Method, device and equipment for predicting demand of home broadband networking product and medium

Abstract

The application discloses a demand prediction method, device and equipment of home broadband networking products and a medium. The method comprises the following steps: acquiring broadband usage information of a user, and generating user features according to the broadband usage information; inputting the user features into a preset demand prediction model; and acquiring demand information of home broadband networking products of the user output by the demand prediction model. The application improves the accuracy of user demand prediction.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, device, and medium for predicting the demand of home broadband networking products. Background Technology

[0002] Home networking uses wired access and wireless coverage to form a small network, allowing multiple smart devices within the home to transmit data, providing great convenience for family members and finding wide application in various fields of the Internet of Things (IoT). Home broadband networking products are primarily used to solve problems such as intranetting for home broadband users, providing whole-house WiFi coverage, and increasing bandwidth speed. Operators typically use demand forecasting models to predict the functional requirements of home networking products.

[0003] However, each household's layout, as well as each user's actual location and usage habits, are different. Therefore, existing technologies cannot effectively identify users with potential networking needs and / or users whose home networking needs optimization, thus failing to accurately provide these users with satisfactory broadband, networking, and wireless access services. Summary of the Invention

[0004] The main objective of this invention is to provide a method, apparatus, device, and medium for predicting the demand of home broadband networking products, aiming to solve the problem of how to accurately provide users with satisfactory broadband, networking, and wireless access services.

[0005] To achieve the above objectives, the present invention provides a demand forecasting method for home broadband networking products, the demand forecasting method for home broadband networking products comprising the following steps:

[0006] Obtain the user's broadband usage information and generate user characteristics based on the broadband usage information;

[0007] Input the user characteristics into a preset demand prediction model;

[0008] Obtain the user's home broadband networking product demand information output by the demand prediction model.

[0009] In one embodiment, the step of inputting the user characteristics into a preset demand prediction model includes:

[0010] The user features are used as input values ​​and sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, the preset weight matrix, the preset bias vector, and the preset activation function of the fully connected layer, and then inputs the output value into the next fully connected layer, until the last fully connected layer.

[0011] The output value of the last fully connected layer is used as the required information.

[0012] In one embodiment, the step of obtaining a user's broadband usage information and generating user characteristics based on the broadband usage information includes:

[0013] Obtain broadband usage information of users on a preset platform, wherein the preset platform includes at least one of a smart gateway platform, a broadband TV service quality monitoring system, and a gigabit passive optical network management platform, and the broadband usage information includes at least one of the following: number of times of buffering, buffering duration, terminal Wi-Fi signal strength, number of connected terminals, and router communication quality.

[0014] User characteristics are generated based on the broadband usage information. These user characteristics include at least one of the following: user bandwidth utilization, terminal Wi-Fi signal strength, number of connected terminals, presence of a low-quality router, average latency of the packet explorer gateway, number of abnormal average latency events, average packet loss rate of the packet explorer gateway, cumulative daily packet loss events, number of application service outages, daily number of application service outages, duration of application service outages, application service playback success rate, number of slow network speed complaints, weak light rate of the optical network unit, user online time, and the offline status of the optical network unit.

[0015] In one embodiment, before the step of inputting the user characteristics into a preset demand prediction model, the method further includes:

[0016] Obtain users' broadband usage history information within a preset time interval;

[0017] Training samples are determined based on the broadband historical usage information;

[0018] A preset neural network model is trained based on the training samples;

[0019] Once the neural network model converges, the converged neural network model will be used as the demand prediction model.

[0020] In one embodiment, the step of using the converged neural network model as the demand prediction model includes:

[0021] The verification sample is determined based on the broadband historical usage information;

[0022] Input the verification sample into the converged neural network model;

[0023] Obtain the output value of the converged neural network and the error value of the validation samples;

[0024] If the error value is less than the preset threshold, the converged neural network model is used as the demand prediction model.

[0025] In one embodiment, the step of training a preset neural network model based on the training samples includes:

[0026] The training samples are used as input values ​​and sequentially input into each fully connected layer of the neural network model. Each fully connected layer obtains its output value based on the input value, the initial weight matrix, the initial bias vector, and the preset activation function, and then inputs the output value into the next fully connected layer, until the last fully connected layer.

[0027] Get the output value of the last fully connected layer;

[0028] The loss value between the output value and the true value corresponding to the training sample is determined according to a preset loss function, and the weight matrix and bias vector are determined according to the loss value;

[0029] The demand forecasting model is determined based on the weight matrix and the bias vector.

[0030] In one embodiment, the step of determining the weight matrix and bias vector based on the loss value includes:

[0031] The extreme value of the loss function is calculated according to a preset algorithm;

[0032] The weight matrix and the bias vector are determined based on the extreme values.

[0033] To achieve the above objectives, the present invention also provides a demand forecasting device for home broadband networking products, the demand forecasting device for home broadband networking products comprising:

[0034] The acquisition module is used to acquire the user's broadband usage information and generate user characteristics based on the broadband usage information;

[0035] The input module is used to input the user characteristics into a preset demand prediction model;

[0036] The determination module is used to obtain the user's home broadband networking product demand information output by the demand prediction model.

[0037] To achieve the above objectives, the present invention also provides a demand forecasting device for home broadband networking products. The demand forecasting device for home broadband networking products includes a memory, a processor, and a demand forecasting program for home broadband networking products stored in the memory and executable on the processor. When the demand forecasting program for home broadband networking products is executed by the processor, it implements the various steps of the demand forecasting method for home broadband networking products as described above.

[0038] To achieve the above objectives, the present invention also provides a computer-readable storage medium storing a demand forecasting program for a home broadband networking product, wherein when the demand forecasting program for the home broadband networking product is executed by a processor, it implements the various steps of the demand forecasting method for the home broadband networking product as described above.

[0039] This invention provides a method, apparatus, device, and medium for predicting the demand for home broadband networking products. It acquires users' broadband usage information and generates user characteristics based on this information. The user characteristics are then input into a preset demand prediction model. Finally, the demand prediction model outputs the user's demand information for home broadband networking products. By determining the user's demand for home broadband networking products through user characteristics, and ensuring that these characteristics closely match the user's broadband usage, the accuracy of demand prediction is improved, facilitating the accurate provision of satisfactory broadband, networking, and wireless access services to users. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the hardware structure of the demand forecasting device for home broadband networking products according to an embodiment of the present invention;

[0041] Figure 2 This is a flowchart illustrating the first embodiment of the demand forecasting method for broadband networking products of the present invention.

[0042] Figure 3 This is a schematic diagram of the logical structure of the demand forecasting model of the present invention;

[0043] Figure 4 This is a detailed flowchart of step S20 of the second embodiment of the demand forecasting method for broadband networking products of the present invention.

[0044] Figure 5 This is a flowchart illustrating the third embodiment of the demand forecasting method for broadband networking products of the present invention.

[0045] Figure 6 This is a schematic diagram of the demand forecasting device for the broadband networking product of the present invention;

[0046] Figure 7 This is a schematic diagram of the demand forecasting device for the broadband networking product of the present invention.

[0047] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0048] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0049] The main solution of this invention is: to obtain the user's broadband usage information and generate user characteristics based on the broadband usage information; to input the user characteristics into a preset demand prediction model; and to obtain the user's home broadband networking product demand information output by the demand prediction model.

[0050] By identifying user characteristics to determine their needs for home broadband networking products, and by ensuring that these user characteristics align with their broadband usage patterns, the accuracy of predicting user needs can be improved, making it easier to provide users with satisfactory broadband, networking, and wireless access services.

[0051] As an implementation solution, demand forecasting devices for home broadband networking products can, for example, Figure 1 As shown.

[0052] The present invention relates to a demand forecasting device for home broadband networking products. The demand forecasting device for home broadband networking products includes: a processor 101, such as a CPU, a memory 102, and a communication bus 103. The communication bus 103 is used to enable communication between these components.

[0053] Memory 102 can be high-speed RAM or stable memory (non-volatile memory), such as disk storage. Figure 1 As shown, the memory 102, which is a computer-readable storage medium, may include a demand forecasting program for home broadband networking products; and the processor 101 can be used to call the demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0054] Obtain the user's broadband usage information and generate user characteristics based on the broadband usage information;

[0055] Input the user characteristics into a preset demand prediction model;

[0056] Obtain the user's home broadband networking product demand information output by the demand prediction model.

[0057] In one embodiment, the processor 101 can be used to call a demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0058] The user features are used as input values ​​and sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, the preset weight matrix, the preset bias vector, and the preset activation function of the fully connected layer, and then inputs the output value into the next fully connected layer, until the last fully connected layer.

[0059] The output value of the last fully connected layer is used as the required information.

[0060] In one embodiment, the processor 101 can be used to call a demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0061] Obtain broadband usage information of users on a preset platform, wherein the preset platform includes at least one of a smart gateway platform, a broadband TV service quality monitoring system, and a gigabit passive optical network management platform, and the broadband usage information includes at least one of the following: number of times of buffering, buffering duration, terminal Wi-Fi signal strength, number of connected terminals, and router communication quality.

[0062] User characteristics are generated based on the broadband usage information. These user characteristics include at least one of the following: user bandwidth utilization, terminal Wi-Fi signal strength, number of connected terminals, presence of a low-quality router, average latency of the packet explorer gateway, number of abnormal average latency events, average packet loss rate of the packet explorer gateway, cumulative daily packet loss events, number of application service outages, daily number of application service outages, duration of application service outages, application service playback success rate, number of slow network speed complaints, weak light rate of the optical network unit, user online time, and the offline status of the optical network unit.

[0063] In one embodiment, the processor 101 can be used to call a demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0064] Obtain users' broadband usage history information within a preset time interval;

[0065] Training samples are determined based on the broadband historical usage information;

[0066] A preset neural network model is trained based on the training samples;

[0067] Once the neural network model converges, the converged neural network model will be used as the demand prediction model.

[0068] In one embodiment, the processor 101 can be used to call a demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0069] The verification sample is determined based on the broadband historical usage information;

[0070] Input the verification sample into the converged neural network model;

[0071] Obtain the output value of the converged neural network and the error value of the validation samples;

[0072] If the error value is less than the preset threshold, the converged neural network model is used as the demand prediction model.

[0073] In one embodiment, the processor 101 can be used to call a demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0074] The training samples are used as input values ​​and sequentially input into each fully connected layer of the neural network model. Each fully connected layer obtains its output value based on the input value, the initial weight matrix, the initial bias vector, and the preset activation function, and then inputs the output value into the next fully connected layer, until the last fully connected layer.

[0075] Get the output value of the last fully connected layer;

[0076] The loss value between the output value and the true value corresponding to the training sample is determined according to a preset loss function, and the weight matrix and bias vector are determined according to the loss value;

[0077] The demand forecasting model is determined based on the weight matrix and the bias vector.

[0078] In one embodiment, the processor 101 can be used to call a demand forecasting program for home broadband networking products stored in the memory 102 and perform the following operations:

[0079] The extreme value of the loss function is calculated according to a preset algorithm;

[0080] The weight matrix and the bias vector are determined based on the extreme values.

[0081] Based on the hardware architecture of the demand forecasting device for the aforementioned home broadband networking product, an embodiment of the demand forecasting method for the home broadband networking product of the present invention is proposed.

[0082] Reference Figure 2 , Figure 2 This is a first embodiment of the demand forecasting method for home broadband networking products of the present invention. The demand forecasting method for home broadband networking products includes the following steps:

[0083] Step S10: Obtain the user's broadband usage information and generate user characteristics based on the broadband usage information.

[0084] Specifically, user broadband usage information may include the number of times buffering occurred, the duration of buffering, the Wi-Fi signal strength of the terminal, the number of connected terminals, and the router communication quality. This information is obtained from a preset platform, which includes a smart gateway platform, a broadband TV service quality monitoring system, and / or a GPON (Gigabit-Capable PON) network management platform.

[0085] Performance parameters related to network quality and service perception in broadband usage information are extracted as user features. These user features include user bandwidth utilization, terminal Wi-Fi signal strength, number of connected terminals, presence of a low-quality router, average ping latency to the gateway, number of instances of abnormal average latency, average packet loss rate when pinging the gateway, daily cumulative packet loss count, number of OTT buffering incidents, daily OTT buffering incident count, OTT buffering duration, OTT playback success rate, number of slow network speed complaints, ONU (Optical Network Unit) low light rate, user internet access duration, and ONU offline status. OTT refers to providing various application services to users via the internet.

[0086] Step S20: Input the user characteristics into a preset demand prediction model.

[0087] Specifically, the user characteristics are input into a preset demand prediction model, which is used to predict users' demand information for home broadband networking products. For example... Figure 3 As shown, the demand prediction model can include six fully connected layers, which can be divided into an input layer, a hidden layer, and an output layer. When the user features have 'a' feature parameters, the number of neurons in each fully connected layer can be 'a', 64, 32, 16, 8, and 1, respectively, where 'a' is the number of neurons in the input layer, '1' is the number of neurons in the output layer, and the rest are the number of neurons in the hidden layers.

[0088] Step S30: Obtain the user's home broadband networking product demand information output by the demand prediction model.

[0089] Specifically, the demand information output by the demand prediction model is obtained. The demand information corresponding to home broadband networking products can be represented by 0 or 1. An output result Y of 0 indicates that the user's demand for home networking products is low; an output result Y of 1 indicates that the user's demand for home networking products is very high. Home networking products can be recommended to users, and users can be accurately provided with satisfactory broadband, networking and wireless access services.

[0090] In this embodiment, the technical solution involves acquiring the user's broadband usage information and generating user characteristics based on that information; inputting the user characteristics into a preset demand prediction model; and obtaining the user's home broadband networking product demand information output by the demand prediction model. By determining the user's demand for home broadband networking products through user characteristics, and by ensuring that the user characteristics closely match the user's broadband usage, the accuracy of predicting user demand is improved, facilitating the accurate provision of satisfactory broadband, networking, and wireless access services to the user.

[0091] Reference Figure 4 , Figure 4This is a second embodiment of the demand forecasting method for broadband networking products of the present invention. Based on the first embodiment, step S20 includes:

[0092] Step S21: The user features are used as input values ​​and sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, the preset weight matrix, the preset bias vector, and the preset activation function of the fully connected layer, and inputs the output value into the next fully connected layer, until the last fully connected layer.

[0093] Step S22: Use the output value of the last fully connected layer as the requirement information.

[0094] Specifically, user characteristics are sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, its preset weight matrix, preset bias vector, and preset activation function, and then inputs this output value into the next fully connected layer, until the last fully connected layer. The output value of the last fully connected layer is then used as the demand information. For example, as shown... Figure 3 As shown, each neuron in the fully connected layer of the demand forecasting model corresponds to a weight matrix and a bias vector. The formula for calculating the output value of each fully connected layer is as follows:

[0095] z (l) =W (l) f (l-1) (z (l-1) )+b (l) ;

[0096] Where l is the number of fully connected layers in the demand forecasting model, and W (l) b is the weight matrix from layer (l-1) to layer l; (l) z is the bias vector from layer (l-1) to layer l; (l) f is the output value of the neuron in layer l. l (·) represents the activation function of the neurons in the l-th layer; where the activation function of the hidden layer can be the ReLU function, and the activation function of the output layer is the sigmoid function.

[0097] In this embodiment, user characteristics are sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, a preset weight matrix, a preset bias vector, and a preset activation function, and then inputs the output value to the next fully connected layer, until the last fully connected layer. The output value of the last fully connected layer is used as the demand information. By determining the user's demand for home broadband networking products through the demand prediction model, the accuracy of recommending home broadband networking products to the user is improved.

[0098] Reference Figure 5 , Figure 5 This is a third embodiment of the demand forecasting method for broadband networking products of the present invention. Based on the first or second embodiment, before step S20, it further includes:

[0099] Step S40: Obtain the user's broadband historical usage information within a preset time interval;

[0100] Step S50: Determine training samples based on the broadband historical usage information;

[0101] Step S60: Train a preset neural network model based on the training samples;

[0102] Step S70: After the neural network model converges, the converged neural network model is used as the demand prediction model.

[0103] Specifically, it retrieves users' historical broadband usage information within a preset time interval, which can be six months. This historical information may include the number of buffering incidents, duration of buffering, terminal Wi-Fi signal strength, number of connected devices, and router communication quality. It can also retrieve historical broadband usage information for users on preset platforms, including smart gateway platforms, broadband TV service quality monitoring systems, and / or GPON network management platforms.

[0104] Training samples are determined based on historical broadband usage information. These training samples can include user characteristics and are labeled with real values, representing the user's actual needs. User characteristics included in the training samples can include parameters such as user bandwidth utilization, terminal Wi-Fi signal strength, number of connected devices, presence of a low-quality router, average ping latency to the gateway, number of abnormal average latency occurrences, average packet loss rate when pinging the gateway, daily cumulative packet loss occurrences, number of OTT buffering incidents, daily OTT buffering incident counts, OTT buffering duration, OTT playback success rate, number of slow network speed complaints, ONU low light rate, user online time, and ONU offline status.

[0105] A pre-defined neural network model is trained using training samples. These samples are sequentially input into each fully connected layer of the model. Each fully connected layer outputs its value based on the input value, its initial weight matrix, initial bias vector, and a pre-defined activation function. This output value is then passed to the next fully connected layer, and so on, until the last fully connected layer is reached. The output value of the last fully connected layer is then obtained. A pre-defined loss function, such as exponential loss, logarithmic loss, or cross-entropy loss, is used to determine the loss between the output value and the true value corresponding to the training sample. The pre-defined loss function can be exponential, logarithmic, or cross-entropy loss. The weight matrix and bias vector are determined based on the loss value. This can be done by determining the weight matrix and bias vector based on the minimum loss value, or by calculating the extreme value of the loss function using a pre-defined algorithm and then determining the weight matrix and bias vector based on the extreme value. The pre-defined algorithm can be gradient descent, Newton's method, or a quasi-Newton method. Finally, the demand prediction model is determined based on the obtained weight matrix and bias vector.

[0106] Once the neural network model converges, it can be directly used as the demand prediction model. Alternatively, validation samples can be determined based on historical broadband usage information, including user characteristics of users who have subscribed to home broadband networking products. The validation samples are input into the converged neural network model, and the error value between the output value of the converged neural network and the validation sample is obtained. If the error value is less than a preset threshold, the converged neural network model is used as the demand prediction model. If the error value is greater than or equal to the preset threshold, the neural network model is retrained by adjusting the range or types of user characteristics.

[0107] In this embodiment, the user's historical broadband usage information within a preset time interval is obtained. Training samples are determined based on this information, and a preset neural network model is trained using these samples. Once the neural network model converges, it is used as a demand prediction model. This demand prediction model, trained using training samples, is used to determine the user's demand for home broadband networking products, thus improving the accuracy of recommending suitable products.

[0108] Reference Figure 6 The present invention also provides a demand forecasting device for home broadband networking products, the demand forecasting device for home broadband networking products comprising:

[0109] The acquisition module 100 is used to acquire the user's broadband usage information and generate user characteristics based on the broadband usage information;

[0110] Input module 200 is used to input the user characteristics into a preset demand prediction model;

[0111] The determination module 300 is used to obtain the user's home broadband networking product demand information output by the demand prediction model.

[0112] In one embodiment, the input module 200 is specifically used for inputting the user characteristics into a preset demand prediction model, in which the input module 200 is used to:

[0113] The user features are used as input values ​​and sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, the preset weight matrix, the preset bias vector, and the preset activation function of the fully connected layer, and then inputs the output value into the next fully connected layer, until the last fully connected layer.

[0114] The output value of the last fully connected layer is used as the required information.

[0115] In one embodiment, the acquisition module 100 is specifically used for: obtaining user broadband usage information.

[0116] Obtain broadband usage information of users on a preset platform, wherein the preset platform includes at least one of a smart gateway platform, a broadband TV service quality monitoring system, and a gigabit passive optical network management platform, and the broadband usage information includes at least one of the following: number of times of buffering, buffering duration, terminal Wi-Fi signal strength, number of connected terminals, and router communication quality.

[0117] User characteristics are generated based on the broadband usage information. These user characteristics include at least one of the following: user bandwidth utilization, terminal Wi-Fi signal strength, number of connected terminals, presence of a low-quality router, average latency of the packet explorer gateway, number of abnormal average latency events, average packet loss rate of the packet explorer gateway, cumulative daily packet loss events, number of application service outages, daily number of application service outages, duration of application service outages, application service playback success rate, number of slow network speed complaints, weak light rate of the optical network unit, user online time, and the offline status of the optical network unit.

[0118] Reference Figure 7 In one embodiment, before inputting the user characteristics into a preset demand prediction model, the demand prediction device for the home broadband networking product further includes:

[0119] The training module 400 is used to acquire users' broadband historical usage information within a preset time interval; determine training samples based on the broadband historical usage information; train a preset neural network model based on the training samples; and use the converged neural network model as the demand prediction model after the neural network model converges.

[0120] In one embodiment, the training module 400 is specifically configured to: use the converged neural network model as the demand prediction model.

[0121] The verification sample is determined based on the broadband historical usage information;

[0122] Input the verification sample into the converged neural network model;

[0123] Obtain the output value of the converged neural network and the error value of the validation samples;

[0124] If the error value is less than the preset threshold, the converged neural network model is used as the demand prediction model.

[0125] In one embodiment, in training a preset neural network model based on the training samples, the training module 400 is specifically used for:

[0126] The training samples are used as input values ​​and sequentially input into each fully connected layer of the neural network model. Each fully connected layer obtains its output value based on the input value, the initial weight matrix, the initial bias vector, and the preset activation function, and then inputs the output value into the next fully connected layer, until the last fully connected layer.

[0127] Get the output value of the last fully connected layer;

[0128] The loss value between the output value and the true value corresponding to the training sample is determined according to a preset loss function, and the weight matrix and bias vector are determined according to the loss value;

[0129] The demand forecasting model is determined based on the weight matrix and the bias vector.

[0130] In one embodiment, the training module 400 is specifically used for determining the weight matrix and bias vector based on the loss value:

[0131] The extreme value of the loss function is calculated according to a preset algorithm;

[0132] The weight matrix and the bias vector are determined based on the extreme values.

[0133] The present invention also provides a demand forecasting device for home broadband networking products. The demand forecasting device for home broadband networking products includes a memory, a processor, and a demand forecasting program for home broadband networking products stored in the memory and executable on the processor. When the demand forecasting program for home broadband networking products is executed by the processor, it implements the various steps of the demand forecasting method for home broadband networking products as described in the above embodiments.

[0134] The present invention also provides a computer-readable storage medium storing a demand forecasting program for a home broadband networking product. When the demand forecasting program for the home broadband networking product is executed by a processor, it implements the various steps of the demand forecasting method for the home broadband networking product as described in the above embodiments.

[0135] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0136] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, system, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, system, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, system, article, or apparatus that includes that element.

[0137] Through the above description of the embodiments, those skilled in the art can clearly understand that the systems described in the embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, parking management device, air conditioner, or network device, etc.) to execute the systems described in the various embodiments of the present invention.

[0138] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A demand forecasting method for home broadband networking products, characterized in that, The demand forecasting method for home broadband networking products includes: Obtain broadband usage information of users on a preset platform, wherein the preset platform includes at least one of a smart gateway platform, a broadband TV service quality monitoring system, and a gigabit passive optical network management platform, and the broadband usage information includes at least one of the following: number of times of buffering, buffering duration, terminal Wi-Fi signal strength, number of connected terminals, and router communication quality. User characteristics are generated based on the broadband usage information. The user characteristics include at least one of the following: user bandwidth utilization, terminal Wi-Fi signal strength, number of connected terminals, presence of a poor-quality router, average latency of the packet explorer gateway, number of abnormal average latency events, average packet loss rate of the packet explorer gateway, cumulative daily packet loss events, number of application service lag events, daily number of application service lag events, duration of application service lag events, application service playback success rate, number of slow network speed complaints, weak light rate of optical network unit, user online time, and offline status of optical network unit. Input the user characteristics into a preset demand prediction model; Obtain the user's home broadband networking product demand information output by the demand prediction model; The step of inputting the user characteristics into a preset demand prediction model includes: The user features are sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, its preset weight matrix, preset bias vector, and preset activation function, and then inputs the output value into the next fully connected layer, until the last fully connected layer. The demand prediction model consists of six fully connected layers. When the user features are 'a' feature parameters, the number of neurons in each fully connected layer is 'a', 64, 32, 16, 8, and 1, respectively. The activation function of the hidden layer of the demand prediction model is the ReLU function, and the activation function of the output layer is the sigmoid function. The training process of the demand prediction model is supervised learning. The output value of the last fully connected layer is used as the requirement information, which is represented by 0 or 1.

2. The demand forecasting method for home broadband networking products as described in claim 1, characterized in that, Before the step of inputting the user characteristics into the preset demand prediction model, the method further includes: Obtain users' broadband usage history information within a preset time interval; Training samples are determined based on the broadband historical usage information; A preset neural network model is trained based on the training samples; Once the neural network model converges, the converged neural network model will be used as the demand prediction model.

3. The demand forecasting method for home broadband networking products as described in claim 2, characterized in that, The step of using the converged neural network model as the demand forecasting model includes: The verification sample is determined based on the broadband historical usage information; Input the verification sample into the converged neural network model; Obtain the output value of the converged neural network and the error value of the validation samples; If the error value is less than the preset threshold, the converged neural network model is used as the demand prediction model.

4. The demand forecasting method for home broadband networking products as described in claim 2, characterized in that, The step of training a preset neural network model based on the training samples includes: The training samples are used as input values ​​and sequentially input into each fully connected layer of the neural network model. Each fully connected layer obtains its output value based on the input value, the initial weight matrix, the initial bias vector, and the preset activation function, and then inputs the output value into the next fully connected layer, until the last fully connected layer. Get the output value of the last fully connected layer; The loss value between the output value and the true value corresponding to the training sample is determined according to a preset loss function, and the weight matrix and bias vector are determined according to the loss value; The demand forecasting model is determined based on the weight matrix and the bias vector.

5. The demand forecasting method for home broadband networking products as described in claim 4, characterized in that, The step of determining the weight matrix and bias vector based on the loss value includes: The extreme value of the loss function is calculated according to a preset algorithm; The weight matrix and the bias vector are determined based on the extreme values.

6. A demand forecasting device for home broadband networking products, characterized in that, The demand forecasting device for the home broadband networking product includes: The acquisition module is used to acquire broadband usage information of users on a preset platform. The preset platform includes at least one of a smart gateway platform, a broadband TV service quality monitoring system, and a gigabit passive optical network management platform. The broadband usage information includes at least one of the following: number of times the connection was interrupted, duration of interruption, terminal Wi-Fi signal strength, number of connected terminals, and router communication quality. User characteristics are generated based on the broadband usage information. The user characteristics include at least one of the following: user bandwidth utilization, terminal Wi-Fi signal strength, number of connected terminals, presence of a poor-quality router, average latency of the packet explorer gateway, number of abnormal average latency events, average packet loss rate of the packet explorer gateway, cumulative daily packet loss events, number of application service lag events, daily number of application service lag events, duration of application service lag events, application service playback success rate, number of slow network speed complaints, weak light rate of optical network unit, user online time, and offline status of optical network unit. The input module is used to input the user characteristics into a preset demand prediction model; The determination module is used to obtain the user's home broadband networking product demand information output by the demand prediction model; The step of inputting the user characteristics into a preset demand prediction model includes: The user features are sequentially input into each fully connected layer of the demand prediction model. Each fully connected layer obtains its output value based on the input value, its preset weight matrix, preset bias vector, and preset activation function, and then inputs the output value into the next fully connected layer, until the last fully connected layer. The demand prediction model consists of six fully connected layers. When the user features are 'a' feature parameters, the number of neurons in each fully connected layer is 'a', 64, 32, 16, 8, and 1, respectively. The activation function of the hidden layer of the demand prediction model is the ReLU function, and the activation function of the output layer is the sigmoid function. The training process of the demand prediction model is supervised learning. The output value of the last fully connected layer is used as the requirement information, which is represented by 0 or 1.

7. A demand forecasting device for home broadband networking products, characterized in that, The demand forecasting device for the home broadband networking product includes a memory, a processor, and a demand forecasting program for the home broadband networking product stored in the memory and executable on the processor. When the demand forecasting program for the home broadband networking product is executed by the processor, it implements the various steps of the demand forecasting method for the home broadband networking product as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a demand forecasting program for a home broadband networking product, which, when executed by a processor, implements the various steps of the demand forecasting method for a home broadband networking product as described in any one of claims 1-5.

Images