Online homestay equipment linkage control method and system based on multi-scene perception

Abstract

The invention discloses an online homestay equipment linkage method based on multi-scene perception, and the method comprises the following steps: obtaining and integrating multi-source sensor data, and generating a fused perception feature; inputting the fused sensing features and the initial running state of the equipment into a scene adaptation controller, and generating an equipment linkage control vector, a scene adaptation control vector and corresponding time sequence information; inputting the control vector, the time sequence information, the sensing feature and the time information into an equipment linkage decision maker and a feedback optimizer, and outputting an optimized feature vector; and according to the optimized feature vector, updating an equipment operation state, and outputting an equipment linkage result. According to the invention, the comfort level of a user is improved, the energy utilization efficiency is also remarkably improved, and the system has a relatively high application value.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular discloses a method and system for controlling the linkage of online homestay equipment based on multi-scenario perception. Background Art

[0002] With the rapid development of smart home and Internet of Things technologies, device linkage methods based on multi-scenario perception are showing great significance in improving user experience and optimizing resource utilization. In the online homestay scenario, in particular, users are increasingly demanding an intelligent living environment. For example, by sensing environmental conditions, user behavior, or usage habits, intelligent linkage between devices can be achieved to provide more comfortable and convenient services. However, existing device linkage technologies still have significant shortcomings in terms of multi-scenario adaptability, intelligence, and user experience optimization.

[0003] After searching, it was found that the patent with publication number CN113467261B proposed a linkage control method based on the operating time of the main device. By monitoring the operating time of the main device, the control information of the linkage device that matches it is determined, and the operating mode of the linkage device is adjusted accordingly. This technical solution simplifies the device control method to a certain extent and improves the user experience in specific scenarios. However, this method mainly relies on single-dimensional operating time data and lacks the ability to comprehensively analyze complex perception data of multiple scenarios (such as ambient temperature, light intensity, user behavior, etc.), making it difficult to meet the needs of diversified scenarios in online homestays. In addition, the system has a slow response speed to scene switching and may not be able to adapt to dynamically changing user needs in real time.

[0004] Another patent with the publication number CN106372412B proposes a method for analyzing the energy efficiency of linked devices. By collecting the energy consumption-related parameters of the linked devices, combined with data preprocessing, framing operations and correlation coefficient calculations, it realizes the troubleshooting of device linkage anomalies and energy efficiency optimization. This technical solution has certain advantages in improving the overall operational stability and energy-saving effects of the system. However, this method mainly focuses on the energy efficiency optimization of the equipment, lacks the ability to deeply perceive user behavior and environmental conditions, and cannot dynamically adjust the device linkage strategy according to specific scenario requirements. At the same time, the solution does not involve the intelligent design of multi-device collaboration, and in actual applications may not meet the complex needs of multi-device linkage in online homestays.

[0005] The above problems show that the existing device linkage technology still has obvious deficiencies in multi-scenario perception capabilities, intelligence, and dynamic adaptability. Especially in the online homestay scenario, users have high requirements for the flexibility, real-time, and personalization of device linkage, and existing technologies are difficult to fully meet these needs. Therefore, there is an urgent need for a device linkage method and system that can integrate multi-dimensional perception data and optimize scene adaptation algorithms to achieve intelligent linkage between devices, thereby improving user experience and reducing operating costs, and meeting the needs of modern online homestays for intelligent management. Summary of the Invention

[0006] In view of this, the purpose of the present invention is to provide a method and system for linking online homestay equipment based on multi-scenario perception to improve the above-mentioned problems.

[0007] A method for linking online homestay equipment based on multi-scenario perception, characterized by comprising the following steps:

[0008] Acquire and integrate multi-source sensor data to generate fused perception features;

[0009] The fused perception features and the initial operating state of the device are input into the scene adaptation controller to generate the device linkage control vector and the scene adaptation control vector as well as the corresponding time series information.

[0010] The control vector, time series information, perception features, and time information are input into the device linkage decision maker and feedback optimizer, and the optimized feature vector is output;

[0011] Update the device operating status according to the optimized feature vector and output the device linkage result.

[0012] Preferably, the acquiring and integrating multi-source sensor data comprises the following steps:

[0013] De-noising and normalizing multi-source sensory data;

[0014] Generate a dynamic weight matrix of the same dimension as the importance distribution information of the data;

[0015] The dynamic weight matrix and the preprocessed data are input into the adaptive feature extraction network to obtain the fused perceptual features.

[0016] Preferably, the scene adaptation controller includes the following sub-steps:

[0017] Obtain the device operation area characteristics and environmental background characteristics based on the initial operation status of the device, and merge them into the regional characteristics of the device-environment pair;

[0018] Use convolutional layers and multi-layer perceptrons to predict device start and stop points, operating intensity points, and scene switching points;

[0019] The three types of key points are sampled to obtain running features, and the time relationship between the three types of points is constructed to generate time features;

[0020] Aggregate operation characteristics and time characteristics to obtain the final device characteristics, environment characteristics and scene characteristics;

[0021] Merge the three features and generate device linkage control vectors and scene adaptation control vectors through a multi-layer perceptron;

[0022] The cosine embedding function is used on the device linkage control vector and the scene adaptation control vector to obtain the time series information of the device control vector and the scene control vector respectively.

[0023] Preferably, the device start and stop points, operation intensity points and scene switching points are normalized to between zero and one through a Sigmoid activation function.

[0024] Preferably, the device linkage decision maker and feedback optimizer are composed of stacked multi-layer dynamic programming networks, and each layer of the network uses the feature vector output by the previous layer to update the control vector of the current layer.

[0025] Preferably, the device linkage result includes device operating status, environmental adaptation status, scene switching time and user satisfaction.

[0026] The embodiment of the present invention further provides an online homestay device linkage system based on multi-scenario perception, which includes a perception data fusion device, a scenario adaptation controller, a device linkage decision maker, and a feedback optimizer, wherein:

[0027] Perception data fusion unit, used to acquire and integrate multi-source sensor data to generate fused perception features;

[0028] The scene adaptation controller is used to input the fused perception features and the initial operating state of the device into the scene adaptation controller to generate the device linkage control vector and the scene adaptation control vector as well as the corresponding time series information;

[0029] The device linkage decision maker is used to input the control vector, time series information, perception features and time information into the device linkage decision maker and feedback optimizer, and output the optimized feature vector;

[0030] The feedback optimizer is used to update the device operating status according to the optimized feature vector and output the device linkage results.

[0031] Preferably, the perception data fusion device is specifically used to: perform denoising and normalization processing on multi-source perception data;

[0032] Generate a dynamic weight matrix of the same dimension as the importance distribution information of the data;

[0033] The dynamic weight matrix and the preprocessed data are input into the adaptive feature extraction network to obtain the fused perceptual features.

[0034] Preferably, the scene adaptation controller is specifically used to:

[0035] Obtain the device operation area characteristics and environmental background characteristics based on the initial operation status of the device, and merge them into the regional characteristics of the device-environment pair;

[0036] Use convolutional layers and multi-layer perceptrons to predict three key points: device start and stop points, operating intensity points, and scene switching points;

[0037] The three types of key points are sampled to obtain running features, and the time relationship between the three types of points is constructed to generate time features;

[0038] Aggregate operation characteristics and time characteristics to obtain device characteristics, environment characteristics and scene characteristics;

[0039] Combine device features, environment features, and scene features and generate device linkage control vectors and scene adaptation control vectors through a multi-layer perceptron;

[0040] The cosine embedding function is used on the device linkage control vector and the scene adaptation control vector to obtain the time series information of the device control vector and the scene control vector respectively.

[0041] Preferably, the device linkage decision maker and feedback optimizer are implemented by stacking multi-layer dynamic programming networks, and output the results of device operation status, environmental adaptation status, scene switching time and user satisfaction respectively.

[0042] In summary, the present invention achieves multi-scenario perception-based device linkage for online homestays through the collaborative work of a perception data amalgamator, a scenario adaptation controller, a device linkage decision maker, and a feedback optimizer. In practical application scenarios, for example, when a user enters a homestay room, the system can automatically adjust the air conditioning temperature, start background music, and adjust the lighting brightness based on real-time perception data, thereby providing a personalized living experience. This intelligent device linkage not only enhances user comfort but also significantly improves energy efficiency, thus possessing high application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1This is a structural diagram of the online homestay equipment linkage system based on multi-scenario perception provided by the first embodiment of the present invention.

[0045] Figure 2 A flowchart of a method for linking online homestay equipment based on multi-scenario perception provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention.

[0047] The present invention proposes a method and system for linking online homestay equipment based on multi-scene perception. The core of the method is to achieve intelligent collaborative linkage between devices through multimodal data fusion and dynamic scene adaptation algorithm. Figure 1 To the attached Figure 2 The specific structure and process of the present invention are described in detail.

[0048] First, if Figure 1 As shown, the multi-scenario perception-based device linkage method and system for online homestays of the present invention consists of four main modules: a perception data fusion unit 1, a scene adaptation controller 2, a device linkage decision maker 3, and a feedback optimizer 4. The data flow and linkage process between these modules constitute the operating mechanism of the entire system. In practical applications, taking an online homestay environment as an example, assume that the environment is equipped with multi-source sensing devices such as temperature sensors, humidity sensors, light sensors, and sound sensors, as well as controllable devices such as air conditioners, humidifiers, lighting, and background music players. After the perception data fusion unit acquires and integrates multi-source sensor data, it initializes the initial operating states of a group of devices. The fused perception data and the initial device states are then input into the scene adaptation controller. Based on the input, the scene adaptation controller 2 generates two sets of control vectors and their time series information. These are then processed by the device linkage decision maker and the feedback optimizer, ultimately outputting an optimized linkage control instruction set, achieving intelligent collaborative linkage between devices.

[0049] Next, each part will be introduced in detail:

[0050] The first part is the perception data fuser 1.

[0051] In this embodiment, the sensory data fusion module 1 consists of a multimodal data preprocessing module and an adaptive feature extraction network. In actual operation, multi-source sensory data includes data from multiple dimensions, such as temperature, humidity, light, and sound. This data first undergoes denoising and normalization in the preprocessing module to ensure data quality and consistency. For example, the raw data collected by a temperature sensor may contain noise or outliers. The preprocessing module uses a filtering algorithm to remove these interferences and normalizes the data to a range of 0 to 1 for subsequent processing. The preprocessed data is used to generate a dynamic weight matrix of the same dimension, which serves as information on the data's importance distribution. The dynamic weight matrix is generated by calculating the relative importance of each sensory data type in the current scene. For example, in a nighttime scene, light data may be less important, while sound data may be more important. The generated dynamic weight matrix is input into the adaptive feature extraction network along with the preprocessed data. After processing through a multi-layer neural network, fused sensory features are ultimately generated. This process not only integrates multi-source data but also provides high-quality feature input for subsequent scene adaptation.

[0052] Part 2, learnable scene adaptation controller 2.

[0053] The main function of the scene adaptation controller 2 is to generate two sets of control vectors: device linkage control vectors and scene adaptation control vectors. To achieve this goal, the controller needs to input the fused perception features and the initial operating state of the device. Taking the air conditioner as an example, assume that the initial state is that the air conditioner is off and the temperature is set to 25°C. The controller first obtains the corresponding device operating area feature F based on the initial state. device and environmental background characteristics F env And combine these two features according to the time dimension to obtain the regional feature F of the device-environment pair device-env , the formula is defined as:

[0054] F device-env =Concat(F device , F env )

[0055] Subsequently, these three features are respectively predicted by a convolutional layer (Conv) and a multi-layer perceptron (MLP) to obtain three types of key points, namely the equipment start and stop points P start , Operation intensity point P intensity and scene switching point P switch Since these key points need to be normalized to between 0 and 1 during sampling, the predicted key points are also normalized using the Sigmoid activation function. The predicted results of these key points can accurately reflect the operating requirements of the device in different scenarios.

[0056] After obtaining the key points, the scene adaptation controller 2 further performs feature sampling and aggregation operations. In order to obtain richer information, the scene adaptation controller 2 samples the three types of key points to obtain operation features, and also constructs the time features of these three types of points. For operation features, the scene adaptation controller 2 uses the device start and stop points P start , Operation intensity point P intensity and scene switching point P switch Equipment-environment regional characteristics F device-env Perform bilinear interpolation sampling operations to generate device operation characteristics F run 、Environmental adaptation characteristics F adapt and scene switching feature F switch The specific calculation steps are as follows:

[0057] F run =Bilinear(P star ,F device -env)

[0058] F adapt =Bilinear(P instensity ,F device -env)

[0059] F switch =Bilinear(P switch ,F device -env)

[0060] Among them, Bilinear represents the bilinear interpolation sampling operation.

[0061] For the time feature, the scene adaptation controller 2 constructs the time relationship of the three types of key points, regards the scene switching point as the origin between the device start and stop points and the operation intensity point, and then connects the device start and stop points and the operation intensity point with the origin to form two corresponding vectors V start and V instensity , the formula is defined as:

[0062] V start =P star -P switch

[0063] V instensity =P instensity -P switch

[0064] Next, the time difference between the two vectors is further calculated to generate a time matrix of the two vectors, which is used to represent the time relationship of the three types of key points. Finally, the time matrix is flattened from two dimensions to one dimension and passed through a fully connected layer (FC) to obtain the time feature Tfeature , the formula is defined as:

[0065] T relation =Flatten(V start ×V instensity )

[0066] T feature =FC(T relation )

[0067] Based on the above operations, the scene adaptation controller 2 further aggregates the running features F run and time characteristics T feature To get the final device characteristics F final-device 、Environmental characteristics F final-env and scene features F final-switch , the formula is defined as:

[0068] F final-device =F run ⊙T feature

[0069] F final-env =F adapt ⊙T feature

[0070] F final-switch =F switch ⊙T feature

[0071] Where ⊙ is the element-wise multiplication operation.

[0072] After completing feature sampling and aggregation, the scene adaptation controller 2 merges the final device features, environment features, and scene features, and obtains the mixed features F through a multi-layer perceptron (MLP). mixed , and finally the mixed feature F mixed After segmentation, the device linkage control vector C generated by the scene adaptation controller 2 is obtained device and scene adaptation control vector C scene The specific calculation steps are as follows:

[0073] F mixed =MLP(Concat(F final-device ,F final-device F final-env ,F final-switch ))

[0074] The calculation formula of the segmentation control vector is:

[0075] [C device ,C scene ]=Split(F mixed )

[0076] Concat is a merge operation, and Split is a split operation.

[0077] In addition, the scene adaptation controller 2 also controls the device linkage control vector C device and scene adaptation control vector C scene Use the cosine embedding function to obtain the time series information T of the device control vector and the scene control vector device and T scene , the formula is defined as:

[0078] T device =CosineEmbed(C device )

[0079] T scene =CosineEmbed(C scene )

[0080] CosineEmbed is a cosine embedding function that converts time information into a feature vector.

[0081] Next, we enter the implementation process of the device linkage decision maker 3 and the feedback optimizer 4. Both modules are composed of stacked multi-layer dynamic programming networks and need to use the device linkage control vector C generated by the scene adaptation controller 2. device and scene adaptation control vector C scene Processing is performed. Except for the first layer, the generated control vector is multiplied by the optimized feature vector output by the decision maker in the previous layer to obtain the control vector that preserves the previous valid information. The control vector, time series information, perception features, and corresponding time information are then input into the corresponding decision maker, which outputs the optimized feature vector. The specific calculation steps are as follows:

[0082]

[0083] in is the output of the device linkage decision maker of the current layer, is the output of the feedback optimizer of the current layer, and are the features output by the previous layer respectively.

[0084] After each layer outputs the feature vector, the feature vector of the device linkage will be used to update the operating status S of the device device and S scene , the formula is defined as:

[0085] S device =φ(Update(D device ))

[0086] S scene=φ(Update(D scene ))

[0087] Where Update is the forward network that updates the device's operating status, and φ is the activation function that normalizes the value to 0 to 1.

[0088] Finally, the feature vector output by the last layer of decision makers will pass through the head forward network for optimization to obtain the device operation status R device , environmental adaptation state R env , scene switching time R time The equipment linkage result and user satisfaction. The calculation formula of equipment operation status is:

[0089] R device =φ(Output(D device ))

[0090] R env =φ(Output(D scene ))

[0091] R time =φ(Output(T scene ))

[0092] R satisfaction =φ(Output(S device ,S scene ))

[0093] Where Output is the final predicted forward network and φ is the activation function.

[0094] In summary, the embodiment of the present invention realizes the linkage of online homestay equipment based on multi-scene perception through the collaborative work of the perception data aggregator 1, the scene adaptation controller 2, the device linkage decision maker 3 and the feedback optimizer 4. In actual application scenarios, for example, when a user enters a homestay room, the system can automatically adjust the air-conditioning temperature, turn on the background music and adjust the light brightness according to the real-time perception data, thereby providing a personalized living experience. This intelligent device linkage not only improves the user's comfort, but also significantly improves energy utilization efficiency, fully reflecting the technical advantages and application value of the present invention.

[0095] See also Figure 2 The second embodiment of the present invention provides a method for linking online homestay equipment based on multi-scenario perception, which includes the following steps:

[0096] S201, acquiring and integrating multi-source sensor data to generate fused perception features;

[0097] S202: Input the fused perception features and the initial operating state of the device into the scene adaptation controller to generate a device linkage control vector and a scene adaptation control vector as well as corresponding time series information;

[0098] S203, inputting the control vector, time series information, perception features, and time information into the device linkage decision maker and the feedback optimizer, and outputting an optimized feature vector;

[0099] S204: Update the device operating status according to the optimized feature vector and output the device linkage result.

[0100] The third embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. The computer program can be executed by a processor of the device where the computer-readable storage medium is located to implement the above-mentioned online homestay device linkage method based on multi-scene perception.

[0101] For example, the computer program described in the third embodiment of the present invention can be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the error correction device. For example, the apparatus described in the second embodiment of the present invention.

[0102] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the error correction method, and utilizes various interfaces and lines to connect the various parts of the processing method.

[0103] The memory can be used to store the computer program and / or module, and the processor implements various functions of an error correction method by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function (such as a sound playback function, a text conversion function, etc.); the data storage area can store data created based on the use of the mobile phone (such as audio data, text message data, etc.). In addition, the memory can include a high-speed random access memory and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0104] Wherein, if the implemented module is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of each of the above-mentioned method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0105] It should be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive effort.

[0106] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention.

Claims

1. A method for linking online homestay equipment based on multi-scenario perception, characterized in that: The following steps are involved: Acquire and integrate multi-source sensor data to generate fused perception features; The fused perception features and the initial operating state of the device are input into the scene adaptation controller to generate the device linkage control vector and the scene adaptation control vector as well as the corresponding time series information. The control vector, time series information, perception features, and time information are input into the device linkage decision maker and feedback optimizer, and the optimized feature vector is output; Update the device operating status according to the optimized feature vector and output the device linkage result.

2. The method according to claim 1, characterized in that Acquiring and integrating multi-source sensor data includes the following steps: De-noising and normalizing multi-source sensory data; Generate a dynamic weight matrix of the same dimension as the importance distribution information of the data; The dynamic weight matrix and the preprocessed data are input into the adaptive feature extraction network to obtain the fused perceptual features.

3. The method according to claim 1, characterized in that The scene adaptation controller is used to: Obtain the device operation area characteristics and environmental background characteristics based on the initial operation status of the device, and merge them into the regional characteristics of the device-environment pair; Use convolutional layers and multi-layer perceptrons to predict three key points: device start and stop points, operating intensity points, and scene switching points; The three types of key points are sampled to obtain running features, and the time relationship between the three types of key points is constructed to generate time features; Aggregate operation characteristics and time characteristics to obtain the final device characteristics, environment characteristics and scene characteristics; Combine the final device features, environment features, and scene features and generate device linkage control vectors and scene adaptation control vectors through a multi-layer perceptron; The cosine embedding function is used on the device linkage control vector and the scene adaptation control vector to obtain the time series information of the device control vector and the scene control vector respectively.

4. The method according to claim 3, characterized in that The device start and stop points, operation intensity points, and scene switching points are normalized to between zero and one through a Sigmoid activation function.

5. The method according to claim 1, wherein The device linkage decision maker and feedback optimizer are composed of stacked multi-layer dynamic programming networks, and each layer of the network uses the feature vector output by the previous layer to update the control vector of the current layer.

6. The method according to claim 1, characterized in that The device linkage results include device operating status, environmental adaptation status, scene switching time and user satisfaction.

7. A multi-scenario perception-based online homestay equipment linkage system, characterized in that: It includes a perception data fusion unit, a scene adaptation controller, a device linkage decision maker, and a feedback optimizer, among which: Perception data fusion unit, used to acquire and integrate multi-source sensor data to generate fused perception features; The scene adaptation controller is used to input the fused perception features and the initial operating state of the device into the scene adaptation controller to generate the device linkage control vector and the scene adaptation control vector as well as the corresponding time series information; The device linkage decision maker is used to input the control vector, time series information, perception features and time information into the device linkage decision maker and feedback optimizer, and output the optimized feature vector; The feedback optimizer is used to update the device operating status according to the optimized feature vector and output the device linkage results.

8. The system according to claim 7, characterized in that The perception data fusion device is specifically used to: perform denoising and normalization processing on multi-source perception data; Generate a dynamic weight matrix of the same dimension as the importance distribution information of the data; The dynamic weight matrix and the preprocessed data are input into the adaptive feature extraction network to obtain the fused perceptual features.

9. The system according to claim 7, wherein: The scene adaptation controller is specifically used to: Obtain the device operation area characteristics and environmental background characteristics based on the initial operation status of the device, and merge them into the regional characteristics of the device-environment pair; Use convolutional layers and multi-layer perceptrons to predict three key points: device start and stop points, operating intensity points, and scene switching points; The three types of key points are sampled to obtain running features, and the time relationship between the three types of points is constructed to generate time features; Aggregate operation characteristics and time characteristics to obtain device characteristics, environment characteristics and scene characteristics; Combine device features, environment features, and scene features and generate device linkage control vectors and scene adaptation control vectors through a multi-layer perceptron; The cosine embedding function is used on the device linkage control vector and the scene adaptation control vector to obtain the time series information of the device control vector and the scene control vector respectively.

10. The system according to claim 7, wherein: The device linkage decision maker and feedback optimizer are implemented through a stacked multi-layer dynamic programming network, and output the results of device operation status, environmental adaptation status, scene switching time and user satisfaction respectively.

Images