Multi-device cooperative control method, device and system for edge computing

Abstract

The invention discloses a multi-device cooperative control method for edge computing, which comprises the following steps that: an edge device module acquires environment information and sends the environment information to an edge computing module through a distributed soft bus; the module extracts information features, generates a decision instruction based on a preset machine learning algorithm, determines a target device group and a response device in the group, and issues an operation instruction through a distributed soft bus; and the response device executes and feeds back a result to the edge calculation module, and the edge calculation module controls the target feedback device to broadcast. According to the method, an edge calculation mode is adopted, data processing and instruction generation are completed on the edge side, the real-time performance of multi-device cooperative control is improved, and the method is suitable for scenes with high response speed requirements. The target device group and the response device are intelligently selected through a scoring mechanism, the device resource utilization rate and the cooperation efficiency are improved, data cloud transmission is reduced through a distributed soft bus, and network bandwidth occupation and communication cost are reduced.

Description

Technical Field

[0001] The present invention belongs to the field of edge computing technology, and specifically relates to a multi-device collaborative control method, device and system for edge computing. Background Art

[0002] In traditional centralized computing architectures, multi-device collaborative control typically relies on cloud servers for data processing and command issuance. This model has numerous limitations: First, data transmission to and from the cloud incurs significant latency, making it difficult to meet real-time requirements in scenarios such as emergency response and real-time control. Second, centrally uploading large amounts of device data to the cloud consumes significant network bandwidth, increasing communication costs and easily leading to control failures when the network is unstable.

[0003] Furthermore, existing multi-device collaborative control methods lack flexibility and intelligence in terms of device grouping and response device selection, making it difficult to dynamically adjust strategies based on real-time circumstances. For example, in a smart home scenario, when a user issues a command, it's impossible to efficiently select the optimal device from multiple devices to perform the task, potentially leading to delayed responses and wasted resources. Furthermore, existing technologies lack comprehensive processing mechanisms for emergency scenarios (such as a crying baby or a user calling for help) and complex scenarios (such as ingredient identification and cooking process coordination), making it difficult to meet diverse application needs. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-device collaborative control method, device and system for edge computing to solve the problems of high latency, large bandwidth occupancy, low device collaboration efficiency and insufficient emergency scenario processing capability in the existing centralized control mode proposed in the background technology.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A multi-device collaborative control method for edge computing includes the following steps: Step S1: Receive user instructions through the edge computing module and decompose the user instructions into a time-series task chain containing multiple subtasks; Step S2: querying the status parameters of all devices related to the time-series task chain in real time, where the status parameters include at least the busy / idle status of the device, real-time working parameters, and connection status; Step S3: Screening out devices in an idle state based on the device state parameters to form a candidate device set; Step S4: Calculate the scoring values ​​of the devices in the candidate device set according to the preset scoring mechanism, and select the device with the highest score as the optimal device to perform the corresponding subtask; In step S5, an operation instruction is sent to the optimal device through the distributed soft bus. After the optimal device executes the instruction, it feeds back the status to the edge computing module, triggering the device status judgment process of the next sequential task.

[0006] According to the above technical solution, the preset scoring mechanism takes into account at least the device's sound pickup intensity, processing capability, and pickup angle, and calculates the scoring value using a preset calculation formula. The scoring value calculation formula is: When the pickup angle r is in the first range (60°≤r<80° or 100°<r≤120°), ; When the pickup angle r is in the second range (80°≤r≤100°), ; Where s is the sound pickup intensity, a and b are preset weights with a>b, and c=10a+b / 10.

[0007] According to the above technical solution, when the edge computing module disassembles user instructions, it analyzes the logical relationship between instructions through machine learning algorithms to determine the sequence and dependency of each subtask.

[0008] According to the above technical solution, the distributed soft bus marks the instruction sequence through timestamps to ensure that multiple devices work together according to the established timing task chain and avoid logical conflicts in instruction execution. The edge computing module monitors and analyzes the status of device feedback in real time. If an abnormal device execution is detected, it will promptly adjust the subsequent instruction issuance strategy or reselect the device to execute the task.

[0009] According to the above technical solution, the process of issuing the operation instruction includes: If the answering device is an edge terminal without broadcasting function, the edge computing module selects the smart home appliance closest to the user from the target device group as the feedback device; The edge computing module controls the feedback device through a distributed soft bus to broadcast the execution results in voice.

[0010] According to the above technical solution, the edge computing module also parses the audio data in the environmental information, identifies emergency scenarios, and generates emergency response instructions; for emergency instructions, the edge computing module prioritizes interrupting the regular timing task chain, generating and executing the emergency instruction chain.

[0011] A multi-device collaborative control device based on edge computing, comprising: A data acquisition module, used to collect environmental information through edge device modules; The edge processing module is used to break down user instructions into time-series task chains, extract environmental information features, and generate decision instructions; The device scheduling module is used to screen idle devices based on their status, calculate the score, select the best responding device, and issue instructions; The feedback control module is used to control the target feedback device to broadcast the execution results.

[0012] According to the above technical solution, the equipment scheduling module includes: Status query unit, used to obtain the busy and idle status, working parameters and connection status of the device; The scoring calculation unit is used to calculate the device scoring value based on the sound pickup intensity, processing capability and sound pickup angle.

[0013] A multi-device collaborative control system based on edge computing, comprising: Edge device modules, including sensor units, camera units, and voice recognition modules; an edge computing module for controlling the method; Distributed soft bus, used to realize communication between edge device modules and edge computing modules.

[0014] According to the above technical solution, the edge computing module is also used to: trigger a waiting mechanism or send a "device busy" prompt when there is no idle device; receive user cancellation or modification instructions, terminate the current process and re-execute collaborative control.

[0015] Compared with the prior art, the present invention has the following beneficial effects: In this invention, an edge computing model is adopted, and data processing and instruction generation are completed on the edge side, avoiding the delay of the round trip to the cloud, significantly improving the real-time performance of multi-device collaborative control, and is suitable for scenarios with high response speed requirements. Through a scoring mechanism based on user location, device distribution, and device status, intelligent selection of target device groups and response devices is achieved, improving the utilization rate of device resources and collaborative efficiency. For edge terminals without broadcast functions, the nearest smart home appliance can be automatically selected as the feedback device to ensure timely broadcast of execution results and enhance user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of the multi-device collaborative control method of the present invention; Figure 2 This is the control block diagram of the multi-device collaborative control system of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 efforts are within the scope of protection of the present invention.

[0018] Example 1 like Figure 1 As shown, a multi-device collaborative control method for edge computing includes the following steps: Step S1: Receive user instructions through the edge computing module and decompose the user instructions into a time-series task chain containing multiple subtasks; In step S1, the edge computing module receives user commands (user commands include voice commands, text commands, etc., where voice commands are collected and initially converted by the voice recognition module of the edge device module. The conversion process is implemented using a lightweight natural language processing (NLP) model) and decomposes the user commands into a time-series task chain containing multiple subtasks. The specific conversion process includes: First, the speech recognition module converts user voice commands into text sequences. Then, the pre-trained BERT lightweight model is used to extract the core elements of the command, including the target device type (such as "air conditioner," "light"), action type (such as "open," "adjust"), parameter information (such as "26°C," "living room"), and time logic (such as "first... then...", "at the same time"). Based on the parsing results, natural language commands are mapped into standardized commands that can be executed by the device (for example, "adjust the air conditioner in the living room to 26°C" is mapped into "air conditioner - switch status = on; temperature parameter = 26°C; location identifier = living room"); Then, based on the time logic elements (such as "turn on the air conditioner first and then turn on the lights"), the execution order of the subtasks is determined to form a temporal task chain that includes execution timing and dependencies.

[0019] Step S2: querying the status parameters of all devices related to the time-series task chain in real time, where the status parameters include at least the busy / idle status of the device, real-time working parameters, and connection status; In step S2, the status parameters of all devices related to the sequential task chain are queried in real time. The status parameters include at least the device busy / idle status, real-time working parameters, and connection status. At the same time, for each subtask in the sequential task chain (corresponding to a single device instruction), its functional support requirements and performance requirements are determined respectively: Functional support requirements refer to the core functions that a device must possess (e.g., "temperature adjustment" requires a temperature control module, and "music playback" requires an audio output module). Matching is performed by querying the device's preset function tags (e.g., "air conditioning = temperature adjustment; speakers = audio playback"). Performance requirements refer to the performance indicators of the equipment in executing tasks (such as response delay ≤ 500ms, processing accuracy ≥ 95%, and continuous working stability ≥ 99%), which are dynamically set according to the task type (for example, the response delay for emergency tasks must be ≤ 200ms, and for routine tasks it can be relaxed to 1s). Step S3: Screening out devices in an idle state based on the device state parameters to form a candidate device set; Specifically, based on device status parameters, idle devices are screened out to form a candidate device set. At the same time, the candidate device set is screened again based on functional support requirements to eliminate devices that do not meet the functional requirements (for example, when performing the "temperature adjustment" task, lighting devices without temperature control modules are eliminated), thus obtaining a candidate subset that meets the functional requirements.

[0020] Step S4: Calculate the scoring values ​​of the devices in the candidate device set according to the preset scoring mechanism, and select the device with the highest score as the optimal device to perform the corresponding subtask; For the devices in the candidate device set, the score is calculated according to the preset scoring mechanism, and the device with the highest score is selected as the optimal device to perform the corresponding subtask; the scoring mechanism adds the weights of function matching degree and performance compliance rate: Functional matching (weight 0.3): If the device fully meets the functional requirements, it will be scored 1 point; if it partially meets the functional requirements (for example, only some parameter adjustments are supported), it will be scored 0.5 points; if it does not meet the functional requirements, it will be scored 0 points; Performance compliance rate (weight 0.4): Based on the device's historical execution data, the probability of meeting the current task's performance requirements is calculated (e.g., the number of response delays that meet the standard / total execution times), and is directly incorporated into the scoring formula. Original parameters (pickup sound intensity, angle, etc., weight 0.3): Combined with the above functions and performance parameters to calculate the comprehensive score.

[0021] In step S5, an operation instruction is sent to the optimal device through the distributed soft bus. After the optimal device executes the instruction, it feeds back the status to the edge computing module, triggering the device status judgment process of the next sequential task.

[0022] Specifically, an operation instruction is sent to the optimal device via a distributed soft bus. After the optimal device executes the instruction, it feeds back the result to the edge computing module, triggering the device status judgment process for the next sequential task. At the same time, the edge computing module monitors the execution process in real time, including: Based on real-time operating parameters fed back by the equipment (e.g., the rate of change of air conditioning temperature, the progress of light brightness adjustment), the actual completion of the task is calculated (calculation formula: completion = deviation rate between current state and target state × 100%. For example, if the target is 26°C and the current temperature is 25°C, the completion is 80%). If the actual completion rate is inconsistent with the preset progress (e.g., completion rate should be ≥90% within 10 seconds), and the duration exceeds the threshold (e.g., 5 seconds), it is considered an "execution abnormality"; If it is determined that the execution is abnormal, the edge computing module immediately re-screens the optimal device from the candidate device set (giving priority to the device with the second highest matching degree between function and performance requirements); If there are no available devices in the candidate device set, then add devices that are not included in the initial screening (need to meet basic functional requirements) to the target device group, recalculate the scores, and issue the command; During the switching process, the switching information (such as "air conditioning response delay, has been switched to the backup device for execution") is fed back to the user through the distributed soft bus.

[0023] In the present invention, an edge computing mode is adopted, and data processing and instruction generation are completed on the edge side, avoiding the delay of round-trip to the cloud, significantly improving the real-time performance of multi-device collaborative control, and is suitable for scenarios with high requirements on response speed.

[0024] Through a scoring mechanism based on user location, device distribution and device status, intelligent selection of target device groups and response devices is achieved, improving the utilization rate and collaborative efficiency of device resources.

[0025] For edge terminals without broadcasting functions, the system can automatically select the nearest smart home appliance as the feedback device to ensure timely broadcasting of execution results and improve user experience.

[0026] It has the ability to identify and respond to emergency scenarios such as babies crying or users calling for help. Emergency response instructions are executed first and trigger a global alarm, enhancing the safety and reliability of the system.

[0027] The distributed soft bus enables direct communication between edge devices, reduces the centralized transmission of data to the cloud, and reduces network bandwidth usage and communication costs.

[0028] Example 2 This embodiment further refines the first embodiment. It provides a specific preset scoring algorithm. In addition to considering the device's sound pickup intensity, processing capabilities, and pickup angle, the preset scoring mechanism also considers the device's implementation efficiency (e), implementation effect (g), and processing convenience (i). This multi-dimensional scoring system is formed by integrating weight coefficients (d, f, h) with the original parameters (sound pickup intensity s, pickup angle r).

[0029] The scoring algorithm consists of two stages. The first stage is: Closest device screening based on sound pickup intensity, processing capabilities, and pickup angle: Filters the device closest to the user in terms of physical location or signal reception capabilities from the candidate device set to form a preliminary candidate set.

[0030] Sound pickup intensity-distance mapping: Assume that the sound pickup intensity is s (unit: dB). The inverse relationship between sound intensity and distance can be expressed as: Where k is a constant, s>0; the greater the sound intensity, The smaller the mapping distance, the closer the device is to the user.

[0031] Calculation of pickup angle deviation: The optimal pickup angle is defined as, and the deviation between the actual angle r and the optimal angle is , the deviation distance mapping is: The smaller the deviation, that is, the closer r is, the mapping distance The smaller.

[0032] Processing capacity filtering threshold: Set the processing capacity to p (0-1 normalized value), and preset the threshold (such as 0.6), only retain equipment.

[0033] Comprehensive distance index: The distance index of the comprehensive sound intensity and angle is: in , when sound intensity is the priority .

[0034] Filtering rule: Sort by D from small to large, select the first n devices, and form the filtered candidate set.

[0035] Phase II: Multi-dimensional scoring based on implementation efficiency, implementation effect, and processing convenience.

[0036] In the preliminary candidate set, the comprehensive score is calculated based on the equipment's task execution efficiency, effectiveness and ease of operation, and the optimal equipment is selected.

[0037] Implementation efficiency e: Based on the device's historical task processing time t, it is normalized to: in The shortest processing time for similar tasks.

[0038] Implementation effect g: Based on the task completion quality score (such as user feedback, success rate) normalized to Interval value.

[0039] Processing convenience i: Based on the device operation complexity (such as the number of command response steps and interface compatibility), it is reverse-normalized to: The scoring formula combining the original parameters and the new dimensions is: When the pickup angle r is in the first range (60°≤r<80° or 100°<r≤120°): When r is in the second range (80°≤r≤100°): Where: a, b are the original weights (a>b), such as (a=0.4, b=0.2); (d, f, h) are the new dimension weights (e.g., d = 0.2, f = 0.15, h = 0.05, and a + b + d + f + h = 1); (c = 10a + b / 10).

[0040] Optimal device selection: Calculate the P values ​​of all devices in the filtered candidate set and select the device with the largest P as the execution device.

[0041] This embodiment also provides a specific implementation method of a machine learning algorithm, which uses the machine learning algorithm to analyze the logical relationship between instructions and determine the sequence and dependency relationship of each subtask.

[0042] The edge computing module obtains the instruction sequence through the speech recognition module and extracts the feature vector of each instruction through a lightweight NLP model, as follows: Instruction feature representation: Each instruction Expressed as ,in, Indicates that a collection of devices is involved; Indicates the action type; Represents parameters; Represents time characteristics; Represents contextual features.

[0043] For dependent instructions: define dependencies Quantify the dependency strength of two instructions (range: ), which is weighted by the following submodules: If two instructions involve the same device, there is a competition dependency on device resources. The calculation formula is: If the action / parameter of instruction j depends on the execution result of instruction i (e.g. "adjust" depends on "open", "delay operation" depends on the completion of the previous instruction), then: The BERT model is used to extract time-related words in instructions (such as "first, then, etc...") and output the time dependency probability: .

[0044] Use machine learning models (such as random forests) to optimize weights (By training with labeled data, the predicted dependency is consistent with the actual dependency), and finally: Setting dependency thresholds (Determined through training data, it was determined that there was a dependency relationship at the time): For example, There is a dependency relationship, that is Must be in Executed before.

[0045] For directives without dependencies, define the degree of parallelism: , set the parallel threshold (Parallelism can be executed in parallel); For example, , which can be executed in parallel.

[0046] Based on the above judgment, build a task chain: Parallel phase: and Can be started simultaneously (no dependencies and parallelism is achieved), so they are executed simultaneously at time 0 and ; Serial phase: After the execution is completed (for example, 2 seconds), it enters the waiting period and executes after the waiting period ends. .

[0047] Example 3 This embodiment is a further refinement of the second embodiment. This embodiment takes a smart home scenario as an example to explain in detail the multi-device collaborative control method of the present invention.

[0048] In a smart home environment, a user issues a voice command "Turn on the air conditioner in the living room". After the voice recognition module of the edge device module collects the command and the sound recognition module converts the command into text, the BERT model parses the core elements: target device = "living room air conditioner", action 1 = "turn on", action 2 = "adjust the temperature to 26°C", time logic = "execute action 2 after action 1 is completed"; mapped to device instructions: "air conditioner - switch = on (sequence 1); air conditioner - temperature = 26°C (sequence 2, depends on sequence 1 to be completed)".

[0049] Functional requirement = "with switch control + temperature adjustment functions"; performance requirement = "switch response time ≤ 1s, temperature adjustment accuracy ±1°C".

[0050] The initial candidate devices include living room air conditioners A and B (spare). First, eliminate the living room fan without temperature control (function mismatch). Then calculate the scores: Air conditioner A has a function match score of 1, a performance compliance rate of 98% (98 out of 100 historical executions met the response time), and an overall score of 85. Air conditioner B has a function match score of 1, a performance compliance rate of 90%, and an overall score of 78. Therefore, air conditioner A is selected.

[0051] If air conditioner A does not respond "powered on" within 1 second (completion rate 0%, inconsistent with the preset progress), the edge computing module immediately triggers the switch and selects air conditioner B for execution. At the same time, the smart speaker announces "Air conditioner A did not respond, and the execution has been switched to air conditioner B."

[0052] The command is transmitted to the edge computing module via a distributed soft bus. This module needs to select the best response device from the smart speakers, air conditioners, lights, and other devices in the living room, execute the command, and feedback the results. The specific steps include: Step 1: Initial screening based on physical location and signal capabilities.

[0053] Sound pickup intensity distance mapping: Smart speaker sound pickup intensity , according to the formula , mapping distance ; Air conditioner pickup sound intensity .

[0054] Pickup Angle Deviation Calculation: Smart Speaker Pickup Angle (First range 60°≤r<80°), deviation , mapping distance ; Air conditioner sound pickup angle (Second range 80°≤r≤100°), deviation ; Mapping distance .

[0055] Processing capacity filtering: If the processing capacity p of smart speakers and air conditioners is greater than the preset threshold of 0.6, both are retained.

[0056] Comprehensive distance index: taking sound intensity weight , angle weight , Comprehensive distance of smart speakers (D1=0.7×0.0125+0.3×0.111≈0.042\); Air conditioning comprehensive distance (D2=0.7×0.0167+0.3×0.0278≈0.020\).

[0057] Sorting by D from small to large, air conditioners (D2=0.020) and smart speakers (D1=0.042) enter the preliminary candidate set.

[0058] Step 2: Comprehensive scoring across multiple dimensions.

[0059] Achieved efficiency e: the shortest time the smart speaker has historically processed similar commands , current processing time ,but ; The shortest time in history of air conditioning ,current ,but .

[0060] Implementation effect g: The success rate of historical commands of smart speakers is 95%, which is standardized as ; Air conditioning success rate 90%, standardized .

[0061] Processing convenience i: Smart speaker operation complexity (number of steps), ,but ; Air conditioning operation complexity ,but .

[0062] Scoring formula calculation: set up Smart speakers ( , first range): ; air conditioner( , second range): .

[0063] Step 3: Optimal device selection. The smart speaker's score P1≈-57.66 is higher than the air conditioner's P2≈-61.67, so the smart speaker is selected as the response device. The distributed soft bus sends instructions, and the smart speaker controls the air conditioner to start and broadcast the results.

[0064] Example 4 This embodiment is a further refinement of the third embodiment, focusing on the process of handling emergency scenarios.

[0065] The kitchen gas concentration sensor detects that the concentration exceeds the threshold (such as 500ppm), and the camera captures the gas leak. The edge computing module needs to urgently dispatch gas alarms, smart valves, living room speakers and other equipment, and give priority to executing alarms and closing valves.

[0066] Step 1: Dynamic Weight Adjustment: In emergency scenarios, increase the weight of implementation efficiency (e) to d = 0.5 and the processing capacity (p) threshold to 0.8 to prioritize high-responsiveness devices. Ignore the impact of pickup angle deviation on the score (i.e., b = 0), and focus on the device's real-time status and execution speed.

[0067] Step 2: Calculate the device score for the gas alarm: s = 90 dB (strong sound intensity), p = 0.9 (> 0.8), e = 1 (minimum processing time 0.5 s), g = 1 (alarm success rate 100%), i = 1 (one-button trigger), because r = 95° (second range), c = 10 × 0.4 + 0 = 4.

[0068] .

[0069] Step 3: The emergency command chain executes the edge computing module, issuing commands based on the scoring priority. First, the gas alarm (P = 40.7) is triggered, which then controls the smart valve (P = 32.64) to close the pipeline. Simultaneously, the living room speaker broadcasts an emergency message. This dynamically interrupts the regular task chain (for example, the ongoing soup-making process) to ensure real-time response to the emergency command.

[0070] Example 5 This embodiment is a further refinement of the fourth embodiment, and details the functions of food identification and cooking process coordination in the smart kitchen scenario.

[0071] like Figure 2 As shown in the figure, in a smart kitchen scenario, the user places ingredients on the kitchen counter. The camera unit collects image data of the ingredients and sends it to the edge computing module via a distributed soft bus. The edge computing module processes the image data based on a preset image recognition algorithm (such as a convolutional neural network algorithm) to identify the type of ingredient (such as tomatoes, eggs, flour, etc.) and the expiration date (this can be determined by identifying the production date and expiration date on the packaging or judging the freshness of the ingredients based on their appearance).

[0072] Based on the recognition results, the edge computing module generates a shopping list or recipe recommendations. For example, if it identifies tomatoes and eggs as ingredients, and the eggs are nearing their expiration date, it generates a shopping list containing "eggs" and recommends a recipe for "scrambled eggs with tomatoes." After the user confirms the recipe, the edge computing module coordinates the kitchen equipment to execute the cooking process.

[0073] The specific coordination process is as follows: Image acquisition: The camera unit above the kitchen worktop (resolution 1920×1080, frame rate 30fps) captures food images in real time and transmits them to the edge computing module via a distributed soft bus using the UDP protocol, with a transmission delay of ≤50ms.

[0074] Edge computing processing: The YOLOv8 lightweight model was used to identify food ingredients. The input image was resized to 640×640, and the recognition time was ≤150ms. Tomatoes (98% confidence) and eggs (95% confidence) were successfully identified. The production date and expiration date (30 days) on the egg packaging were parsed, and the remaining shelf life was determined to be 1 day based on the current date, triggering a "priority use" reminder.

[0075] Task chain generation: The edge computing module uses the LSTM algorithm to analyze the user's historical cooking habits (for example, making scrambled eggs with tomatoes three times in the past week) and generate a time-series task chain: weighing ingredients → preheating the ceramic hob → beating eggs → dicing tomatoes → mixing and stir-frying → serving reminder.

[0076] Device collaborative execution phase: The edge computing module sends a weighing instruction via the distributed soft bus: "Weigh 2 tomatoes, 3 eggs." The instruction includes a timestamp. The smart scale performs weighing: tomatoes weigh 350g, and eggs weigh a total of 180g. The data is fed back to the edge computing module via the soft bus, with a feedback delay of ≤80ms. The edge computing module compares the recipe standards (300-400g of tomatoes, 150-200g of eggs) and generates a “proper weight” prompt.

[0077] Candidate device screening and scoring: Step 1, device status query: Electric ceramic stove (model ET-02): Idle, temperature sensor normal, maximum power 2000W; Gas stove (model GS-03): Idle, gas valve open, flame detector normal.

[0078] Device screening (physical location and signal capabilities are shown in Table 1).

[0079] Table 1 Physical location of equipment and signal capabilities Step 2: Multi-dimensional scoring (weights a=0.4, b=0.2, d=0.2, f=0.15, h=0.05).

[0080] Step 3: Device collaborative control: The edge computing module sends a command to the ceramic cooker via the distributed soft bus: "Preheat to 200°C, countdown 3 minutes." The command carries a timing dependency flag (depending on the "weighing completed" status). After receiving the command, the ceramic cooker sends status feedback to the edge computing module: "Started, current temperature 25°C, expected to reach 200°C in 1.5 minutes." The edge computing module synchronously schedules the smart egg beater (candidate device score P=-45.3) to enter standby mode, waiting for the "preheating completion" signal to trigger the next loop.

[0081] Each instruction is accompanied by a timestamp accurate to milliseconds to ensure the timing correctness of "weighing completed (12:01:30) → preheating started (12:02:15)"; The edge computing module maintains a task status lock: When the ceramic stove returns "preheating", the "beat the eggs" task is locked and not executed to avoid device resource conflicts.

[0082] If the ceramic stove preheats for more than 2 minutes but does not reach 200°C, the edge computing module detects an abnormal temperature rise rate (<5°C / 10s) and immediately executes: Send the command "ceramic stove stops heating" and mark the device as faulty; After rescreening the candidate devices, the gas stove was activated due to its second highest score (P=-63.7055), and a preheating instruction was issued; The smart speaker announces: "There is an abnormality with the ceramic stove, and the system has switched to preheating the gas stove."

[0083] The entire process delay from ingredient identification to equipment scheduling is ≤300ms, meeting the instant response requirements of cooking scenarios; Through a multi-dimensional scoring algorithm, the selection efficiency of electric ceramic stoves in preheating tasks is improved by 42% compared with random scheduling; The abnormal equipment switching mechanism makes the cooking process interruption recovery time ≤ 10s, ensuring task continuity; The smart speaker's real-time broadcast combined with equipment status feedback allows users to monitor the cooking progress throughout the entire process.

[0084] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0085] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A multi-device collaborative control method for edge computing, characterized by: The following steps are involved: Step S1: Receive user instructions through the edge computing module and decompose the user instructions into a time-series task chain containing multiple subtasks; Step S2: querying the status parameters of all devices related to the time-series task chain in real time, where the status parameters include at least the busy / idle status of the device, real-time working parameters, and connection status; Step S3: Screening out devices in an idle state based on the device state parameters to form a candidate device set; Step S4: Calculate the scoring values ​​of the devices in the candidate device set according to the preset scoring mechanism, and select the device with the highest score as the optimal device to perform the corresponding subtask; In step S5, an operation instruction is sent to the optimal device through the distributed soft bus. After the optimal device executes the instruction, it feeds back the status to the edge computing module, triggering the device status judgment process of the next sequential task.

2. The multi-device collaborative control method for edge computing according to claim 1, characterized in that: The preset scoring mechanism considers at least the device's sound pickup intensity, processing capabilities, and pickup angle, and calculates the score using a preset calculation formula. The scoring formula is: When the pickup angle r is in the first range (60°≤r<80° or 100°<r≤120°), ; When the pickup angle r is in the second range (80°≤r≤100°), ; Where s is the sound pickup intensity, a and b are preset weights with a>b, and c=10a+b / 10.

3. The multi-device collaborative control method for edge computing according to claim 2, characterized in that: When the edge computing module disassembles user instructions, it uses machine learning algorithms to analyze the logical relationship between instructions and determine the sequence and dependency of each subtask.

4. The multi-device collaborative control method for edge computing according to claim 3, characterized in that: The distributed soft bus uses timestamps to mark the instruction sequence, ensuring that multiple devices work together according to the established timing task chain to avoid logical conflicts in instruction execution; The edge computing module monitors and analyzes the status of device feedback in real time. If an abnormal device execution is detected, it will promptly adjust the subsequent instruction issuance strategy or reselect the device to execute the task.

5. The multi-device collaborative control method for edge computing according to claim 4, characterized in that: The process of issuing an operation instruction includes: If the answering device is an edge terminal without broadcasting function, the edge computing module selects the smart home appliance closest to the user from the target device group as the feedback device; The edge computing module controls the feedback device through a distributed soft bus to broadcast the execution results in voice.

6. The multi-device collaborative control method for edge computing according to claim 1, characterized in that: It also includes an edge computing module that parses audio data in environmental information, identifies emergency scenarios, and generates emergency response instructions; for emergency instructions, the edge computing module prioritizes interrupting the regular timing task chain, generating and executing the emergency instruction chain.

7. A multi-device collaborative control device based on edge computing, characterized by: include: A data acquisition module, used to collect environmental information through edge device modules; The edge processing module is used to break down user instructions into time-series task chains, extract environmental information features, and generate decision instructions; The device scheduling module is used to screen idle devices based on their status, calculate the score, select the best responding device, and issue instructions; The feedback control module is used to control the target feedback device to broadcast the execution results.

8. The multi-device collaborative control device based on edge computing according to claim 7, characterized in that: The equipment scheduling module includes: Status query unit, used to obtain the busy and idle status, working parameters and connection status of the device; The scoring calculation unit is used to calculate the device scoring value based on the sound pickup intensity, processing capability and sound pickup angle.

9. A multi-device collaborative control system based on edge computing, characterized by: include: Edge device modules, including sensor units, camera units, and voice recognition modules; An edge computing module, configured to execute the control method according to any one of claims 1 to 6; Distributed soft bus, used to realize communication between edge device modules and edge computing modules.

10. The multi-device collaborative control system based on edge computing according to claim 9, characterized in that: The edge computing module is also used to: trigger a waiting mechanism or send a "device busy" prompt when there are no idle devices; receive user cancellation or modification instructions, terminate the current process and re-execute collaborative control.

Images