WO2024082767A1

WO2024082767A1 - Gain adjustment method and apparatus, computer device and storage medium

Info

Publication number: WO2024082767A1
Application number: PCT/CN2023/110153
Authority: WO
Inventors: 王一凡; 陈金利
Original assignee: 三维通信股份有限公司
Priority date: 2022-10-21
Filing date: 2023-07-31
Publication date: 2024-04-25
Also published as: CN115688569A

Abstract

A gain adjustment method and apparatus, a computer device and a storage medium. The gain adjustment method comprises: acquiring a signal to be adjusted; determining a first gain value on the basis of the power of said signal; inputting said signal into a gain determination model, so as to obtain a second gain value, the gain determination model being obtained by means of deep learning training; and obtaining a target gain value on the basis of the first gain value and the second gain value.

Description

Gain adjustment method, device, computer equipment and storage medium

Related Applications

This application claims priority to Chinese patent application number 202211296132.4, filed on October 21, 2022, entitled “Gain adjustment method, device, computer equipment and storage medium”, the entire text of which is hereby incorporated by reference.

Technical Field

The present application relates to the field of communication technology, and in particular to a gain adjustment method, device, computer equipment and storage medium.

Background technique

Traditional signal gain technology does not have the ability to predict power, and generally adjusts the signal gain at a constant rate based on experience. This method requires more lag time to adjust to the appropriate gain when the signal changes greatly and changes quickly.

The related technology is to adjust the gain by fast decay and slow release. However, the gain decay speed or release speed of this method is a constant value designed based on experience, which may not be able to respond to signal changes in time in the actual system.

Summary of the invention

According to various embodiments of the present application, a gain adjustment method, apparatus, computer device, and storage medium are provided.

In a first aspect, the present application provides a gain adjustment method. The method comprises:

Obtain the signal to be adjusted;

Determining a first gain value based on the power of the signal to be adjusted;

Inputting the signal to be adjusted into a gain determination model obtained through deep learning training to obtain a second gain value, where the second gain value is a predicted gain value at the next moment;

A target gain value is obtained based on the first gain value and the second gain value.

In one embodiment, obtaining a target gain value based on the first gain value and the second gain value comprises:

Acquire an accuracy evaluation result at a previous moment, where the accuracy evaluation result is determined based on the second gain value and a standard gain value at the previous moment;

If the accuracy evaluation result meets the preset threshold, the second gain value is used as the target gain value;

If the accuracy evaluation result does not meet the preset threshold, the first gain value is used as the target gain value.

In one embodiment, obtaining the accuracy evaluation result of the previous moment includes: The first gain value is used as a standard gain value; and the accuracy score is determined based on the degree of deviation between the second gain value at a previous moment and the standard gain value.

In one of the embodiments, after obtaining the target gain value based on the first gain value and the second gain value, the method further includes: updating the model parameters of the gain determination model based on the accuracy evaluation result.

In one of the embodiments, after obtaining the accuracy evaluation result of the previous moment, the method further includes: if the accuracy evaluation result meets a reset threshold, restoring the gain determination model to an initial model.

In one of the embodiments, determining the first gain value based on the power of the signal to be adjusted includes: comparing the power of the signal to be adjusted with a preset power; and determining the first gain value based on the comparison result.

In one of the embodiments, before inputting the signal to be adjusted into the gain determination model to obtain the second gain value, it also includes: obtaining an initial model, wherein the initial model is a long short-term memory network; training the initial model using a reference signal and a reference gain value as a training set to obtain the gain determination model.

In a second aspect, the present application also provides a gain adjustment device. The device comprises:

An acquisition module, used for acquiring a signal to be adjusted;

A first determining module, configured to determine a first gain value based on the power of the signal to be adjusted;

A second determination module, configured to input the signal to be adjusted into a gain determination model to obtain a second gain value, wherein the gain determination model is obtained through deep learning training;

The gain determination module is configured to obtain a target gain value based on the first gain value and the second gain value.

In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

Obtain the signal to be adjusted;

Determining a first gain value based on the power of the signal to be adjusted;

In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

Obtain the signal to be adjusted;

Determining a first gain value based on the power of the signal to be adjusted;

The details of one or more embodiments of the present application are set forth in the following drawings and description. Other features, objects, and advantages of the present application will become apparent from the description, drawings, and claims.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to better describe and illustrate the embodiments and/or examples of the inventions disclosed herein, reference may be made to one or more drawings. The additional details or examples used to describe the drawings should not be considered as limiting the scope of the disclosed inventions, the embodiments and/or examples currently described, and any of the best modes of these inventions currently understood.

FIG. 1 is a diagram showing an application environment of a gain adjustment method according to some embodiments.

FIG. 2 is a flow chart of a gain adjustment method according to some embodiments.

FIG3 is a schematic diagram of the prediction principle of a long short-term memory network according to some embodiments.

FIG. 4 is a structural block diagram of a gain determination model of a gain adjustment method according to some embodiments.

FIG. 5 is a block diagram of a mid-gain adjustment system according to some embodiments.

FIG6 is a structural block diagram of a gain adjustment device according to some embodiments.

FIG. 7 is a diagram showing the internal structure of a computer device according to some embodiments.

In the accompanying drawings: 102, terminal; 104, server; 600, gain adjustment device; 610, acquisition module; 620, first determination module; 630, second determination module; 640, gain determination module; 700, computer device; 701, processor; 702, internal memory; 703, non-volatile storage medium; 704, system bus; 705, I/O interface; 706, communication interface; 707, input device; 708, display unit.

Detailed ways

In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

The gain adjustment method provided in the embodiment of the present application can be applied in the application environment shown in Figure 1. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the terminal 102 can be but is not limited to various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

In one embodiment, as shown in FIG2 , a gain adjustment method is provided, comprising the following steps:

Step S201, obtaining a signal to be adjusted.

It can be understood that the signal to be adjusted is a real-time signal currently obtained. At this time, the amplitude of the signal to be adjusted changes. The gain of the signal to be adjusted needs to be adjusted according to the situation of the signal to be adjusted to ensure the stability of the output of the signal to be adjusted.

Step S202: determining a first gain value based on the power of the signal to be adjusted.

Exemplarily, the method for determining the first gain value is a traditional gain control method, that is, according to the current power value of the signal to be adjusted, the corresponding gain value is determined, and the signal to be adjusted is gained. However, when the signal to be adjusted changes greatly and changes quickly, if the gain value is determined only when the signal to be adjusted is obtained, and then the signal to be adjusted is gained, then the signal to be adjusted has been maintained at the current power for a period of time, and the signal to be adjusted in the previous period of time has not been appropriately gained, or the power of the signal to be adjusted will change again at the next moment, and the currently determined gain value cannot adapt to the signal to be adjusted at the next moment, which can easily lead to unstable output of the signal to be adjusted and produce distortion.

Step S203: input the signal to be adjusted into a gain determination model obtained through deep learning training to obtain a second gain value, where the second gain value is a predicted gain value at the next moment.

Exemplarily, the gain determination model is trained to determine the predicted gain value at the next moment based on the signal to be adjusted at the current moment. It can be understood that the initial model of the deep learning model here can adopt the neural network model in the relevant technology, which is not specifically limited here.

Step S204: obtaining a target gain value based on the first gain value and the second gain value.

It is understandable that after obtaining two gain values in two ways, a comparison can be performed and a more suitable gain value can be determined as the final target gain value. For example, the matching degree between the two gain values and the actual demand can be judged respectively, and a more matching target gain value can be selected to gain the signal to be adjusted.

The above-mentioned gain adjustment method obtains a signal to be adjusted; determines a first gain value based on the power of the signal to be adjusted; inputs the signal to be adjusted into a gain determination model to obtain a second gain value, and the gain determination model is obtained through deep learning training; obtains a target gain value based on the first gain value and the second gain value, combines the deep learning model prediction gain and the traditional power detection adjustment gain, and determines the final target gain value according to the gain values obtained in the two ways. It can respond to changes in the signal in a timely manner and provide appropriate gain. It meets the requirements of high precision and convergence speed by predicting the future situation of the signal to be adjusted, thereby ensuring the stability of the signal to be adjusted.

Step 1: Obtain the accuracy evaluation result of the previous moment, the accuracy evaluation result is based on the The second gain value and the standard gain value are determined;

Step 2: if the accuracy evaluation result meets a preset threshold, the second gain value is used as the target gain value;

Step 3: If the accuracy evaluation result does not meet a preset threshold, the first gain value is used as the target gain value.

In this embodiment, the first gain value or the second gain value is determined as the target gain value according to the accuracy evaluation result at the previous moment. Exemplarily, the accuracy evaluation result may be an accuracy score.

It can be understood that comparing the second gain value at the previous moment with the standard gain value to determine the accuracy evaluation result is actually comparing the power value corresponding to the second gain value at the previous moment with the actually observed power value, wherein the actually observed power value is the output power actually required by the signal to be adjusted. If the accuracy evaluation result meets the preset threshold, it means that the second gain value at the previous moment matches the actually observed gain value, and the second gain value can meet the actual demand, then the second gain value is still used as the final output gain value at the current moment; if the accuracy evaluation result does not meet the preset threshold, it means that the second gain value at the previous moment deviates from the actually observed gain value and cannot meet the signal gain adjustment demand, then the first gain value obtained by the traditional method is used as the final output gain value at this moment.

In the above embodiment, the evaluation result of the second gain value at the previous moment is used as a reference standard to determine whether the second gain value can meet the signal gain requirement, and the first gain value is used as the final output gain when the second gain value cannot meet the requirement, so as to avoid signal distortion when the second gain value deviates from the actual requirement, and can effectively ensure the accuracy of the signal gain and the stability of the system.

In other embodiments, the basis for selecting the first gain value or the second gain value as the target gain value may also be the degree of matching between the first gain value and the actual demand, or comparing the evaluation scores of the first gain value and the second gain value, and selecting the one with the higher evaluation score as the target gain value; in addition, the target gain value may also be determined by taking the first gain value and the second gain value into consideration by taking a weighted sum of the two gain values. It is understandable that the specific method of determining the target gain value based on the first gain value and the second gain value can be determined by the user according to actual needs, and will not be repeated here.

In one embodiment, obtaining the accuracy evaluation result at the previous moment includes:

Step 1, taking the first gain value at the current moment as the standard gain value;

Step 2: determining an accuracy score based on the degree of deviation between the second gain value at a previous moment and the standard gain value.

It can be understood that the first gain value lags behind the signal to be adjusted at the current moment, and it can only meet the signal gain requirements before the current moment. Therefore, the first gain value at the current moment corresponds to the actual observed gain value at the previous moment, that is, the standard gain value at the previous moment.

In other embodiments, the accuracy evaluation result may also be in other forms, such as accuracy level, which is not specifically limited here.

The above embodiment utilizes the hysteresis of the traditional gain control and uses the first gain value at the current moment as the standard gain value at the previous moment, which can accurately reflect the actual demand of the signal gain at the previous moment and make the accuracy evaluation result more reliable.

In one embodiment, determining the accuracy score based on the degree of deviation between the second gain value at the previous moment and the standard gain value includes: scoring according to an inverse proportional function of the difference between the power corresponding to the second gain value at the previous moment and the power corresponding to the standard gain value at the previous moment, the closer the power corresponding to the second gain value at the previous moment is to the power corresponding to the standard gain value at the previous moment, the higher the accuracy score.

It is understandable that in other embodiments, other calculation methods may be used to determine the accuracy score, as long as it can reflect the degree of deviation between the second gain value at the previous moment and the standard gain value, and no specific limitation is made here.

In one embodiment, after obtaining the target gain value based on the first gain value and the second gain value, the method further includes:

Step 1: Update the model parameters of the gain determination model based on the accuracy evaluation result.

It can be understood that the accuracy evaluation result can reflect the deviation between the output result of the gain determination model and the actual demand. According to this deviation, the gain determination model is reinforced learning, and the model parameters of the gain determination model are updated. In real-time operation, the weights can be updated according to the prediction of the gain determination model to adapt to other unconsidered issues in the online operation, which can effectively improve the accuracy of the output results of the gain determination model.

In one embodiment, after obtaining the accuracy evaluation result at the previous moment, the method further includes:

Step 1: If the accuracy evaluation result meets the reset threshold, the gain determination model is restored to the initial model.

It can be understood that if the accuracy evaluation result meets the reset threshold, it means that the gain value output by the gain determination model is far away from the actual required gain value, which means that the current gain determination model has reached a singularity point and cannot provide accurate predictions. Therefore, the gain determination model is reset and retrained. At the same time, before the gain determination model is retrained, the first gain value is used as the target gain value for the final output. The reset threshold can be set by the user according to the actual The requirements are set, and no specific restrictions are given here.

In another embodiment, the gain determination model may be reset when the accuracy evaluation result is less than a preset threshold for multiple times. The specific preset threshold may be set by the user according to actual needs and is not specifically limited here.

In the above embodiment, when it is determined that the prediction result of the gain determination model is inaccurate, the gain determination model is reset, and the first gain value is used as the target gain value of the final output, degenerating to the traditional gain control method to avoid the problem of network locking, thereby protecting the system and making the signal output more stable.

In another embodiment, determining the first gain value based on the power of the signal to be adjusted includes:

Step 1, comparing the power of the signal to be adjusted with a preset power;

Step 2: determine the first gain value based on the comparison result.

It can be understood that the first gain value is determined by the traditional gain control method, that is, based on the power of the current real-time signal and the preset power, it is judged whether the power of the current signal is too large or too small, and the degree of being too large or too small, and the corresponding gain value is output according to the degree of being too large or too small, so as to adjust the power value of the signal to be adjusted.

In another embodiment, before inputting the signal to be adjusted into a gain determination model to obtain a second gain value, the method further includes:

Step 1, obtaining an initial model, wherein the initial model is a long short-term memory network;

Step 2: Use the reference signal and the reference gain value as a training set to train the initial model to obtain the gain determination model.

It can be understood that the reference signal is the signal information obtained based on historical data, and the reference gain value is obtained based on historical data. The corresponding reference signal is the gain value actually required at the next moment. The reference signal and the reference gain value are used as training sets to train the initial model, and the gain determination model is obtained by iterating through parameter updates until the convergence requirements are met.

Long Short-Term Memory (LSTM) is a time-recurrent neural network designed to solve the long-term dependency problem of general RNN (Recurrent Neural Network). LSTM adds memory and forgetting control to RNN. It can provide better predictions for long sequences.

In the above embodiment, a long short-term memory network is used for training and a gain determination model is obtained, which can comprehensively consider the signal amplitude change before the current moment to make a signal amplitude prediction at the next moment, thereby determining the gain value, with better prediction effect and more accurate results.

In other embodiments, the gain determination model can also be a model composed of a combination of a neural network and a gain adjuster, wherein the neural network is used to receive signal input and output the signal power at the next moment, and the gain adjuster is used to determine the required gain value based on the signal power at the next moment and output it.

Please refer to Figure 3, which is a schematic diagram of the prediction principle of the long short-term memory network of an embodiment of the present application. It can be understood that the long short-term memory network makes a prediction of the maximum amplitude of the future signal based on the change of the signal amplitude at the current moment and before the current moment, that is, historical information.

Please refer to Figure 4, which is a structural block diagram of the gain determination model of the gain adjustment method of an embodiment of the present application. Exemplarily, the A-layer network is a long short-term memory network, and the B-layer network is a fully connected network. Specifically, A1 to An are recurrent networks with memory, and B1 to Bn are networks without memory. The A1 layer and the B1 layer are directly connected, and finally summed through C0 to obtain the predicted output. The input-output relationship between the A-layer network and the B-layer network is as follows:

b _t =σ(W _b ·[y _t-1 ,x _t ])
a _t =σ(W _a ·[y _t-1 ,x _t ])

Among them, _xt is the input of An, _yt is the output of An, W is the model parameter weight, σ is the hyperparameter, and is the user-defined prediction bias weight.

The input and output relationship between the B-layer network and the C-layer network is as follows:
m _t = ^AT n _t + c

Among them, m _t is the input of Bn, _nt is the output of C0, A ^T is used to transform the input of C0 into a matrix and reduce the data dimension to obtain the output data of the expected dimension, and c is the model parameter.

It can be understood that the final output of the neural network is a prediction of the signal power, which is between -1 and 1, used to reflect the degree of change in the signal power, and the gain adjuster is used to map -1 to 1 to the actual gain value.

Please refer to FIG. 5 , which is a structural block diagram of a gain adjustment system according to an embodiment of the present application.

In this embodiment, the gain adjustment system includes a long short-term memory network, a gain adjustment module B, a power detection module, a gain adjustment module A, a scoring module, a gain output module and a gain control module, wherein the long short-term memory network is used to provide a predicted value of the signal amplitude according to the signal input, and the gain adjustment module B is used to output the gain B at the next moment according to the predicted value of the signal amplitude; the power detection module is a traditional detection module, which is used to detect the amplitude of the current signal power, and the gain adjustment module A is a traditional gain control module, which is used to output the gain according to the current signal amplitude; the scoring module is used to A score is made through the outputs of gain adjustment module A and gain adjustment module B, as well as the current signal amplitude, to determine which output of gain adjustment module A and gain adjustment module B is better, that is, which one better meets the signal adjustment requirements, and a larger score is given. The score is used as the final gain output and transmitted to the gain control module through the gain output module. The gain control module is used to gain the signal input signal according to the final gain output to obtain an output signal.

Exemplarily, the entire system can be abstracted into an input-output black box, whose overall function is to add a suitable gain to the signal input and then output the signal. First, the input signal passes through the gain calculator to calculate a currently required gain, which is output to the gain control module to control the gain of the current signal and output it. Inside the gain calculator, the signal is divided into two streams. The power detection module and the power adjustment module A are traditional gain controllers, and the long short-term memory network and the gain adjustment module B are a power prediction model. Both output their own determined gain values and score them in the scoring module. A comprehensive gain is given to the gain output module, and the scoring results are also returned to the long short-term memory network for reinforcement learning to optimize the results.

In this embodiment, the input and output relationship of the gain adjustment module B is as follows:

The input of the gain adjustment module B is g, the output of the gain adjustment module B is f(g), and k is set according to parameters related to the maximum output amplitude of the hardware system.

It should be understood that, although the steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear description in this application, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be based on the same inventive concept with other steps or steps or stages in other steps. The embodiment of the present application also provides a gain adjustment device for realizing the gain adjustment method involved in the above-mentioned. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific definition in the one or more gain adjustment device embodiments provided below can refer to the definition of the gain adjustment method in the above text, and will not be repeated here.

In one embodiment, as shown in FIG. 6 , a gain adjustment device 600 is provided, comprising: an acquisition module 610, a first determination module 620, a second determination module 630 and a gain determination module 640, wherein:

The acquisition module 610 is used to acquire the signal to be adjusted.

The first determination module 620 is configured to determine a first gain value based on the power of the signal to be adjusted.

The first determination module 620 is further configured to: compare the power of the signal to be adjusted with a preset power; and determine the first gain value based on the comparison result.

The second determination module 630 is used to input the signal to be adjusted into a gain determination model to obtain a second gain value, and the gain determination model is obtained through deep learning training.

The gain determination module 640 is configured to obtain a target gain value based on the first gain value and the second gain value.

The gain determination module 640 is further configured to:

The gain determination module 640 is further configured to: use the first gain value at the current moment as a standard gain value; and determine an accuracy score based on a degree of deviation between the second gain value at the previous moment and the standard gain value.

The gain adjustment device 600 further includes: an updating module.

An updating module is used to update the model parameters of the gain determination model based on the accuracy evaluation result.

The gain adjustment device 600 further includes: a reset module.

A reset module is used to restore the gain determination model to an initial model if the accuracy evaluation result meets a reset threshold.

The gain adjustment device 600 further includes: a training module.

The training module is used to: obtain an initial model, which is a long short-term memory network; train the initial model using a reference signal and a reference gain value as a training set to obtain the gain determination model.

Each module in the gain adjustment device 600 can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor 701 in the computer device 700 in the form of hardware, or can be stored in the memory in the computer device 700 in the form of software, so that the processor 701 can call and execute the operations corresponding to each module.

In one embodiment, a computer device 700 is provided, which may be a terminal, and its internal structure diagram may be shown in FIG7. The computer device 700 includes a processor 701, a memory, a communication interface 706, a display unit 708, and an input device 707 connected via a system bus 704. The communication interface 706, the display unit 708, and the input device 707 are connected to the system bus 704 via an I/O interface 705. Among them, the processor 701 of the computer device 700 is used to provide computing and control capabilities. The memory of the computer device 700 includes a non-volatile storage medium 703 and an internal memory 702. The non-volatile storage medium 703 stores an operating system and a computer program. The internal memory 702 provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium 703. The communication interface 706 of the computer device 700 is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor 701, a gain adjustment method is implemented. The display unit 708 of the computer device 700 may be a liquid crystal display or an electronic ink display, and the input device 707 of the computer device 700 may be a touch layer covering the display unit 708, or a button, trackball or touchpad provided on the housing of the computer device 700, or an external keyboard, touchpad or mouse.

Those skilled in the art will appreciate that the structure shown in FIG. 7 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device 700 to which the solution of the present application is applied. The specific computer device 700 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device 700 is provided, including a memory and a processor 701, wherein a computer program is stored in the memory, and when the processor 701 executes the computer program, the following steps are implemented:

Obtain the signal to be adjusted;

Determining a first gain value based on the power of the signal to be adjusted;

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor 701, the following steps are implemented:

Obtain the signal to be adjusted;

Determining a first gain value based on the power of the signal to be adjusted;

It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchain, etc., but are not limited to this. The processor 701 involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of this application should be based on the attached claims.

Claims

A gain adjustment method, characterized in that the method comprises:

Obtain the signal to be adjusted;

Determining a first gain value based on the power of the signal to be adjusted;

Inputting the signal to be adjusted into a gain determination model obtained through deep learning training to obtain a second gain value, where the second gain value is a predicted gain value at the next moment;

A target gain value is obtained based on the first gain value and the second gain value.
The method according to claim 1, wherein obtaining a target gain value based on the first gain value and the second gain value comprises:

Acquire an accuracy evaluation result at a previous moment, where the accuracy evaluation result is determined based on the second gain value and a standard gain value at the previous moment;

If the accuracy evaluation result meets the preset threshold, the second gain value is used as the target gain value;

If the accuracy evaluation result does not meet the preset threshold, the first gain value is used as the target gain value.
The method according to claim 2, wherein obtaining the accuracy evaluation result at the previous moment comprises:

Using the first gain value at the current moment as the standard gain value;

The accuracy score is determined based on the degree of deviation between the second gain value at a previous moment and the standard gain value.
The method according to claim 2, wherein after obtaining the target gain value based on the first gain value and the second gain value, the method further comprises:

Model parameters of the gain determination model are updated based on the accuracy evaluation result.
The method according to claim 2, wherein after obtaining the accuracy evaluation result at the previous moment, the method further comprises:

If the accuracy evaluation result meets the reset threshold, the gain determination model is restored to the initial model.
The method according to claim 1, wherein determining the first gain value based on the power of the signal to be adjusted comprises:

Comparing the power of the signal to be adjusted with a preset power;

The first gain value is determined based on the comparison result.
The method according to claim 1, wherein before inputting the signal to be adjusted into the gain determination model to obtain the second gain value, the method further comprises:

Acquire an initial model, wherein the initial model is a long short-term memory network;

The reference signal and the reference gain value are used as a training set to train the initial model to obtain the gain determination model.
A gain adjustment device, characterized in that the device comprises:

An acquisition module, used for acquiring a signal to be adjusted;

A first determining module, configured to determine a first gain value based on the power of the signal to be adjusted;

A second determination module, configured to input the signal to be adjusted into a gain determination model to obtain a second gain value, wherein the gain determination model is obtained through deep learning training;

The gain determination module is configured to obtain a target gain value based on the first gain value and the second gain value.
A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.
A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.