediction of the multiple parameters.

[0006] In a first aspect, an embodiment of the present application provides a parameter prediction method, comprising:

[0007] Obtaining input parameters for predicting multiple parameters;

[0008] Determining a target prediction model for predicting the multiple parameters based on first prediction errors corresponding to the multiple prediction models running in parallel and second prediction errors corresponding to the single prediction model running alone;

[0009] The multiple prediction models correspond to the multiple parameters one by one, each prediction model in the multiple prediction models is used to predict the parameter corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time;

[0010] Based on the input parameters and the target prediction model, the multiple parameters are predicted.

[0011] In a second aspect, an embodiment of the present application provides a prediction server, including:

[0012] An acquisition unit, used for acquiring input parameters for predicting multiple parameters;

[0013] A determination unit, configured to determine a target prediction model for predicting the multiple parameters based on first prediction errors corresponding to the multiple prediction models running in parallel and a second prediction error corresponding to the single prediction model running independently;

[0014] The multiple prediction models correspond to the multiple parameters one by one, each prediction model in the multiple prediction models is used to predict the parameter corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time;

[0015] The prediction unit is used to predict the multiple parameters based on the input parameters and the target prediction model.

[0016] In the third aspect, an embodiment of the present application provides a prediction system, which includes: a terminal device, a network device, a core network device, a prediction server, and a remote control server; wherein the core network device processes the collected data sent by the network device or the terminal device into structured data, and sends the structured data to the prediction server, and the prediction server executes the uplink first aspect method based on the structured data sent by the core network device, and sends the prediction result to the remote control server or the terminal device.

[0017] In a fourth aspect, an embodiment of the present application provides an electronic device, including:

[0018] a processor adapted to implement computer instructions; and,

[0019] A computer-readable storage medium stores computer instructions, wherein the computer instructions are suitable for being loaded by a processor and executing the method of the first aspect.

[0020] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are read and executed by a processor of a computer device, the computer device executes the method of the first aspect above.

[0021] In a sixth aspect, an embodiment of the present application provides a computer program product or a computer program, the computer program product or the computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method of the first aspect above.

[0022] Based on the above technical solution, by introducing multiple prediction models that are allowed in parallel and a single prediction model that runs separately, and designing each of the multiple prediction models as a dedicated prediction model corresponding to a parameter, and designing the single prediction model as a general prediction model for one-time prediction of the multiple parameters, the multiple prediction models and the single prediction model can both achieve simultaneous prediction of multiple parameters. Further, considering that the energy consumption cost of the multiple prediction models running in parallel is greater than the energy consumption cost of the single prediction model running separately, that is, the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model, based on this, the present application introduces the first prediction error corresponding to the multiple prediction models and the second prediction error corresponding to the single prediction model, and based on the first prediction error and the second prediction error, determines the target prediction model for predicting the multiple parameters, which is equivalent to predicting the multiple parameters based on different prediction costs and different prediction accuracies, thereby achieving diverse predictions of the multiple parameters.

[0023] In short, the solution provided in the present application can not only realize the simultaneous prediction of multiple parameters, but also realize the diversified prediction of the multiple parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is an example of the system framework provided in the embodiments of the present application.

[0025] Figure 2 It is a schematic flow chart of the parameter prediction method provided in the embodiment of the present application.

[0026] Figure 3 It is a schematic flow chart of a parameter prediction method under remote control innovation provided in an embodiment of the present application.

[0027] Figure 4 It is a schematic flowchart of the test training process provided in an embodiment of the present application.

[0028] Figure 5 It is a schematic block diagram of the prediction server provided in an embodiment of the present application.

[0029] Figure 6 It is a schematic block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[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.

[0031] The terms "first", "second", "third" and "fourth" etc. in the specification and claims of the present application and the drawings are used to distinguish different objects, rather than to describe a specific order. 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 limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0032] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0033] The following is an introduction and description of the technical fields involved in the embodiments of the present application:

[0034] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0035] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0036] Computer Vision Technology (CV): Computer vision is a science that studies how to make machines "see". To be more specific, it refers to machine vision such as using cameras and computers to replace human eyes to identify and measure targets, and further perform graphic processing so that the computer processing becomes an image that is more suitable for human eye observation or transmission to instrument detection. As a scientific discipline, computer vision studies related theories and technologies, and attempts to establish an artificial intelligence system that can obtain information from images or multi-dimensional data. Computer vision technology usually includes image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous positioning and map construction, and other technologies, as well as common biometric recognition technologies such as face recognition and fingerprint recognition.

[0037] Machine Learning (ML): A multi-disciplinary interdisciplinary subject involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning and other technologies.

[0038] Figure 1 It is an example of the system framework 100 provided in an embodiment of the present application.

[0039] like Figure 1 As shown, the system framework 100 may include a terminal device 111 , a terminal device 112 , a network device 121 , a network device 122 , a core network device 130 , a prediction server 140 , a remote control server 150 , and a remote control cockpit 160 .

[0040] Among them, the terminal device 111 can communicate with the core network device 130 through the network device 121 and / or the network device 122, and the prediction server 140 can predict various parameters of the terminal device 111 (or the terminal device 112) to obtain the prediction result, and send the predicted prediction result to the remote control server 150 or the remote control cockpit 160. The remote control server 150 or the remote control cockpit 160 sends a control instruction to the terminal device 111 (or the terminal device 112) based on the received prediction result to control the terminal device 111 (or the terminal device 112) to perform the corresponding operation. For example, the control instruction is used to control the terminal device 111 (or the terminal device 112) to perform at least one of the following: adjusting the uplink video bit rate, reducing the remote control speed, or safely docking and exiting the remote control operation, etc. Among them, the remote control server 150 can be used to be responsible for information processing, for example, to undertake service requests and responses, and for example, to send control instructions to the terminal device 111 (or the terminal device 112) based on the prediction results fed back by the prediction server 140; the driver in the remote control cockpit 160 can send control instructions to the terminal device 111 (or the terminal device 112) according to the video information displayed on the screen or the prediction results fed back by the prediction server 140.

[0041] Of course, in other alternative embodiments, the remote control server 150 can also analyze the traffic environment where the terminal device 111 (or the terminal device 112) is located based on the obtained prediction results, and send the analysis results to the remote cockpit 160, so that the driver in the remote cockpit 160 can remotely drive the vehicle based on the analysis results.

[0042] Exemplarily, the prediction server 140 can predict the quality of service (QoS) parameters used to characterize the terminal device 111 (or the terminal device 112), and the QoS parameters include information used to characterize the network status of the terminal device 111 (or the terminal device 112), which can also be referred to as network status information. It includes but is not limited to: throughput, signal to interference plus noise ratio (SINR), received signal strength indication (RSSI), reference signal receiving power (RSRP), reference signal receiving quality (RSRQ), and delay. Of course, other parameters of the terminal device 111 (or the terminal device 112) can also be predicted, including but not limited to: the location information of the terminal device, the azimuth of the terminal device and the longitude and latitude information of the terminal device, video resolution, video segment size, video segment download time, etc.

[0043] Exemplarily, the prediction server 140 may predict the current data of the QoS parameters based on the historical data of the QoS parameters. For example, the prediction server 140 may predict the delay of the terminal device 111 (or the terminal device 112) at the current moment based on the historical delay data of the terminal device 111 (or the terminal device 112).

[0044] Exemplarily, the system framework 100 can be applicable to any scenario requiring remote control or requiring prediction of various parameter indicators.

[0045] For example, the system framework 100 can be applied to remote driving or high-speed scenarios. At this time, the terminal device 111 (or the terminal device 112) can be a vehicle, including but not limited to: a manned vehicle, an intelligent networked vehicle, and an unmanned vehicle. The vehicle can establish communication with the network device 121 or the network device 122 through the radio communication subsystem built into its vehicle-mounted unit. The network device 121 or the network device 122 can act as a transmission source. The network device 121 or the network device 122 can be an active communication device, for example, including but not limited to: a 4G / 5G base station, a road side unit (Road Side Unit, RSU), WiFi, etc. Of course, the terminal device 111 (or the terminal device 112) can be a device equipped with one or more of an application (Application, APP), an on-board Internet Protocol (Internet Protocol, IP) camera, and other terminal software. The core network device 130 includes but is not limited to a 4G core network device or a 5G core network (5G Core, 5GC) device, and the 5G core network device can be a 5G cloud core network. The prediction server may be referred to as a QoS prediction server, and the remote control cockpit 160 may be a cockpit for remotely controlling a vehicle.

[0046] Of course, the system architecture can also support various traffic flow services, for example, including but not limited to: unmanned trucks, vehicle-road collaboration, real-time twins, and remote control, etc. This application does not make specific restrictions on this.

[0047] also, Figure 1 Two terminal devices, two network devices, and one core network device are exemplarily shown, but the present application is not limited thereto. For example, in other alternative embodiments, the system framework 100 may include other number of network devices and each network device may include other number of terminal devices within its coverage area.

[0048] It is worth noting that the prediction server 140 is usually only able to predict a single QoS parameter (such as throughput or latency), while actual application scenarios often require predictions for multiple QoS parameters. For example, in a remote control scenario, since the throughput and latency of the uplink of the terminal device are both sensitive variables, both throughput and latency need to be predicted.

[0049] In addition, different scenarios have different prediction requirements for QoS parameters. For example, when both remote control scenarios and high-speed scenarios need to predict latency, the high-speed scenario has higher requirements for latency prediction accuracy compared to the remote control scenario. However, the existing prediction schemes require the accuracy of QoS parameters to be as high as possible, and there are no relevant schemes for multiple accuracy predictions. Based on this, for some scenarios with lower requirements for parameter prediction accuracy, using prediction algorithms with higher prediction accuracy for prediction may lead to excess computing power and increase prediction costs.

[0050] In view of this, the present application provides a parameter prediction method, which enables the prediction server 140 to not only achieve simultaneous prediction of multiple parameters, but also achieve diversified prediction of the multiple parameters. In particular, the prediction server 140 can simultaneously predict multiple parameters such as throughput and latency. Specifically, multiple prediction models running in parallel or an enhanced single prediction model can be used to predict multiple parameters simultaneously, which can not only achieve simultaneous prediction of multiple parameters, but also achieve diversified prediction of the multiple parameters.

[0051] Figure 2 A schematic flow chart of a parameter prediction method 200 according to an embodiment of the present application is shown. The parameter prediction 200 can be performed by any electronic device with data processing capabilities. For example, the electronic device can be implemented as a server. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, and big data and artificial intelligence platforms. The server can be directly or indirectly connected via wired or wireless communication, and the present application does not make specific restrictions here. For example, the parameter prediction 200 can be performed by Figure 1 The prediction server 140 is shown executing.

[0052] like Figure 2 As shown, the method 200 may include some or all of the following contents:

[0053] S210, obtaining input parameters for predicting multiple parameters;

[0054] S220, determining a target prediction model for predicting the multiple parameters based on first prediction errors corresponding to the multiple prediction models running in parallel and second prediction errors corresponding to the single prediction model running alone;

[0055] The multiple prediction models correspond to the multiple parameters one by one, each prediction model in the multiple prediction models is used to predict the parameter corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time;

[0056] S230: Predict the multiple parameters based on the input parameters and the target prediction model.

[0057] Exemplarily, when the target prediction model is the single prediction model, the single prediction model is used to make a one-time prediction of the multiple parameters; wherein, the term "one-time prediction" is intended to explain: by executing a prediction process once using the single prediction model, the predicted values ​​of the multiple parameters can be obtained; or, for the single prediction model, it is only necessary to input the input parameters once to obtain the predicted values ​​of the multiple parameters. Correspondingly, when the target prediction model is the multiple prediction models, each of the multiple prediction models is used to predict the parameters corresponding to each prediction model. In other words, it is necessary to execute multiple prediction processes using the multiple prediction models in a traversal manner to obtain the predicted values ​​of the multiple parameters; or, for each of the multiple prediction models, it is necessary to input the input parameters once, so that each prediction model can output the predicted values ​​of the parameters corresponding to each prediction model.

[0058] In this embodiment, by introducing multiple prediction models that are allowed in parallel and a single prediction model that runs separately, and designing each of the multiple prediction models as a dedicated prediction model corresponding to a parameter, and designing the single prediction model as a general prediction model for one-time prediction of the multiple parameters, the multiple prediction models and the single prediction model can both achieve simultaneous prediction of multiple parameters. Further, considering that the energy consumption cost of the multiple prediction models running in parallel is greater than the energy consumption cost of the single prediction model running separately, that is, the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model, based on this, the present application introduces the first prediction error corresponding to the multiple prediction models and the second prediction error corresponding to the single prediction model, and based on the first prediction error and the second prediction error, determines the target prediction model for predicting the multiple parameters, which is equivalent to predicting the multiple parameters based on different prediction costs and different prediction accuracies, thereby achieving diversified prediction of the multiple parameters.

[0059] In short, the solution provided in the present application can not only realize the simultaneous prediction of multiple parameters, but also realize the diversified prediction of the multiple parameters.

[0060] Taking the remote control scenario as an example, through the solution provided by the present application, the changes in multiple parameters (such as throughput and latency) can be predicted, and the vehicle can be remotely controlled based on the predicted values ​​of the multiple parameters. For example, before the network change moment arrives, corresponding measures can be taken in time based on the predicted values ​​of the multiple parameters (such as adjusting the uplink video bit rate, reducing the remote control speed, safely parking and exiting remote control, etc.), thereby improving the safety of the remote-driven vehicle and enhancing the driving decision-making performance of the remote-driven vehicle.

[0061] It should be understood that the first prediction error and the second prediction error involved in the present application may be parameters or indicators used to characterize the prediction accuracy of the prediction model. For example, the first prediction error and the second prediction error may be relative mean square error (RMSE) or other parameters.

[0062] In addition, the multiple prediction models and the single prediction model involved in the present application can be any model or algorithm with a prediction function, for example, including but not limited to: a traditional learning model, an integrated learning model or a deep learning model. Optionally, the traditional learning model includes but is not limited to: a tree model (regression tree) or a logistic regression (logistic regression, LR) model; the integrated learning model includes but is not limited to: an improved model of the gradient boosting algorithm (XGBoost) or a random forest model; the deep learning model includes but is not limited to: a long short-term memory network (Long Short-Term Memory, LSTM) or a neural network. Of course, in other embodiments of the present application, other machine learning models can also be used, and the present application does not specifically limit this.

[0063] It should also be understood that the present application does not specifically limit the role or use of the prediction results of multiple parameters. For example, in some implementations, after obtaining the prediction results of the multiple parameters, a control instruction for controlling the terminal device can be generated based on the prediction results. The control instruction is used to control the terminal device to perform at least one of the following: adjusting the uplink video bit rate, reducing the remote control speed, or safely parking and exiting the remote control operation, etc. In other implementations, the prediction results of the multiple parameters can also be used to generate a network environment for the terminal device, so that the remote control personnel can remotely control the terminal device based on the analyzed network environment. For example, in a remote driving scenario, the terminal device is a remote driving vehicle, and the network environment is a traffic environment; that is, the prediction results of the multiple parameters can also be used to generate a traffic environment (such as congestion) for the remote driving vehicle, so that the remote control personnel can control the remote driving vehicle based on the analyzed traffic environment (such as congestion).

[0064] In some embodiments, the S220 may include:

[0065] If the first prediction error is greater than or equal to the second prediction error, the single prediction model is determined as the target prediction model; if the first prediction error is less than the second prediction error, the multiple prediction models are determined as the target prediction model or the single prediction model is determined as the target prediction model.

[0066] Exemplarily, if the first prediction error is greater than or equal to the second prediction error, it means that the prediction accuracy of the multiple prediction models is less than or equal to the prediction accuracy of the single prediction model. In addition, since the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model, determining the single prediction model as the target prediction model, compared with determining the multiple prediction models as the target prediction model, can control the prediction cost of the multiple parameters while ensuring the prediction accuracy.

[0067] Exemplarily, if the first prediction error is smaller than the second prediction error, it means that the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model. In addition, the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model. At this time, the single prediction model is determined as the target prediction model, which is equivalent to that when the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model, the target prediction model can be determined by preferentially controlling the prediction cost of the multiple parameters.

[0068] Exemplarily, if the first prediction error is smaller than the second prediction error, it means that the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model. In addition, the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model. At this time, the multiple prediction models are determined as the target prediction model, which is equivalent to, when the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model, the target prediction model can be determined by preferentially ensuring the prediction accuracy of the multiple parameters.

[0069] In some embodiments, if the first prediction error is smaller than the second prediction error, based on the second prediction error, the multiple prediction models are determined as the target prediction model or the single prediction model is determined as the target prediction model.

[0070] Exemplarily, if the first prediction error is smaller than the second prediction error, then when the second prediction error meets the accuracy requirements of the multiple parameters, the single prediction model is determined as the target prediction model; when the second prediction error does not meet the accuracy requirements of the multiple parameters, the multiple prediction models are determined as the target prediction model.

[0071] In some embodiments, if the second prediction error is less than or equal to a preset error, the single prediction model is determined as the target prediction model; if the second prediction error is greater than the preset error, the multiple prediction models are determined as the target prediction model.

[0072] Exemplarily, the preset error may represent the accuracy requirement of the plurality of parameters.

[0073] Exemplarily, if the second prediction error is less than or equal to the preset error, it means that the prediction accuracy of the single prediction model can meet the accuracy requirement to a certain extent. At this time, the single prediction model is determined as the target prediction model, which can control the prediction cost of the multiple parameters while ensuring the prediction accuracy of the multiple parameters.

[0074] Exemplarily, if the second prediction error is greater than the preset error, it means that the prediction accuracy of the single prediction model may not be able to meet the accuracy requirements to a certain extent. At this time, determining the multiple prediction models as the target prediction models can prioritize the prediction accuracy of the multiple parameters by sacrificing the prediction cost.

[0075] Of course, the accuracy requirements of the multiple parameters may also be expressed as parameters in other forms, and this application does not make any specific limitation on this.

[0076] For example, in other alternative embodiments, if the second prediction error is within the preset error range, the single prediction model is determined as the target prediction model; if the second prediction error is outside the preset error range, the multiple prediction models are determined as the target prediction model. In other words, the accuracy requirements of the multiple parameters can be represented by the preset error range. Specifically, if the second prediction error is within the preset error range, it means that the prediction accuracy of the single prediction model can meet the accuracy requirements to a certain extent. At this time, the single prediction model is determined as the target prediction model, which can control the prediction cost of the multiple parameters on the basis of ensuring the prediction accuracy of the multiple parameters. If the second prediction error is outside the preset error range, it means that the prediction accuracy of the single prediction model may not meet the accuracy requirements to a certain extent. At this time, the multiple prediction models are determined as the target prediction model, which can give priority to ensuring the prediction accuracy of the multiple parameters by sacrificing the prediction cost.

[0077] In some embodiments, if the first prediction error is smaller than the second prediction error, based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model.

[0078] In this embodiment, if the first prediction error is smaller than the second prediction error, it means that the prediction cost of the multiple prediction models is high and the prediction accuracy is high, and the prediction cost of the single prediction model is low and the prediction accuracy is low; in this case, since the application scenarios of the multiple parameters can reflect the accuracy requirements of the multiple parameters to a certain extent, this embodiment determines the target prediction model by introducing the application scenarios of the multiple parameters, which can ensure that the prediction accuracy of the target prediction model is consistent with the accuracy requirements of the multiple parameters, avoiding the problem of low prediction accuracy of the multiple parameters or excessively high prediction cost of the multiple parameters due to inconsistency between the prediction accuracy of the target prediction model and the accuracy requirements of the multiple parameters.

[0079] Exemplarily, the application scenarios of the multiple parameters are scenarios determined by accuracy requirements of the multiple parameters.

[0080] In this embodiment, the application scenarios of the multiple parameters are designed to be scenarios determined by the accuracy requirements of the multiple parameters, which can ensure that the prediction accuracy of the target prediction model is consistent with the accuracy requirements of the multiple parameters, thereby avoiding the problem of too low prediction accuracy of the multiple parameters or too high prediction cost of the multiple parameters due to the inconsistency between the prediction accuracy of the target prediction model and the accuracy requirements of the multiple parameters.

[0081] In some embodiments, if the application scenario of the multiple parameters is a scenario in which the terminal device applies the multiple parameters, the multiple prediction models are determined as the target prediction models; if the application scenario of the multiple parameters is a scenario in which the prediction server applies the multiple parameters, the single prediction model is determined as the target prediction model.

[0082] In the present embodiment, taking into account that different devices usually have different requirements for the prediction accuracy of the multiple parameters, the embodiment of the present application distinguishes the application scenarios of the multiple parameters by using the devices applying the multiple parameters, which not only reduces the classification of the application scenarios of the multiple parameters, thereby reducing the complexity of determining the target prediction model, but also ensures that the prediction accuracy of the target prediction model is consistent with the accuracy requirements of the multiple parameters, thereby avoiding the problem of too low prediction accuracy of the multiple parameters or too high prediction cost of the multiple parameters due to the inconsistency between the prediction accuracy of the target prediction model and the accuracy requirements of the multiple parameters.

[0083] Exemplarily, if the application scenario of the multiple parameters is a scenario in which the terminal device applies the multiple parameters, it means that the multiple parameters are parameters that can affect the user experience. At this time, when both the multiple prediction models and the single prediction model are available, the multiple prediction models can be used as the target prediction model to prioritize the prediction accuracy of the multiple parameters.

[0084] Exemplarily, if the application scenario of the multiple parameters is a scenario in which the prediction server applies the multiple parameters, it means that the multiple parameters may not be parameters that affect the user experience. At this time, when both the multiple prediction models and the single prediction model are available, the single prediction model can be used as the target prediction model to prioritize controlling the prediction cost of the multiple parameters.

[0085] Of course, in other alternative embodiments, if the first prediction error is less than the second prediction error, the multiple prediction models may be determined as the target prediction model or the single prediction model may be determined as the target prediction model based on the second prediction error and the application scenario of the multiple parameters. This application does not make specific limitations on this.

[0086] Exemplarily, if the first prediction error is less than the second prediction error, and the second prediction error is less than or equal to the preset error, then based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model. In other words, if the first prediction error is less than the second prediction error, it means that the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model. If the second prediction error is less than or equal to the preset error, it means that the prediction accuracy of the single prediction model can meet the accuracy requirements to a certain extent. At this time, based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model, which can make the target prediction model a model that matches the application scenarios of the multiple parameters, and can avoid the problem of too low prediction accuracy or too high prediction cost when the target prediction model does not match the application scenarios of the multiple parameters, thereby improving the prediction effect.

[0087] Exemplarily, if the first prediction error is smaller than the second prediction error, and the second prediction error is within a preset error range, then based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model. In other words, if the first prediction error is smaller than the second prediction error, it means that the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model. If the second prediction error is within the preset error range, it means that the prediction accuracy of the single prediction model can meet the accuracy requirements to a certain extent. At this time, based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model, so that the target prediction model can be a model that matches the application scenarios of the multiple parameters, and the problem of too low prediction accuracy or too high prediction cost caused by the mismatch between the target prediction model and the application scenarios of the multiple parameters can be avoided, thereby improving the prediction effect.

[0088] In some embodiments, S210 may include:

[0089] Obtain structured data of terminal devices;

[0090] The structured data includes m historical moments and a parameter set of each of the m historical moments; the parameter set includes the plurality of parameters, and m is an integer greater than 0;

[0091] Obtain the first historical moment that is before the moments corresponding to the multiple parameters and whose interval length is greater than or equal to the preset length;

[0092] Based on the m historical moments, taking the first historical moment as the starting moment, n historical moments are selected in chronological order; n is an integer greater than 0 and less than or equal to m;

[0093] Based on the structured data, obtain n parameter sets corresponding to the n historical moments;

[0094] Based on the target prediction model and the n parameter sets, the input parameter is determined.

[0095] In this embodiment, the input parameters are determined by n parameter sets. Since the parameter sets are relatively easy to obtain, the acquisition cost of the input parameters can be controlled. In addition, when the value of n is an integer greater than 1, it is equivalent to determining the input parameters by multiple parameter sets. At this time, the input parameters can not only reflect the parameter set of the first historical moment in the past, but also improve the change trend of the parameter set before the first historical moment, enrich the information used to predict the multiple parameters, and improve the prediction accuracy of the multiple parameters.

[0096] Exemplarily, the parameter set of each historical moment includes the values ​​of the multiple parameters at each historical moment. Further, the parameter set of each historical moment may also include the values ​​of the auxiliary parameters at each historical moment. For example, the parameter set at time t0 includes the values ​​of the multiple parameters at time t0, and further, may also include the values ​​of the auxiliary parameters at time t0. Optionally, the multiple parameters include but are not limited to: network status information and throughput for characterizing the network status of the terminal device. The network status information includes but is not limited to: signal to interference plus noise ratio (SINR), received signal strength indication (RSSI), reference signal receiving power (RSRP), reference signal receiving quality (RSRQ), delay. The auxiliary information includes but is not limited to: location information of the terminal device, azimuth of the terminal device and longitude and latitude information of the terminal device, video resolution, video segment size, video segment download time, etc.

[0097] Exemplarily, the structured data may be two-dimensional structured data of historical moments and parameter sets.

[0098] Exemplarily, the structured data may be data obtained by processing time series data by a terminal device or a network device, or may be data obtained by processing time series data by a core network device or a prediction server. The time series data may be a sequence formed by counting certain indicators or parameters at different times and arranging them in chronological order.

[0099] Exemplarily, the time series data may be data collected by a terminal device or a network device.

[0100] Exemplarily, the method of processing time series data into received data includes but is not limited to: various data processing technologies for converting unstructured data into structured data. For example, the data processing technologies mainly include the following three:

[0101] 1. Decision Tree:

[0102] Decision trees are widely used in various commercial application software and systems, such as product data storage, transaction logs, etc. This method requires manual feature extraction, which is cumbersome and requires a lot of manpower for data labeling.

[0103] 2. Deep Learning:

[0104] Deep learning can find the semantic features of these structured data, and by solving the problems of manual data cleaning and preparation, find automated methods with little or no human intervention to make the data scalable.

[0105] 3. Structured data cleaning:

[0106] Structured data cleaning is to clean up some erroneous data according to certain rules. The task of data cleaning is to filter out those data that do not meet the requirements, and hand over the filtering results to the business department to confirm whether to filter them out or correct them by the business unit before extraction.

[0107] In some embodiments, if the target prediction model is the single prediction model, a second historical moment before the first historical moment is obtained; based on the m historical moments, k historical moments are selected forward in chronological order with the second historical moment as the starting moment; k is an integer greater than 0 and less than or equal to m; based on the structured data, k parameter sets corresponding to the k historical moments are obtained; and a differential operation is performed on the n parameter sets and the k parameter sets to obtain the input parameters.

[0108] Exemplarily, input parameters of the multiple prediction models are different from input parameters of the single prediction model.

[0109] Exemplarily, after obtaining the n parameter sets, it can be determined whether to process the n parameter sets based on whether the target prediction model is the single prediction model. If the target prediction model is the single prediction model, the output of the target prediction model is the predicted values ​​of the multiple parameters. Since the correlation between the predicted values ​​of the multiple parameters and the n parameter sets is large when the target prediction model predicts the multiple parameters based on historical data, the prediction accuracy of the multiple parameters is affected. In view of this, the present application eliminates or reduces the correlation between the predicted values ​​of the multiple parameters and the n parameter sets by differential operations, which can improve the prediction accuracy of the multiple parameters.

[0110] In some embodiments, if the target prediction model is the multiple prediction models, the n parameter sets are determined as the input parameters.

[0111] For example, after obtaining the n parameter sets, it can be determined whether to process the n parameter sets based on whether the target prediction model is the single prediction model. If the target prediction model is the multiple prediction models, the n parameter sets are directly determined as the input parameters of each prediction model in the multiple prediction models.

[0112] In this embodiment, considering that the target prediction model is the multiple prediction models, the multiple prediction models are dedicated prediction models corresponding to the parameters, that is, the prediction accuracy of the multiple prediction models is relatively high. At this time, the n parameter sets can be directly determined as the input parameters to reduce the prediction complexity of the multiple parameters.

[0113] In some embodiments, before S210, the method 200 may further include:

[0114] Get training data set and test data set;

[0115] Training the multiple prediction models and the single prediction model based on the training data set;

[0116] Testing each of the multiple prediction models based on the test data set to obtain multiple prediction errors;

[0117] Based on the multiple prediction errors, obtaining the first prediction error;

[0118] The single prediction model is tested based on the test data set to obtain the second prediction error.

[0119] In this embodiment, after the multiple prediction models and the single prediction model are trained based on the training set, the multiple prediction models and the single prediction model can also be tested based on the test set, so that the subsequent prediction process can determine the target prediction model for prediction based on the first prediction error and the second prediction error obtained by the test.

[0120] Figure 3 It is a schematic flow chart of a parameter prediction method 300 under remote control innovation provided by an embodiment of the present application. The parameter prediction 300 can be interactively executed by terminal devices, network devices, core network devices, prediction servers and remote control servers, wherein the prediction server can be executed by any electronic device with data processing capabilities. For example, the electronic device can be implemented as a server. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, and big data and artificial intelligence platforms. The server can be directly or indirectly connected via wired or wireless communication, and the present application does not make specific restrictions here. For example, the parameter prediction 300 can be performed by Figure 1 The system architecture 100 shown performs.

[0121] like Figure 3 As shown, the method 300 may include:

[0122] S310, start.

[0123] S320, the terminal device obtains time series data and converts the time series data into structured data.

[0124] Exemplarily, the terminal device may be a vehicle, including but not limited to: a manned vehicle, an intelligent connected vehicle, and an unmanned vehicle.

[0125] Exemplarily, the time series data may be a sequence formed by counting certain indicators or parameters at different times and arranging them in chronological order.

[0126] Exemplarily, the certain indicator or parameter includes network status information and throughput for characterizing the network status of the terminal device. The network status information includes, but is not limited to: Signal to Interference plus Noise Ratio (SINR), Received Signal Strength Indication (RSSI), Reference Signal Receiving Power (RSRP), Reference Signal Receiving Quality (RSRQ), and latency. Furthermore, the certain indicator or parameter may also include, but is not limited to: location information of the terminal device, azimuth of the terminal device and longitude and latitude information of the terminal device, video resolution, video segment size, video segment download time, etc.

[0127] Exemplarily, the method by which the terminal device processes the time series data into received data includes, but is not limited to: various data processing technologies for converting unstructured data into structured data. For example, the data processing technologies mainly include the following three:

[0128] 1. Decision Tree:

[0129] Decision trees are widely used in various commercial application software and systems, such as product data storage, transaction logs, etc. This method requires manual feature extraction, which is cumbersome and requires a lot of manpower for data labeling.

[0130] 2. Deep Learning:

[0131] Deep learning can find the semantic features of these structured data, and by solving the problems of manual data cleaning and preparation, find automated methods with little or no human intervention to make the data scalable.

[0132] 3. Structured data cleaning:

[0133] Structured data cleaning is to clean up some erroneous data according to certain rules. The task of data cleaning is to filter out those data that do not meet the requirements, and hand over the filtering results to the business department to confirm whether to filter them out or correct them by the business unit before extraction.

[0134] S330, the terminal device sends the structured data to the network device via an uplink.

[0135] Exemplarily, the terminal device may upload the acquired collected data to the network device via a 5G uplink as structured data. The network device may include one or more of a 4G / 5G base station, RSU, WiFi, etc.

[0136] It is worth noting that this embodiment intends to use the terminal device to obtain the collected data, and send the collected data to the core network device through the network device. In other alternative embodiments, the collected data can also be obtained through the network device (such as a base station), and the obtained collected data can be sent to the core network device in real time. In this case, the operator needs to participate and provide relevant data support, which can reduce the software and hardware requirements of the terminal device.

[0137] S340, the network device forwards the structured data to the prediction server through the core network device.

[0138] Exemplarily, the network device forwards the structured data to the prediction server through the core network device using the relevant interface to train multiple prediction models and a single prediction model. The core network device includes but is not limited to a 4G core network device or a 5G core network device, and the 5G core network device can be a 5G cloud core network.

[0139] S350, the prediction server preprocesses the structured data to obtain a training set and a test set.

[0140] Exemplarily, the prediction server processes the structured data into a sample set, and divides the sample set into a training set and a test set.

[0141] Exemplarily, the process of preprocessing structured data by the prediction server can be understood as: the prediction server processes the structured data into inputs of multiple prediction models or a single prediction model and configures a corresponding label for each input, wherein each input and the corresponding label can be used to form a sample.

[0142] S360: The prediction server trains multiple prediction models running in parallel and a single prediction model running independently based on the training set.

[0143] Exemplarily, the multiple prediction models are two LSTM (Parallel-LSTM) networks running in parallel, the single prediction model is an enhanced LSTM (Improved-LSTM) network, the prediction server receives structured data and preprocesses the structured data, and then trains the Parallel-LSTM and Improved-LSTM based on the training set.

[0144] Exemplarily, when the prediction server trains the Parallel-LSTM and Improved-LSTM based on the training set, the algorithm module may be started first and the training parameters may be configured.

[0145] Exemplarily, when the prediction server configures the training parameters, it includes but is not limited to setting at least one of the following parameters: input features, output features, prediction features, model parameters, algorithm average times, prediction time, model training and testing ratio, number of iterations, model loss function, model optimization function, number of machine learning neurons, etc. Among them, the configuration method can be a default configuration method, or configured by the user, and this application does not specifically limit this.

[0146] It should be understood that the prediction model involved in the embodiment of the present application is not specifically limited. As an example, the prediction model includes but is not limited to: a traditional learning model, an integrated learning model or a deep learning model. Optionally, the traditional learning model includes but is not limited to: a tree model (regression tree) or a logistic regression (logistic regression, LR) model; the integrated learning model includes but is not limited to: an improved model of the gradient boosting algorithm (XGBoost) or a random forest model; the deep learning model includes but is not limited to: a long short-term memory network (Long Short-Term Memory, LSTM) or a neural network. Of course, in other embodiments of the present application, other machine learning models can also be used, and the present application does not specifically limit this. LSTM is a time recursive neural network (RNN), which can be used to solve the gradient disappearance and gradient explosion problems in the long sequence training process compared to ordinary RNN, and can perform better in longer sequences. Systems based on LSTM can learn to translate languages, control robots, image analysis, document summarization, speech recognition image recognition, handwriting recognition, control chat robots, predict diseases, click-through rates and stocks, and synthesize music and other tasks.

[0147] S370, the prediction server tests the multiple prediction models and the single prediction model based on the test set to obtain a first prediction error and a second prediction error.

[0148] Exemplarily, the multiple prediction models are two LSTMs (Parallel-LSTM) running in parallel, the single prediction model is an enhanced LSTM (Improved-LSTM), the prediction server receives structured data and preprocesses the structured data, and then tests the Parallel-LSTM and Improved-LSTM based on the test set. Among them, the first prediction error is the average prediction error of the Parallel-LSTM, which can be recorded as ARMSE1, and the second prediction error is the prediction error of the Improved-LSTM, which can be recorded as RMSE2.

[0149] After the training process and the testing process are performed on the multiple prediction models and the single prediction model, in the actual prediction process, multiple parameters can be predicted based on the first prediction errors corresponding to the multiple prediction models and the second prediction error corresponding to the single prediction model. The prediction process is described below.

[0150] like Figure 3 As shown, the prediction process may include:

[0151] S381, when predicting multiple parameters, the prediction server determines whether the first prediction error is smaller than the second prediction error?

[0152] S382: If the first prediction error is smaller than the second prediction error, the prediction server determines whether the second prediction error is smaller than a preset error.

[0153] Exemplarily, if the first prediction error is smaller than the second prediction error, it means that the prediction accuracy of the multiple prediction models is greater than the prediction accuracy of the single prediction model. In addition, the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model. At this time, the prediction server can further determine the target prediction model used to predict the multiple parameters by judging whether the second prediction error is smaller than the preset error.

[0154] Exemplarily, the preset error may represent the accuracy requirement of the plurality of parameters.

[0155] S383: If the first prediction error is smaller than the second prediction error, and the second prediction error is smaller than the preset error, the prediction server determines the single prediction model as the target prediction model.

[0156] Exemplarily, if the second prediction error is less than or equal to the preset error, it means that the prediction accuracy of the single prediction model can meet the accuracy requirement to a certain extent. At this time, the single prediction model is determined as the target prediction model, which can control the prediction cost of the multiple parameters while ensuring the prediction accuracy of the multiple parameters.

[0157] S384: If the first prediction error is smaller than the second prediction error, and the second prediction error is greater than or equal to the preset error, the prediction server determines the multiple prediction models as target prediction models.

[0158] Exemplarily, if the second prediction error is greater than the preset error, it means that the prediction accuracy of the single prediction model may not be able to meet the accuracy requirements to a certain extent. At this time, determining the multiple prediction models as the target prediction models can prioritize the prediction accuracy of the multiple parameters by sacrificing the prediction cost.

[0159] Of course, the accuracy requirements of the multiple parameters may also be expressed as parameters in other forms, and this application does not make any specific limitation on this.

[0160] For example, in other alternative embodiments, if the second prediction error is within the preset error range, the single prediction model is determined as the target prediction model; if the second prediction error is outside the preset error range, the multiple prediction models are determined as the target prediction model. In other words, the accuracy requirements of the multiple parameters can be represented by the preset error range. Specifically, if the second prediction error is within the preset error range, it means that the prediction accuracy of the single prediction model can meet the accuracy requirements to a certain extent. At this time, the single prediction model is determined as the target prediction model, which can control the prediction cost of the multiple parameters on the basis of ensuring the prediction accuracy of the multiple parameters. If the second prediction error is outside the preset error range, it means that the prediction accuracy of the single prediction model may not meet the accuracy requirements to a certain extent. At this time, the multiple prediction models are determined as the target prediction model, which can give priority to ensuring the prediction accuracy of the multiple parameters by sacrificing the prediction cost.

[0161] S385: If the first prediction error is greater than or equal to the second prediction error, the prediction server determines the single prediction model as the target prediction model.

[0162] Exemplarily, if the first prediction error is greater than or equal to the second prediction error, it means that the prediction accuracy of the multiple prediction models is less than or equal to the prediction accuracy of the single prediction model. In addition, since the prediction cost of the multiple prediction models is greater than the prediction cost of the single prediction model, determining the single prediction model as the target prediction model, compared with determining the multiple prediction models as the target prediction model, can control the prediction cost of the multiple parameters while ensuring the prediction accuracy.

[0163] It is worth noting that, in S382, it is assumed that this embodiment introduces a preset error as a threshold value of the second prediction error.

[0164] Assume that the first prediction error in this embodiment is recorded as ARMSE1, the second preset error is recorded as ARMSE2, and the preset error is recorded as threshold 1. If ARMSE1 is greater than RMSE2, it means that the performance of a single prediction model is better than that of multiple prediction models, that is, the prediction is more accurate, and the single prediction model occupies less memory than multiple prediction models. At this time, we use the model of the single prediction model as the prediction model, and notify the remote control server and / or terminal of the result of the single prediction model as the subsequent prediction result, and take corresponding remote control operations; if ARMSE1 is less than RMSE2, it is necessary to further determine the value of RMSE2. If the value of RMSE2 is less than the threshold value 1, that is, the prediction errors of multiple prediction models and the single prediction model are within the tolerable threshold range, in order to save memory and energy consumption, this embodiment uses the model of the single prediction model as the prediction model, and notifies the remote control server and / or terminal of the result of the single prediction model as the subsequent prediction result, and takes corresponding remote control operations; if the value of RMSE2 is greater than the threshold value 1, this embodiment uses the model of multiple prediction models as the prediction model, and notifies the remote control server and / or terminal of the result of the multiple prediction models as the subsequent prediction result, and takes corresponding remote control operations.

[0165] S391, the prediction server sends the prediction result to the remote control server or the remote cockpit.

[0166] S392, the remote control server and the remote cockpit send control instructions to the terminal device based on the prediction results.

[0167] Exemplarily, after receiving the prediction result, the remote control server or remote cockpit generates a control instruction based on the prediction result and sends the control instruction to the terminal device. For example, in a remote control scenario, the remote control server or remote cockpit can send a control instruction to a terminal device (such as a remotely driven vehicle) via a downlink (such as a 5G-V2X link).

[0168] Exemplarily, after receiving the control instruction sent by the remote control server and the remote cockpit, the terminal device can be controlled to perform corresponding operations based on the control instruction. For example, the terminal device can be controlled to perform at least one of the following based on the received control instruction: adjusting the uplink video bit rate, reducing the remote control speed, safely parking, and exiting the remote control operation, etc.

[0169] S392, end.

[0170] Figure 4 It is a schematic flowchart of the test training process provided in an embodiment of the present application.

[0171] like Figure 4 As shown, the S350 may include:

[0172] S351, the prediction server converts the time series prediction problem into a supervised learning problem and obtains a data set.

[0173] For example, the time series problem can be transformed into a supervised learning problem by setting the prediction time and lag time.

[0174] It is worth noting that time series refers to a sequence formed by counting certain indicators or parameters at different times and arranging them in chronological order. Supervised learning refers to the process of adjusting the parameters of the classifier using a set of samples of known categories to achieve the required performance. Supervised learning can also be called supervised training or teacher learning. Supervised learning is a machine learning task to infer a function from labeled training data. The training data includes a set of training examples. In supervised learning, each instance consists of an input object (usually a vector) and a desired output value (also called a supervisory signal). The supervised learning algorithm analyzes the training data and generates an inferred function that can be used to map out new instances. An optimal solution will allow the algorithm to correctly determine the class labels of unseen instances.

[0175] S352, the prediction server may split the obtained data set into a training set and a test set.

[0176] Exemplarily, the prediction server may preprocess the data set (i.e., the sample set), including data normalization and segmentation (splitting the training set and the test set) operations on the data set. After the prediction server obtains the training set and the test set, it may set training parameters and train the multiple prediction models and the single prediction model based on the training set and the test set.

[0177] Combine the following Figure 4 The training process of the multiple prediction models and the single prediction model (i.e. Figure 3 S360) shown in the figure is used for explanation.

[0178] like Figure 4 As shown, the S360 may include:

[0179] S361, the prediction server predicts the delay and throughput using a first prediction model among multiple prediction models to obtain a first prediction value of the delay.

[0180] S362: The prediction server determines a prediction error of the first prediction model based on the first predicted value of the delay and the actual value of the delay.

[0181] S363: The prediction server predicts the delay and throughput using a second prediction model among the multiple prediction models to obtain a first prediction value of the throughput.

[0182] S364: The prediction server determines a prediction error of the second prediction model based on the first predicted value of the throughput and the actual value of the throughput.

[0183] S365, the prediction server trains the first prediction model and the second prediction model based on the prediction error of the first prediction model and the error of the second prediction model.

[0184] Exemplarily, for S361 to S364, it is assumed that the multiple prediction models are two parallel LSTM (Parallel-LSTM) networks. For the two parallel LSTM networks, two central processing units (CPU) or graphics processing units (GPU) are used to run the two parallel LSTMs in parallel, and the input is two sets of identical structured data (i.e., training sets). For the first LSTM network in the two parallel LSTM networks, its prediction feature is delay, and its output is the first predicted value of delay. For the second LSTM network in the two parallel LSTM networks, its prediction feature is (uplink) throughput, and its output is the first predicted value of throughput; based on this, the prediction server determines the prediction error of the first LSTM network based on the first predicted value of delay and the true value of delay, and trains the first LSTM network based on the prediction error of the first LSTM network, for example, until the prediction error of the first LSTM network converges to a preset threshold. Similarly, the prediction server determines the prediction error of the second LSTM network based on the first predicted value of the throughput and the actual value of the throughput, and trains the second LSTM network based on the prediction error of the second LSTM network, for example, until the prediction error of the second LSTM network converges to a preset threshold.

[0185] S366, the prediction server performs a differential operation on the training set to obtain a differential training set.

[0186] S367, the prediction server predicts the delay and the throughput using a single prediction model to obtain a second prediction value of the delay and a second prediction value of the throughput.

[0187] S368, the prediction server determines the prediction error of the single prediction model based on the second predicted value of the delay, the second predicted value of the throughput, the true value of the delay, and the true value of the throughput.

[0188] S369: The prediction server trains the second prediction model based on the prediction error of the single prediction model.

[0189] Exemplarily, for S366 to S369, it is assumed that the single prediction model is an enhanced LSTM network (Improved-LSTM), which inputs structured data and whose prediction features are (uplink) throughput and latency, that is, the output of the enhanced LSTM network includes a second predicted value of latency and a second predicted value of throughput. Based on this, the prediction server can determine the prediction error of the enhanced LSTM network based on the second predicted value of latency, the second predicted value of throughput, the true value of latency, and the true value of throughput, and train the enhanced LSTM network based on the prediction error of the enhanced LSTM network, for example, until the prediction error of the enhanced LSTM network converges to a preset threshold.

[0190] It is worth noting that, in the present embodiment, during the training of the enhanced LSTM network, a differential operation can be used to eliminate the correlation between the prediction features to avoid serious time lags in the prediction results.

[0191] Combine the following Figure 4 The testing process of the multiple prediction models and the single prediction model (i.e. Figure 3 S370) shown in the figure is used for explanation.

[0192] like Figure 4 As shown, the S370 may include:

[0193] S371, the prediction server uses the first prediction model to test the delay and throughput to obtain a first test value of the delay.

[0194] S372, the prediction server determines the prediction error of the first prediction model based on the first test value of the delay and the true value of the delay.

[0195] S373, the prediction server uses the second prediction model to test the latency and throughput to obtain a first test value of the throughput.

[0196] S374: The prediction server determines a prediction error of the second prediction model based on the first test value of the throughput and the true value of the throughput.

[0197] S375, the prediction server determines a first prediction error based on the prediction error of the first prediction model and the error of the second prediction model.

[0198] S376, the prediction server performs a differential operation on the test set to obtain a differential test set.

[0199] S377, the prediction server predicts the delay and throughput using a single prediction model to obtain a second test value of the delay and a second test value of the throughput.

[0200] S378, the prediction server determines a second prediction error based on the second test value of the delay, the second test value of the throughput, the true value of the delay, and the true value of the throughput.

[0201] Exemplarily, in S371 to S378, assuming that the prediction error of the first prediction model is recorded as ARMSE_1 and the prediction error of the second prediction model is recorded as ARMSE_2, it can be determined that the first prediction error is ARMSE (=(ARSME_1+ARMSE_2) / 2).

[0202] Exemplarily, the second prediction error is RSME2.

[0203] The preferred embodiments of the present application are described in detail above in conjunction with the accompanying drawings. However, the present application is not limited to the specific details in the above embodiments. Within the technical concept of the present application, the technical solution of the present application can be subjected to a variety of simple modifications, and these simple modifications all belong to the protection scope of the present application. For example, the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present application will not further explain various possible combinations. For another example, the various different embodiments of the present application can also be arbitrarily combined, as long as they do not violate the ideas of the present application, they should also be regarded as the contents disclosed in the present application.

[0204] It should also be understood that in the various method embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0205] The method provided in the embodiment of the present application is described above, and the device provided in the embodiment of the present application is described below.

[0206] Figure 5 It is a schematic block diagram of the prediction server 400 provided in an embodiment of the present application.

[0207] like Figure 5 As shown, the prediction server 400 may include:

[0208] An acquisition unit 410, configured to acquire input parameters for predicting multiple parameters;

[0209] A determination unit 420, configured to determine a target prediction model for predicting the multiple parameters based on first prediction errors corresponding to the multiple prediction models running in parallel and a second prediction error corresponding to the single prediction model running independently;

[0210] The multiple prediction models correspond to the multiple parameters one by one, each prediction model in the multiple prediction models is used to predict the parameter corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time;

[0211] The prediction unit 430 is used to predict the multiple parameters based on the input parameters and the target prediction model.

[0212] In some embodiments, the determining unit 420 is specifically configured to:

[0213] If the first prediction error is greater than or equal to the second prediction error, determining the single prediction model as the target prediction model;

[0214] If the first prediction error is smaller than the second prediction error, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model.

[0215] In some embodiments, the determining unit 420 is specifically configured to:

[0216] If the first prediction error is smaller than the second prediction error, based on the second prediction error, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model.

[0217] In some embodiments, the determining unit 420 is specifically configured to:

[0218] If the second prediction error is less than or equal to the preset error, the single prediction model is determined as the target prediction model; if the second prediction error is greater than the preset error, the multiple prediction models are determined as the target prediction model.

[0219] In some embodiments, the determining unit 420 is specifically configured to:

[0220] If the first prediction error is less than the second prediction error, based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model.

[0221] In some embodiments, the determining unit 420 is specifically configured to:

[0222] If the application scenario of the multiple parameters is a scenario in which the terminal device applies the multiple parameters, the multiple prediction models are determined as the target prediction models; if the application scenario of the multiple parameters is a scenario in which the prediction server applies the multiple parameters, the single prediction model is determined as the target prediction model.

[0223] In some embodiments, the acquisition unit 410 is specifically used to:

[0224] Obtain structured data of terminal devices;

[0225] The structured data includes m historical moments and a parameter set of each of the m historical moments; the parameter set includes the plurality of parameters, and m is an integer greater than 0;

[0226] Obtain the first historical moment that is before the moments corresponding to the multiple parameters and whose interval length is greater than or equal to the preset length;

[0227] Based on the m historical moments, taking the first historical moment as the starting moment, n historical moments are selected in chronological order; n is an integer greater than 0 and less than or equal to m;

[0228] Based on the structured data, obtain n parameter sets corresponding to the n historical moments;

[0229] Based on the target prediction model and the n parameter sets, the input parameter is determined.

[0230] In some embodiments, the acquisition unit 410 is specifically used to:

[0231] If the target prediction model is the single prediction model, obtain the second historical moment before the first historical moment; based on the m historical moments, take the second historical moment as the starting moment, and select k historical moments in chronological order; k is an integer greater than 0 and less than or equal to m; based on the structured data, obtain k parameter sets corresponding to the k historical moments; perform differential operations on the n parameter sets and the k parameter sets to obtain the input parameters.

[0232] In some embodiments, the acquisition unit 410 is specifically used to:

[0233] If the target prediction model is the multiple prediction models, the n parameter sets are determined as the input parameters.

[0234] In some embodiments, before acquiring the input parameters for predicting the multiple parameters, the acquiring unit 410 is further configured to:

[0235] Before, the method 200 may further include:

[0236] Get training data set and test data set;

[0237] Training the multiple prediction models and the single prediction model based on the training data set;

[0238] Testing each of the multiple prediction models based on the test data set to obtain multiple prediction errors;

[0239] Based on the multiple prediction errors, obtaining the first prediction error;

[0240] The single prediction model is tested based on the test data set to obtain the second prediction error.

[0241] It should be understood that the device embodiment and the method embodiment may correspond to each other, and similar descriptions may refer to the method embodiment. To avoid repetition, no further description is given here. Specifically, the prediction server 400 may correspond to the corresponding subject in the method 200 to 300 of the embodiment of the present application, and the various units in the prediction server 400 are respectively for implementing the corresponding processes in the method 200 to 300, and for the sake of brevity, no further description is given here.

[0242] It should also be understood that the various units in the prediction server 400 involved in the embodiment of the present application can be respectively or completely merged into one or several other units to constitute, or some (some) of the units can also be split into multiple smaller units in function to constitute, which can achieve the same operation without affecting the realization of the technical effect of the embodiment of the present application. The above-mentioned units are divided based on logical functions. In practical applications, the function of one unit can also be implemented by multiple units, or the function of multiple units is implemented by one unit. In other embodiments of the present application, the prediction server 400 may also include other units. In practical applications, these functions can also be implemented by other units to assist in implementation, and can be implemented by multiple units in collaboration. According to another embodiment of the present application, the prediction server 400 involved in the embodiment of the present application can be constructed by running a computer program (including program code) capable of executing each step involved in the corresponding method on a general computing device of a general-purpose computer including processing elements and storage elements such as a central processing unit (CPU), a graphics processing unit (GPU), a random access storage medium (RAM), and a read-only storage medium (ROM), to construct the prediction server 400 involved in the embodiment of the present application, and to implement the parameter prediction method of the embodiment of the present application. The computer program can be recorded on, for example, a computer-readable storage medium, and loaded into an electronic device through the computer-readable storage medium and run therein to implement the corresponding method of the embodiment of the present application.

[0243] In other words, the units mentioned above can be implemented in hardware form, can be implemented in software form, or can be implemented in the form of a combination of software and hardware. Specifically, the steps of the method embodiment in the embodiment of the present application can be completed by the hardware integrated logic circuit and / or software form of the processor, and the steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to execute, or the hardware and software combination in the decoding processor can be executed. Optionally, the software can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory, and completes the steps in the above method embodiment in combination with its hardware.

[0244] Figure 6 It is a schematic structural diagram of an electronic device 500 provided in an embodiment of the present application.

[0245] like Figure 6 As shown, the electronic device 500 at least includes a processor 510 and a computer-readable storage medium 520. The processor 510 and the computer-readable storage medium 520 may be connected via a bus or other means. The computer-readable storage medium 520 is used to store a computer program 521, which includes computer instructions, and the processor 510 is used to execute the computer instructions stored in the computer-readable storage medium 520. The processor 510 is the computing core and control core of the electronic device 500, which is suitable for implementing one or more computer instructions, and is specifically suitable for loading and executing one or more computer instructions to implement the corresponding method flow or corresponding function.

[0246] As an example, the processor 510 may also be referred to as a central processing unit (CPU) or a graphics processing unit (GPU). The processor 510 may include, but is not limited to, a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like.

[0247] As an example, the computer-readable storage medium 520 may be a high-speed RAM memory, or a non-volatile memory (Non-Volatile Memory), such as at least one disk memory; optionally, it may also be at least one computer-readable storage medium located away from the aforementioned processor 510. Specifically, the computer-readable storage medium 520 includes, but is not limited to: a volatile memory and / or a non-volatile memory. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus random access memory (DR RAM).

[0248] like Figure 6 As shown, the electronic device 500 may further include a transceiver 530 .

[0249] The processor 510 may control the transceiver 530 to communicate with other devices, specifically, to send information or data to other devices, or to receive information or data sent by other devices. The transceiver 530 may include a transmitter and a receiver. The transceiver 530 may further include an antenna, and the number of antennas may be one or more.

[0250] It should be understood that the various components in the communication device 500 are connected via a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus and a status signal bus.

[0251] In one implementation, the electronic device 500 may be any electronic device with data processing capability; the computer-readable storage medium 520 stores a first computer instruction; the processor 510 loads and executes the first computer instruction stored in the computer-readable storage medium 520 to implement Figure 1 The corresponding steps in the method embodiment shown; in the specific implementation, the first computer instruction in the computer-readable storage medium 520 is loaded by the processor 510 and executes the corresponding steps. To avoid repetition, they are not repeated here.

[0252] According to another aspect of the present application, an embodiment of the present application further provides a computer-readable storage medium (Memory), which is a memory device in the electronic device 500 for storing programs and data. For example, a computer-readable storage medium 520. It can be understood that the computer-readable storage medium 520 here can include both the built-in storage medium in the electronic device 500 and the extended storage medium supported by the electronic device 500. The computer-readable storage medium provides a storage space, which stores the operating system of the electronic device 500. In addition, one or more computer instructions suitable for being loaded and executed by the processor 510 are also stored in the storage space, and these computer instructions can be one or more computer programs 521 (including program codes).

[0253] According to another aspect of the present application, an embodiment of the present application further provides a computer program product or a computer program, which includes a computer instruction, and the computer instruction is stored in a computer-readable storage medium. For example, a computer program 521. At this time, the data processing device 500 can be a computer, and the processor 510 reads the computer instruction from the computer-readable storage medium 520, and the processor 510 executes the computer instruction, so that the computer executes the parameter prediction method provided in the above various optional ways.

[0254] In other words, when implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process of the embodiment of the present application is run in whole or in part or the function of the embodiment of the present application is implemented. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center.

[0255] Those of ordinary skill in the art will appreciate that the units and process steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0256] Finally, it should be noted that the above content is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A parameter prediction method, characterized in that: The method is applicable to a prediction server, comprising: Acquire input parameters for predicting a plurality of parameters, wherein the plurality of parameters are used to characterize quality of service (QoS) parameters of a terminal device, and the input parameters include historical data of the QoS parameters; Based on first prediction errors corresponding to multiple prediction models running in parallel and second prediction errors corresponding to a single prediction model running alone, a prediction model whose prediction errors in the multiple prediction models and the single prediction model match application scenarios of the multiple parameters is determined as a target prediction model for predicting the multiple parameters, wherein the application scenarios of the multiple parameters are scenarios determined by accuracy requirements of the multiple parameters; The multiple prediction models correspond to the multiple parameters one by one, each prediction model in the multiple prediction models is used to predict the parameters corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time; Based on the input parameters and the target prediction model, predicting the multiple parameters and obtaining prediction results of the multiple parameters; The method of determining, based on the first prediction errors corresponding to the multiple prediction models running in parallel and the second prediction error corresponding to the single prediction model running alone, a prediction model whose prediction errors in the multiple prediction models and the single prediction model match the application scenarios of the multiple parameters as a target prediction model for predicting the multiple parameters, comprises: If the first prediction error is greater than or equal to the second prediction error, determining the single prediction model as the target prediction model; If the first prediction error is less than the second prediction error, determining the multiple prediction models as the target prediction models or determining the single prediction model as the target prediction model; The step of obtaining input parameters for predicting multiple parameters includes: Obtain structured data of terminal devices; The structured data includes m historical moments and a parameter set of each of the m historical moments; the parameter set includes the multiple parameters, and m is an integer greater than 0; Acquire a first historical moment that is before the moments corresponding to the multiple parameters and whose interval time is greater than or equal to a preset time; Based on the m historical moments, taking the first historical moment as the starting moment, n historical moments are selected in chronological order; n is an integer greater than 0 and less than or equal to m; Based on the structured data, obtaining n parameter sets corresponding to the n historical moments; The input parameters are determined based on the target prediction model and the n parameter sets.

2. The method according to claim 1, characterized in that If the first prediction error is less than the second prediction error, determining the multiple prediction models as the target prediction models or determining the single prediction model as the target prediction model comprises: If the first prediction error is smaller than the second prediction error, based on the second prediction error, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model.

3. The method according to claim 2, characterized in that The step of determining the multiple prediction models as the target prediction models or determining the single prediction model as the target prediction model based on the second prediction error includes: If the second prediction error is less than or equal to the preset error, the single prediction model is determined as the target prediction model; if the second prediction error is greater than the preset error, the multiple prediction models are determined as the target prediction models.

4. The method according to claim 1, characterized in that: If the first prediction error is less than the second prediction error, determining the multiple prediction models as the target prediction models or determining the single prediction model as the target prediction model comprises: If the first prediction error is smaller than the second prediction error, based on the application scenarios of the multiple parameters, the multiple prediction models are determined as the target prediction models or the single prediction model is determined as the target prediction model.

5. The method according to claim 4, characterized in that The step of determining the multiple prediction models as the target prediction models or determining the single prediction model as the target prediction model based on the application scenarios of the multiple parameters includes: If the application scenario of the multiple parameters is a scenario in which the terminal device applies the multiple parameters, the multiple prediction models are determined as the target prediction models; if the application scenario of the multiple parameters is a scenario in which the prediction server applies the multiple parameters, the single prediction model is determined as the target prediction model.

6. The method according to claim 1, characterized in that The step of determining the input parameters based on the target prediction model and the n parameter sets includes: If the target prediction model is the single prediction model, obtaining a second historical moment before the first historical moment; Based on the m historical moments, taking the second historical moment as the starting moment, k historical moments are selected in chronological order; k is an integer greater than 0 and less than or equal to m; Based on the structured data, obtaining k parameter sets corresponding to the k historical moments; A differential operation is performed on the n parameter sets and the k parameter sets to obtain the input parameters.

7. The method according to claim 1, characterized in that The step of determining the input parameters based on the target prediction model and the n parameter sets includes: If the target prediction model is the multiple prediction models, the n parameter sets are determined as the input parameters.

8. The method according to any one of claims 1 to 7, characterized in that Before obtaining the input parameters for predicting the multiple parameters, the method further includes: Get training data set and test data set; training the multiple prediction models and the single prediction model based on the training data set; Testing each prediction model of the multiple prediction models based on the test data set to obtain multiple prediction errors; Based on the multiple prediction errors, obtaining the first prediction error; The single prediction model is tested based on the test data set to obtain the second prediction error.

9. A prediction server, characterized in that: include: An acquisition unit, configured to acquire input parameters for predicting a plurality of parameters, wherein the plurality of parameters are used to characterize quality of service (QoS) parameters of a terminal device, and the input parameters include historical data of the QoS parameters; a determining unit, configured to determine, based on first prediction errors corresponding to the multiple prediction models running in parallel and second prediction errors corresponding to the single prediction model running alone, a prediction model whose prediction errors in the multiple prediction models and the single prediction model match the application scenarios of the multiple parameters as a target prediction model for predicting the multiple parameters, wherein the application scenarios of the multiple parameters are scenarios determined by the accuracy requirements of the multiple parameters; The multiple prediction models correspond to the multiple parameters one by one, each prediction model in the multiple prediction models is used to predict the parameters corresponding to each prediction model, and the single prediction model is used to predict the multiple parameters at one time; Prediction unit, used to: Based on the input parameters and the target prediction model, predicting the multiple parameters and obtaining prediction results of the multiple parameters; The determining unit is specifically used for: If the first prediction error is greater than or equal to the second prediction error, determining the single prediction model as the target prediction model; If the first prediction error is less than the second prediction error, determining the multiple prediction models as the target prediction models or determining the single prediction model as the target prediction model; The acquisition unit is specifically used for: Obtain structured data of terminal devices; The structured data includes m historical moments and a parameter set of each of the m historical moments; the parameter set includes the multiple parameters, and m is an integer greater than 0; Acquire a first historical moment that is before the moments corresponding to the multiple parameters and whose interval time is greater than or equal to a preset time; Based on the m historical moments, taking the first historical moment as the starting moment, n historical moments are selected in chronological order; n is an integer greater than 0 and less than or equal to m; Based on the structured data, obtaining n parameter sets corresponding to the n historical moments; The input parameters are determined based on the target prediction model and the n parameter sets.

10. A prediction system, characterized in that: The prediction system includes: terminal equipment, network equipment, core network equipment, prediction server, and remote control server; wherein, the core network equipment processes the collected data sent by the network equipment or the terminal equipment into structured data, and sends the structured data to the prediction server, and the prediction server executes the parameter prediction method as described in any one of claims 1 to 8 based on the structured data sent by the core network equipment, and sends the prediction result to the remote control server or the terminal equipment.

11. An electronic device, characterized in that: include: a processor adapted to execute a computer program; A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by the processor, the parameter prediction method according to any one of claims 1 to 8 is implemented.

12. A computer-readable storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, enable the computer to perform the parameter prediction method according to any one of claims 1 to 8.

13. A computer program product comprising instructions, characterized in that When the instructions are executed on a computer, the computer is enabled to perform the parameter prediction method according to any one of claims 1 to 8.

Images