A sensor data collaborative transmission system, method, and medium based on the Internet of Things

Abstract

This invention provides a sensor data collaborative transmission system, method, and medium based on the Internet of Things (IoT). The system includes a management platform and a sensing control platform. The management platform is configured to: determine the data type of a data frame based on its header parameters; determine the priority of the data frame based on its data type, failure time, and information entropy; determine bandwidth adjustment parameters based on channel load data, priority queue backlog data, and data frame priority; and generate a first adjustment command based on the bandwidth adjustment parameters and send it to the sensing control platform. The sensing control platform includes an edge computing device configured to: adjust the channel's transmission bandwidth based on the first adjustment command. This method achieves accurate identification and priority protection of critical commands; solves the problems of high latency in control commands and packet loss of sensing data caused by traditional static allocation; and improves the real-time performance and stability of industrial control systems.

Description

Technical Field

[0001] This specification relates to the field of data transmission, and in particular to a sensor data collaborative transmission system, method and medium based on the Internet of Things. Background Technology

[0002] In existing information systems, high-frequency sensing data acquired by sensors and real-time control commands typically need to be transmitted through shared physical channels, and the bandwidth of each channel is usually fixed. Because the massive amount of sensing data can consume limited transmission bandwidth, critical control commands often experience uncontrollable transmission delays due to network congestion.

[0003] Therefore, it is desirable to provide a sensor data collaborative transmission system and method based on the Internet of Things, which can dynamically isolate and schedule data streams without increasing physical channels. Summary of the Invention

[0004] The system includes an IoT-based sensor data collaborative transmission system, comprising a management platform and a sensing control platform. The management platform is configured to: determine the data type of a data frame based on its header parameters; determine the priority of the data frame based on its data type, failure time, and information entropy; determine bandwidth adjustment parameters based on channel load data, priority queue backlog data, and the priority of the data frame; and generate a first adjustment command based on the bandwidth adjustment parameters and send it to the sensing control platform. The sensing control platform includes an edge computing device configured to: adjust the transmission bandwidth of the channel based on the first adjustment command.

[0005] The present invention includes a sensor data collaborative transmission method based on the Internet of Things (IoT), implemented based on a sensor data collaborative transmission system. The method includes: determining the data type of the data frame based on the header parameters of the data frame; determining the priority of the data frame based on the data type, failure time, and information entropy; determining bandwidth adjustment parameters based on channel load data, priority queue backlog data, and the priority of the data frame; generating a first adjustment command based on the bandwidth adjustment parameters and sending it to the sensing control platform; and adjusting the transmission bandwidth of the channel based on the first adjustment command.

[0006] The invention includes a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions from the storage medium, the computer executes the sensor data collaborative transmission method.

[0007] By analyzing the header parameters of data frames to directly distinguish between control commands and sensing data, and combining failure time and information entropy to construct a multi-dimensional priority evaluation system for data frames, accurate identification and priority protection of key commands are achieved. At the same time, the transmission bandwidth is dynamically adjusted based on channel load and queue backlog, which effectively solves the problems of high latency of control commands and packet loss of sensing data caused by traditional static allocation, and improves the real-time performance and stability of industrial control systems. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the platform structure of an IoT-based sensor data collaborative transmission system according to some embodiments of this specification;

[0009] Figure 2 This is an exemplary flowchart of an IoT-based sensor data collaborative transmission method according to some embodiments of this specification;

[0010] Figure 3 This is a schematic diagram illustrating preprocessing operations according to some embodiments of this specification;

[0011] Figure 4 This is an exemplary flowchart illustrating the determination of bandwidth adjustment parameters according to some embodiments of this specification;

[0012] Figure 5 These are exemplary schematic diagrams of effect prediction models shown in some embodiments of this specification;

[0013] Figure 6 This is an exemplary schematic diagram of a bandwidth prediction model according to some embodiments of this specification. Detailed Implementation

[0014] The accompanying drawings used in the description of the embodiments will be briefly introduced below. The drawings do not represent all embodiments.

[0015] The terms “system,” “device,” “unit,” and / or “module” as used herein are a method of distinguishing different components, elements, parts, sections, or assemblies at different levels. If other terms can achieve the same purpose, they may be replaced with other expressions.

[0016] Unless the context clearly indicates an exception, words such as "a," "an," "a kind," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0017] In the embodiments described in this specification, the order of the steps is interchangeable unless otherwise specified, and steps may be omitted. Other steps may also be included in the operation process.

[0018] Figure 1 This is a schematic diagram of the platform structure of an IoT-based sensor data collaborative transmission system, as shown in some embodiments of this specification.

[0019] In some embodiments, such as Figure 1 As shown, the IoT-based sensor data collaborative transmission system 100 may include a management platform 110, a sensor network platform 120, and a perception control platform 130.

[0020] Management Platform 110 refers to a digital monitoring and management platform that oversees transmission bandwidth.

[0021] In some embodiments, the management platform 110 may be configured in a processor and / or server. The processor and / or server may process data and / or information obtained from other platforms. Based on this data, information, and / or processing results, the processor and / or server may execute program instructions to perform one or more functions described in this specification.

[0022] In some embodiments, the management platform 110 includes inter-platform sub-platforms and a data center that communicate with each other. The sub-platforms can process data and / or information obtained from the data center.

[0023] In some embodiments, the sub-platforms of the management platform 110 include at least one of a data perception sub-platform, a data analysis sub-platform, and a control sub-platform.

[0024] The data-aware sub-platform refers to a platform used to receive bandwidth management data. Bandwidth management data refers to data related to bandwidth management, such as header parameters and data types of data frames.

[0025] The data analytics sub-platform refers to a management platform that assesses and predicts the demand for adjusting transmission bandwidth. In some embodiments, the data analytics sub-platform can be configured to determine predicted bandwidth requirements. See [link to documentation] for an explanation of predicted bandwidth requirements. Figure 5 And its contents.

[0026] A control sub-platform refers to a platform used to remind users to adjust transmission bandwidth after determining the need for such adjustment. In some embodiments, the control sub-platform includes a mobile / computer control terminal access interface, supporting visualization of axial force / torque data, alarms, etc. For example, when the sensor data load exceeds a preset load threshold, the control sub-platform can remind users to pay attention and adjust the bandwidth in a timely manner through pop-up reminders on the terminal device, SMS notifications, etc.

[0027] In some embodiments, a data center includes a database, a data processing model library, and computing units.

[0028] Databases are used to collect, store, manage, and control related data. Examples include MySQL, PostgreSQL, InfluxDB, and Prometheus.

[0029] A data processing model library refers to a collection of data processing models used to manage data processing. In some embodiments, data processing models may include effect prediction models, bandwidth prediction models, etc. For an explanation of effect prediction models, see [link to documentation]. Figure 5 For details regarding the bandwidth prediction model, please refer to [link / reference]. Figure 5 And its contents.

[0030] A computing unit is a functional module used to perform arithmetic, logical, and other instruction operations. Computing units may include, but are not limited to, central processing units (CPUs).

[0031] The sensor network platform 120 refers to a platform for sensing and communicating bandwidth management data, used for bidirectional data interaction and communication between the management platform 110 and the sensing and control platform 130.

[0032] For example, the sensor network platform 120 may include communication equipment, servers, and various gateway devices. Bandwidth management data may include the data frame data type, priority, priority queue, and bandwidth adjustment parameters. For a description of the data frame data type, priority, and priority queue, please refer to [link to documentation]. Figure 2 And its contents.

[0033] The sensing and control platform 130 refers to an information processing platform for security supervision of various managed and regulated objects. A regulated object refers to the object to which management is implemented. In some embodiments, the regulated object may be a controlled device. A controlled device refers to a device whose transmission bandwidth needs to be adjusted, such as a robotic arm or environmental auxiliary equipment. Environmental auxiliary equipment may include a temperature and humidity control system. The sensing and control platform 130 may include an independent transmission channel, edge computing device, and sensors corresponding to each regulated object. The edge computing device may include a gateway, a computer, or an intelligent network device. Sensors may include humidity sensors, temperature sensors, vibrating wire force gauges, torque sensors, and multi-dimensional sensors; the measurement range of the sensors can be determined according to factory settings.

[0034] In some embodiments, the sensing and control platform deploys sensors on the support shaft or drive shaft to convert force parameters into frequency signals. Each sensor integrates a radio frequency chip, forming a tree-like network, where leaf nodes are only responsible for data acquisition, and routing nodes have data relay functions.

[0035] In some embodiments of this specification, the IoT-based sensor data collaborative transmission system 100 can form an information operation closed loop between various functional platforms and operate in a coordinated and regular manner under the unified management of the management platform.

[0036] It should be noted that the above description of the IoT-based sensor data collaborative transmission system 100 and its platform is for ease of description only and should not be construed as limiting this specification to the scope of the embodiments described. It is understood that those skilled in the art, after understanding the principles of this system, may arbitrarily combine the various platforms or construct subsystems to connect with other platforms without departing from these principles.

[0037] Figure 2 This is an exemplary flowchart illustrating an IoT-based sensor data collaborative transmission method according to some embodiments of this specification. Figure 2 As shown, process 200 includes the following steps. In some embodiments, process 200 may be executed by a management platform in an IoT-based sensor data collaborative transmission system.

[0038] Step 210: Determine the data type of the data frame based on the header parameters of the data frame.

[0039] A data frame is the basic unit of data transmission. In some embodiments, a data frame includes header parameters.

[0040] Header parameters refer to the protocol information in the header of a data frame. Examples include protocol identifiers (such as protocol controller (EtherCAT), bus controller (CAN), transmission control protocol (TCP / IP)), command words (such as command prompt, Cmd), and logical addresses.

[0041] Data type refers to the business attributes of a data frame. In some embodiments, data type includes control instructions and sensing data.

[0042] Control commands are instructions used to control the actions of controlled equipment. Examples include motion control commands and environmental control commands. Motion control commands can be used to control a robotic arm to perform physical actions such as material handling, while environmental control commands can be used to control environmental auxiliary equipment to adjust temperature and humidity.

[0043] Perception data refers to data acquired by sensors related to spatial environment perception. For example, sensors mounted on a robotic arm can acquire multidimensional sensor perception data, which may include the position data of various objects in a three-dimensional map of the environment in which the robotic arm is located, used to determine the obstacle situation around the robotic arm; sensors mounted on environmental assistance devices can acquire environmental monitoring data, including data such as temperature and humidity, used to determine the real-time environmental conditions.

[0044] In some embodiments, the management platform can determine the data type of a data frame by querying a first preset table based on the header parameters of the data frame.

[0045] The first preset table is a mapping table between the header parameters of the data frame and the data type of the data frame.

[0046] In some embodiments, the first preset table may include a mapping relationship between different header parameters and data types. The first preset table can be constructed through empirical presets. For example, if the Cmd of a data frame is a write operation (e.g., LWR, APWR) and the Address belongs to the motor register region (e.g., [0x0000-0x0FFF]), the corresponding data type is a control instruction according to the mapping relationship in the first preset table. As another example, if the Cmd of a data frame is a read operation (e.g., LRD, APRD) or the Address belongs to the sensor data region (e.g., [0x1000-0xFFFF]), the corresponding data type is sensing data according to the mapping relationship in the first preset table.

[0047] Step 220: Determine the priority of the data frame based on the data type, expiration time, and information entropy.

[0048] Failure time refers to the time difference between the current moment and the failure time.

[0049] The failure moment refers to the moment when a data frame loses its business value or causes a system failure.

[0050] A data frame losing its business value means that it can no longer support system decisions or business objectives. For example, the moment a control command completes execution, the control command loses its business value, and the management platform can determine this moment as the data frame's failure point. Another example is when the difference between a certain moment and its creation moment exceeds the data frame's maximum lifetime; the management platform can then determine this moment as the data frame's failure point. The maximum lifetime refers to the maximum duration a data frame can exist, and it can be determined based on empirical presets. For example, the maximum lifetime of a control command is 10ms.

[0051] The generation time refers to the moment when the data frame is generated. In some embodiments, the management platform can obtain the generation timestamp from the header parameters of the data frame and determine the generation timestamp as the generation time of the data frame.

[0052] System failure due to data frames refers to situations where errors in data frames cause system malfunctions, crashes, or performance degradation. For example, incorrect data frame format can lead to protocol parsing errors. Another example is sensor malfunction caused by data field verification failures due to incorrect data frame content.

[0053] Information entropy is a quantification of the degree of deviation between the current data frame and historical data.

[0054] In some embodiments, the information entropy can be determined by system presets. For example, the information entropy of a control command is preset to 1. The information entropy of the sensed data can be obtained by calculating the average of multiple sensed data. For example, the information entropy of the sensed data satisfies the following formula (1):

[0055] a3 = (|b1-b0|) / c(1)

[0056] Where a3 is the information entropy; b1 is the current value, i.e., the average value of the sensed data collected in the current data frame within the preset time period; b0 is the previous value, i.e., the average value of the sensed data collected in the previous data frame within the preset time period; and c is the physical range, i.e., the difference between the maximum and minimum values ​​of the sensor's measurement range. For more information on sensors and measurement ranges, please refer to [link to relevant documentation]. Figure 1 Corresponding description.

[0057] Priority refers to data used to measure the importance of a data frame. The higher the priority, the earlier the data frame is processed.

[0058] In some embodiments, the management platform can normalize the initial priority, urgency, and information entropy, perform a weighted summation of the normalization results, and determine the summation result as the priority.

[0059] Initial priority refers to the initial value of the priority of a data frame.

[0060] In some embodiments, the initial priority of a data frame can be determined based on an empirical preset. For example, the initial priority of a control command is 8, and the initial priority of a sensed data is 2.

[0061] Urgency refers to the degree of urgency in processing data frames. In some embodiments, urgency can be determined based on formula (2):

[0062] a2=(t max -t now ) / t max (2)

[0063] Where a2 refers to the degree of urgency, t max t represents the maximum expiration time of all data frames currently in use.now This is the expiration time of the currently calculated data frame. The management platform can obtain the expiration times of all data frames and take the maximum value as t. max .

[0064] Step 230: Determine the bandwidth adjustment parameters based on the channel load data, the priority queue backlog data, and the data frame priority.

[0065] A channel refers to a logical transmission channel or physical link between control sensors or edge computing devices, or between edge computing devices and a management platform.

[0066] Load data reflects the physical transmission capacity of a channel, such as the channel's current signal-to-noise ratio (SNR), physical bandwidth utilization, and packet loss rate.

[0067] In some embodiments, the management platform can acquire load data controlling the sensor output at the interface between the channel and the sensor. In some embodiments, the management platform can acquire the communication protocol between the platform and the device, and determine the load data by parsing the communication protocol.

[0068] A priority queue is a scheduling queue corresponding to a channel. In some embodiments, a priority queue includes a high-priority queue and a low-priority queue.

[0069] Backlogged data refers to the amount of data waiting to be processed in a priority queue. For example, the storage space occupied by data frames waiting to be sent. In some embodiments, the management platform can determine backlogged data by retrieving data stored in the data center to obtain the queue length of data frames waiting to be sent.

[0070] A high-priority queue refers to a priority queue that is processed first. In some embodiments, in response to the presence of data in a high-priority queue, the management platform prioritizes processing the data in the high-priority queue.

[0071] A low-priority queue refers to a priority queue that requires less processing. In some embodiments, the management platform processes data in the low-priority queue when there is no data in the high-priority queue or when there is available bandwidth on the channel.

[0072] Bandwidth adjustment parameters refer to parameters that adjust the transmission bandwidth of a channel. For example, the token generation rate of a channel.

[0073] A token refers to the transmission permission of a data frame. The token generation rate refers to the number of tokens added to the token bucket per unit of time. The token generation rate is related to the maximum number of data frames allowed to pass through the channel per unit of time; the higher the token generation rate, the greater the maximum number of data frames allowed to pass through the channel. The token bucket is a dataset that holds tokens, and it has a preset capacity, such as 100. Adding a token to the token bucket indicates that the data frame is allowed to be transmitted; the token is consumed when the data frame is transmitted.

[0074] In some embodiments, a data frame exists on a channel, and a channel can transmit multiple data frames.

[0075] In some embodiments, the management platform can determine bandwidth adjustment parameters based on a variety of methods.

[0076] In some embodiments, the management platform may determine bandwidth adjustment parameters based on a first vector database, according to load data of multiple channels, backlog data of priority queues, and priority of data frames.

[0077] In some embodiments, the first vector database includes multiple sets of first feature vectors and their corresponding first vector labels. For example, the management platform can filter historical data to construct multiple reference feature vectors based on load data, priority queue backlog data, and data frame priorities when the channel meets preset conditions within a certain period. The reference feature vectors that meet the preset conditions are determined as the first feature vectors, and the bandwidth adjustment parameters corresponding to the first feature vectors are used as their corresponding first vector labels.

[0078] In some embodiments, the preset condition may be: the channel corresponding to the reference feature vector maintains stable, fault-free operation within a preset time window. The preset time window may be set based on experience. Stable, fault-free operation may mean that no abnormal warnings are triggered.

[0079] An anomaly warning is an alert indicating an anomaly in channel transmission. In some embodiments, in response to channel load data meeting warning conditions, the management platform can send an anomaly warning to administrators. Warning conditions can be set based on experience.

[0080] The management platform can construct a first target vector based on the load data of each channel, the backlog data of the priority queue, and the priority of the data frames. Based on this first target vector, a target vector is retrieved from a first vector database. In some embodiments, the management platform can determine the target vector based on the similarity between the first target vector and multiple first feature vectors in the first vector database. For example, a first feature vector whose similarity to the first target vector meets a preset similarity condition can be used as the first target vector. The preset similarity condition can be set as needed. For example, it could be the maximum similarity or a similarity greater than a first preset similarity threshold. The first preset similarity threshold is set based on human experience.

[0081] In some embodiments, for a first target vector to be matched corresponding to a channel, if a first target vector exists in the first vector database, the bandwidth adjustment parameter of the first target vector is determined to be the bandwidth adjustment parameter of the channel.

[0082] In some embodiments, the management platform determines bandwidth adjustment parameters based on candidate adjustment strategies. For more information on this topic, please refer to [link to relevant documentation]. Figure 4 And its related descriptions.

[0083] Step 240: Generate the first adjustment command based on the bandwidth adjustment parameters and send it to the perception control platform.

[0084] The first adjustment instruction refers to an instruction to adjust the channel transmission bandwidth. In some embodiments, the first adjustment instruction includes adjusting the channel transmission bandwidth according to bandwidth adjustment parameters.

[0085] The channels corresponding to the robotic arm carry motion control commands and multi-dimensional sensor data, while the channels corresponding to environmental auxiliary equipment carry environmental adjustment commands and environmental monitoring data. Because the control commands and sensor data differ for different controlled devices, the bandwidth adjustment parameters also differ for each device.

[0086] In some embodiments, the management platform can adjust parameters based on the bandwidth of different channels to generate differentiated first adjustment instructions. For example, for a robotic arm, the first adjustment instruction generated by the management platform can adjust the total bandwidth of the channel according to the stage of the robotic arm's movement to avoid motion delay. As another example, for environmental assistance equipment, the first adjustment instruction generated by the management platform can maintain or appropriately reduce the total bandwidth of the channel according to network load conditions to ensure the smooth operation of critical equipment.

[0087] In some embodiments, the management platform is further configured to: determine acquisition parameters based on load data and accumulated data; generate acquisition instructions based on the acquisition parameters and send them to the perception control platform; and control sensors to acquire data at a sampling frequency based on the acquisition instructions and upload the acquired data to the edge computing device at an upload frequency.

[0088] Acquisition parameters refer to the parameters that control the sensor's data acquisition operation. Examples include sampling frequency and upload frequency.

[0089] Sampling frequency refers to the frequency at which a sensor collects data.

[0090] Upload frequency refers to the frequency at which sensors upload data to edge computing devices.

[0091] A higher signal-to-noise ratio (SNR) indicates better signal quality and a more stable channel, allowing for higher acquisition and upload frequencies. Conversely, higher utilization and higher packet loss rates indicate less available bandwidth and poorer transmission quality, necessitating a reduction in acquisition and upload frequencies.

[0092] Load data can be characterized by load scores. A load score is a numerical value used to quantify the channel load status. The greater the transmission capacity, the higher the load score.

[0093] In some embodiments, the management platform can obtain the normalized current signal-to-noise ratio, bandwidth utilization, and packet loss rate to determine the load score. Since the sampling frequency and upload frequency are positively correlated with the signal-to-noise ratio and negatively correlated with the utilization and packet loss rate, the management platform can calculate the load score based on formula (3):

[0094] Load score = k1 Current signal-to-noise ratio + k2 (1 - bandwidth utilization) + k3 (1 - packet loss rate) (3)

[0095] Where k1 is the weight corresponding to the current signal-to-noise ratio; k2 is the weight corresponding to the bandwidth utilization rate; and k3 is the weight corresponding to the packet loss rate.

[0096] When the load score exceeds the load threshold score, the management platform can increase the data collection and upload frequency. The load threshold score can be determined based on empirical presets.

[0097] The more data accumulates, the lower the collection and upload frequencies should be. In some embodiments, if the amount of accumulated data in a priority queue exceeds the maximum accumulation limit, the management platform can reduce the collection and upload frequencies, prioritizing the processing of accumulated data in that priority queue to prevent buffer overflow. The buffer refers to the set of priority queues to be processed. The maximum accumulation limit can be determined based on empirical presets.

[0098] Data acquisition instructions are instructions for performing data acquisition.

[0099] In some embodiments, the management platform can generate acquisition instructions based on acquisition parameters. For example, an acquisition instruction could be to acquire data based on acquisition parameters.

[0100] In some embodiments, the management platform can send acquisition instructions to the perception control platform; control the sensors to acquire data at a sampling frequency based on the acquisition instructions, and upload the acquired data to the edge computing device at an upload frequency.

[0101] For more information on edge computing devices, please see [link / reference]. Figure 1 And its related descriptions.

[0102] In some embodiments of this specification, by utilizing load data and combining it with the accumulation status of priority queues, source-end adaptive adjustment of the acquisition frequency and upload frequency of control sensors is achieved. When the channel quality is excellent, the data acquisition accuracy can be automatically improved to capture more details and ensure data integrity at critical moments. When the network is congested or the queue is backlogged, the frequency is intelligently reduced to reduce the incoming network traffic, effectively preventing continuous traffic surges from overwhelming the edge gateway buffer, avoiding disordered packet loss due to buffer overflow, and improving the robustness and transmission stability of the system in fluctuating network environments.

[0103] Step 250: Based on the first adjustment instruction, the edge computing device adjusts the transmission bandwidth of the channel.

[0104] Transmission bandwidth refers to the maximum amount of data that a channel can transmit per unit of time.

[0105] In some embodiments, after receiving the first adjustment instruction, the edge computing device can dynamically adjust the transmission bandwidth of each channel based on the bandwidth adjustment parameters: for the channel corresponding to the robotic arm, when the critical action of the robotic arm corresponding to the control instruction is detected, the edge computing device can increase the transmission bandwidth of the channel and lock the backlog of data in its high-priority queue to ensure that the control instruction passes through quickly and avoid motion delay; for the channel of the environmental auxiliary equipment, the edge computing device can maintain or appropriately reduce its transmission bandwidth according to the load condition to release network resources and ensure the stable operation of other controlled devices.

[0106] In some embodiments of this specification, control commands and sensing data are directly distinguished by parsing the header parameters of the data frame, and a multi-dimensional priority evaluation system for the data frame is constructed by combining failure time and information entropy, thereby achieving accurate identification and priority protection of key commands; at the same time, the transmission bandwidth is dynamically adjusted based on channel load and queue backlog, which effectively solves the problems of high latency of control commands and packet loss of sensing data caused by traditional static allocation, and improves the real-time performance and stability of the industrial control system.

[0107] Figure 3 This is a schematic diagram illustrating preprocessing operations according to some embodiments of this specification.

[0108] like Figure 3 As shown, the management platform performs preprocessing operations on data frames according to their data types to determine the priority queue corresponding to the data frames.

[0109] Preprocessing refers to the processing performed on data frames before transmission. Examples include compressing data frames and placing them into a designated priority queue.

[0110] The management platform can compare the priority of a data frame with a preset priority threshold to determine the corresponding preprocessing operation. The preset priority threshold is based on empirical settings. For example, if the priority of a data frame is less than or equal to the preset priority threshold, the management platform 110 compresses the data frame and places it in a low-priority queue to save bandwidth and reduce data frame redundancy; if the priority of a data frame is greater than the preset priority threshold, the management platform 110 places the data frame in a high-priority queue to ensure that the data frame is processed first and avoids delays.

[0111] In some embodiments, the preprocessing operation includes a first preprocessing 330 and a second preprocessing 340. The management platform is further configured to: perform the first preprocessing 330 on the data frame when the data type of the data frame is a control instruction 310; and perform the second preprocessing 340 on the data frame when the data type of the data frame is sensor data 320.

[0112] For more information on data types, please refer to [link / reference]. Figure 2 And its related descriptions.

[0113] In some embodiments, the first preprocessing includes bypassing the buffer module and the compression module, allowing data frames to directly enter the high-priority queue. The management platform can directly insert data frames of control instruction type into the high-priority queue, saving the time that data frames wait for processing and ensuring that they are processed with priority.

[0114] The buffer module refers to the module in the management platform that temporarily stores data frames and implements data frame shaping and priority scheduling.

[0115] The compression module refers to the module in the management platform that performs lossless compression on data frames.

[0116] In some embodiments, the second preprocessing includes: lossless compression of the data frame using a preset compression algorithm, causing the data frame to enter a low-priority queue. The management platform can losslessly compress data frames of the data type "sensory data" using a preset compression algorithm and insert the data frames into the low-priority queue, reducing data frame redundancy while ensuring that it does not block the priority queue.

[0117] Preset compression algorithms refer to pre-defined compression algorithms, such as LZ4 and ZSTD.

[0118] For explanations of data frames, data types, priorities, and priority queues, please refer to [link / reference]. Figure 2 And its contents.

[0119] In some embodiments of this specification, control commands are directly transmitted to ensure millisecond-level response, and sensing data is compressed at a high rate to reduce bandwidth usage. This "fast-slow separation" processing mechanism, under limited bandwidth resources, not only ensures extremely low latency for core data, but also significantly improves the throughput of massive sensing data.

[0120] In some embodiments of this specification, by performing preprocessing on data frames according to data type, it is possible to ensure rapid transport of core data frames while reducing storage and computational overhead.

[0121] Figure 4 This is an exemplary flowchart illustrating the determination of bandwidth adjustment parameters according to some embodiments of this specification. Figure 4 As shown, process 400 includes the following steps. In some embodiments, process 400 may be executed by a management platform in an IoT-based sensor data collaborative transmission system.

[0122] For more information on bandwidth adjustment parameters, please refer to [link / reference]. Figure 2 And its related descriptions.

[0123] Step 410: Generate candidate adjustment strategies.

[0124] Candidate adjustment strategies refer to the set of alternative solutions for bandwidth adjustment parameters. For more information on bandwidth adjustment parameters, please refer to [link to relevant documentation]. Figure 2 And its related descriptions.

[0125] In some embodiments, the management platform 110 can generate candidate adjustment strategies using a random generation function. For example, the management platform 110 can call the random generation function to generate random values ​​of the transmission bandwidth of all channels, use the set of these random values ​​as a candidate adjustment strategy, and repeat the above random generation process K times (K is the preset number of candidate strategies) to obtain K candidate adjustment strategies.

[0126] In some embodiments, the management platform 110 can also generate candidate adjustment strategies based on the load data, priority queue backlog data, and data frame priorities of all current channels using a first database. For example, the management platform 110 can construct a first matching vector based on the load data, priority queue backlog data, and data frame priorities of all current channels; then, it can obtain the top K first feature vectors (each first feature vector corresponds to the load data, priority queue backlog data, and data frame priorities of all channels) from the first database that rank highly similar to the first matching vector. Each first vector label (each first vector label is a bandwidth adjustment parameter) is used as a candidate adjustment strategy, and the first vector labels corresponding to the K first feature vectors are used as K candidate adjustment strategies. More information on similarity determination and the construction of the first feature vectors can be found in [link to relevant documentation]. Figure 2 And its related descriptions.

[0127] For more information on load data, priority queue backlog data, data frame priority, and the first database, please refer to [link / reference]. Figure 2 And its related descriptions.

[0128] Step 420: Based on the load data, backlog data, data frame priority, and candidate adjustment strategies, determine the first control effect corresponding to the candidate adjustment strategy through the effect prediction model.

[0129] The first regulation effect refers to the regulation effect of the candidate regulation strategy on each channel, which is used to evaluate the performance of the candidate regulation strategy on each channel.

[0130] In some embodiments, a candidate adjustment strategy corresponds to a first regulatory effect.

[0131] The first control effect can be a sequence of control effect values ​​corresponding to multiple channels. Each element in the sequence can be represented as a value between 0 and 1. The larger the value, the better the control effect. For example, after the management platform 110 applies the candidate adjustment strategy to channels A and B, the first control effect of the candidate adjustment strategy includes a control effect of 0.9 for channel A and a control effect of 0.7 for channel B.

[0132] In some embodiments, the management platform 110 can determine the first control effect corresponding to the candidate adjustment strategy by using an effect prediction model based on load data, backlog data, data frame priority, and candidate adjustment strategies.

[0133] In some embodiments, the performance prediction model is a machine learning model. For example, the performance prediction model can be any one or a combination of deep neural networks (DNNs) or other custom model structures.

[0134] like Figure 5 The illustrated schematic diagram shows an example of an effect prediction model. The inputs to the effect prediction model 520 may include load data 511, backlog data 512, data frame priority 513, and candidate adjustment strategies 514. The output may include a first control effect 530. More information regarding load data, backlog data, and data frame priority can be found in the corresponding descriptions above.

[0135] In some embodiments, the input to the effect prediction model 520 further includes the predicted bandwidth requirements 515 for each channel. The input to the effect prediction model 520 includes the predicted bandwidth requirements corresponding to each channel in the first modulation effect.

[0136] Predicted bandwidth requirements refer to the transmission bandwidth required for a given channel to maintain normal operation over a predicted period of time in the future.

[0137] In some embodiments, the management platform 110 can determine the predicted bandwidth requirement in various ways. For example, the management platform 110 can obtain the predicted bandwidth requirement through a machine learning model. More information on how to determine the predicted bandwidth requirement can be found in [link to relevant documentation]. Figure 6 And its related descriptions.

[0138] In some embodiments of this specification, by supplementing the predicted bandwidth requirement of the channel as input to the effect prediction model, the effect prediction model can further combine the future bandwidth requirement of the channel when predicting the first control effect, thereby improving the accuracy of predicting the effect of candidate adjustment strategies.

[0139] In some embodiments, the performance prediction model can be trained based on a large number of first training samples with first training labels. The management platform 110 can input multiple first training samples with first training labels into the initial performance prediction model, construct a loss function using the first training labels and the results of the initial performance prediction model, and iteratively update the parameters of the initial performance prediction model based on the loss function using methods such as gradient descent. When the loss function meets preset training conditions, a trained performance prediction model is obtained. These preset training conditions may include loss function convergence, the number of iterations reaching a threshold, etc.

[0140] The first training sample and the first training label can be obtained based on historical data. The first training sample may include historical load data, historical backlog data, historical data frame priorities, and historical bandwidth adjustment parameters for historical time periods. The first training label may include the first adjustment effect of the candidate adjustment strategy corresponding to the first training sample.

[0141] In some embodiments, after the management platform 110 adjusts the transmission bandwidth of all channels according to the historical bandwidth adjustment parameters, the management platform 110 evaluates the actual first control effect of the historical bandwidth adjustment parameters on each channel based on the idle rate and transmission delay of each channel in the future period, and determines it as the first training label.

[0142] Idle rate refers to the proportion of time a channel will be in a state of no data transmission within a certain period of time in the future, and is used to characterize the degree of idleness of channel bandwidth resources. Total duration is the total length of the future period.

[0143] In some embodiments, the management platform 110 can count the cumulative duration (i.e., idle time) of the channel in a state of no data transmission over a future period of time, and divide the idle time by the total duration to obtain the channel idle rate.

[0144] Transmission delay refers to the total time it takes for data to travel from the priority queue of the channel to its successful delivery to the target node. It is used to measure the transmission response efficiency of the channel. The target node refers to the receiving device or network node that the data will eventually reach after being transmitted through the channel.

[0145] In some embodiments, the management platform 110 can record the initial time when a data frame enters the channel transmission queue and the completion time when the data frame is successfully sent to the target node. The difference between the completion time and the initial time is the transmission delay duration.

[0146] In some embodiments, the management platform 110 can quantify the idle rate and transmission latency into an idle score and a latency score, respectively. The first regulation effect is positively correlated with the idle score and the latency score. For example, the idle score can be obtained based on formula (4):

[0147] (4)

[0148] In some embodiments, the management platform 110 first sets the maximum tolerable delay duration (which can be adjusted according to requirements). The maximum tolerable delay duration refers to the longest acceptable delay threshold for a service when transmitting data through the channel.

[0149] If the transmission delay exceeds the maximum tolerance value, the delay score is recorded as 1; otherwise, the delay score can be calculated based on the transmission delay and the maximum tolerance delay.

[0150] For example, management platform 110 can be obtained based on formula (5):

[0151] (5)

[0152] The management platform 110 performs weighted fusion processing on the idle score and delay score of each channel in the first control effect. The fusion result is the value of each element (between 0 and 1) in the sequence corresponding to the first control effect. The weighted fusion processing can satisfy formula (6):

[0153] (6)

[0154] In formula (6), w is the weight, which can be preset manually based on experience, and w is between 0 and 1.

[0155] The larger the value of the first control effect, the better the first control effect of the historical bandwidth adjustment parameter on a certain channel.

[0156] Step 430: Based on the first regulatory effect, determine the second regulatory effect of the candidate adjustment strategy.

[0157] The second control effect refers to the comprehensive control effect of the candidate control strategy on all channels. For example, if a candidate control strategy involves channels A and B, after applying the candidate control strategy, the first control effect corresponding to this candidate control strategy is 0.7 for channel A and 0.8 for channel B. The management platform 110 combines the weight coefficients of the two channels (both channel A and channel B have a weight of 0.5), performs a weighted calculation and summation, and obtains the comprehensive control effect (i.e., the second control effect) corresponding to the candidate control strategy, which is 0.75. More information about weight coefficients and how to determine the second control effect can be found below and in related descriptions.

[0158] In some embodiments, the management platform 110 may use the lowest numerical value of the first control effect among the candidate adjustment strategies as the second control effect of that candidate adjustment strategy. For example, if the first control effects of the candidate adjustment strategies are 0.7 and 0.8 for channel A and channel B respectively, the management platform 110 may use the lowest numerical value of the first control effect of channel A (i.e., 0.7) as the second control effect of the candidate adjustment strategy. By using the lowest numerical value of the first control effect among the candidate adjustment strategies as the second control effect, the minimum lower limit of the overall control effect can be guaranteed. Further details on how to determine the second control effect can be found below and in the related description.

[0159] In some embodiments, the management platform is further configured to: determine the weighting coefficient of the channel based on the state characteristics and task characteristics of the robotic arm corresponding to the channel; and determine the second control effect based on the weighting coefficient and the first control effect.

[0160] The state characteristics of a robotic arm represent a set of key parameters describing the current operating status of the robotic arm for each channel. For example, the state characteristics of a robotic arm may include its position, posture, and gripping force. Different robotic arms may have different state characteristics, which is not limited here.

[0161] In some embodiments, the state characteristics of the robotic arm can be obtained by torque sensors, position sensors, etc. on the robotic arm corresponding to each channel.

[0162] Task characteristics refer to the attributes of the currently executing task in each channel. For example, task characteristics may include task type, task stage, etc. Different robotic arms may have different task characteristics, which are not limited here.

[0163] In some embodiments, task characteristics can be obtained by the management platform 110 directly accessing data stored in the data center.

[0164] In some embodiments, task types may include high-precision tasks (such as precision parts assembly tasks) and low-precision tasks (such as ordinary material handling tasks).

[0165] In some embodiments, the task phase may include an operation phase, an approach phase, an idle phase, etc.

[0166] The operation phase may include precision welding, high-precision gripping, etc. This phase requires high-frequency data feedback from torque sensors to accurately control the contact force of the robotic arm, thus requiring high real-time performance and data integrity in channel transmission.

[0167] The approach phase can include stages such as no-load return to zero and workpiece approach. During this phase, only the position of the robotic arm needs to be monitored, and the channel is allowed to adopt a certain degree of transmission frequency reduction or data compression strategy.

[0168] The idle phase is a standby state where the robotic arm has no tasks to perform. During this phase, the channel connection only needs to be maintained through low-frequency heartbeat packets.

[0169] The weighting coefficient is a numerical value that characterizes the importance of the corresponding channel.

[0170] In some embodiments, the management platform 110 can determine the weight coefficients of each channel based on the state and task characteristics of the robotic arm corresponding to each channel, using a second database. For example, the management platform 110 constructs a second matching vector based on the state and task characteristics of the robotic arm corresponding to each channel. Then, based on the second matching vector, it searches the second database to obtain the second feature vector with the highest similarity to the second matching vector as the second target vector, and determines the second vector label corresponding to the second target vector as the weight coefficient of the current channel. The management platform 110 repeats the above operation for all channels to obtain the weight coefficients corresponding to each channel.

[0171] In some embodiments, the management platform 110 calculates the vector distance between the second vector to be matched and multiple second feature vectors to determine the similarity, including Euclidean distance, etc.

[0172] The second database includes multiple second feature vectors and multiple second vector labels corresponding to the multiple second feature vectors.

[0173] The second feature vector includes the state and task characteristics of the robotic arm corresponding to a historical channel. The second vector label corresponding to the second feature vector is the weight coefficient of the actual channel corresponding to that second feature vector.

[0174] In some embodiments, the management platform 110 can determine a preset time window based on the system's historical operation logs. This preset time window is set manually based on experience. Within this preset time window, target periods corresponding to each channel where the robotic arm maintains fault-free and stable operation are selected. Fault-free and stable operation means that no abnormal warnings are triggered, or all high-precision tasks are successfully completed. The management platform 110 extracts the actual weighting coefficients used for each channel within the target period and uses these weighting coefficients as the second vector labels corresponding to each second feature vector within the target period.

[0175] The management platform 110 repeats the above-mentioned filtering, extraction and second vector label generation operations, collects the channel weight coefficients under multiple different combinations of robotic arm state features and task features, and forms a sufficient number of second vector labels covering multiple scenarios in the second database to meet the matching requirements of different second feature vectors.

[0176] In some embodiments, the management platform 110 can multiply the corresponding weight coefficient of a single channel with the first control effect of that channel according to the weight coefficient and the first control effect, to obtain the weighted result of that channel; then, the weighted results of all channels are summed to obtain the final summed result as the second control effect.

[0177] Some embodiments in this specification determine the channel weight coefficient by fusing the state characteristics and task characteristics of the robotic arm, and calculate the second control effect by combining the first control effect. This can dynamically adapt to the channel resource allocation requirements of different operation scenarios, allowing the channel to receive resource tilt during critical operation stages, ensuring stable and uninterrupted communication links, reducing the risk of robotic arm movement deviations or safety accidents caused by network fluctuations, and improving the reliability and security of the IoT-based sensor data collaborative transmission system.

[0178] Step 440: Determine the bandwidth adjustment parameters based on the second control effect.

[0179] For more information on bandwidth adjustment parameters, please refer to [link / reference]. Figure 2 And its related descriptions.

[0180] In some embodiments, the management platform 110 calculates the corresponding second control effect for each of the multiple candidate adjustment strategies, selects the candidate adjustment strategy with the highest second control effect value, and determines the candidate adjustment strategy as the final bandwidth adjustment parameter.

[0181] Some embodiments in this specification generate candidate adjustment strategies, combine load data, backlog data, and data frame priority, use an effect prediction model to determine multiple first control effects for multiple channels, and further obtain second control effects, and finally select the optimal bandwidth adjustment parameters. The effect of the candidate adjustment strategies can be predicted and evaluated before the candidate adjustment strategies are executed, and inferior strategies that are prone to causing network congestion can be avoided in advance. With the mechanism of prediction before decision-making, the utilization rate of channel resources and low transmission latency are taken into account, and global optimal scheduling in complex network environments is achieved.

[0182] Figure 6 This is an exemplary schematic diagram of a bandwidth prediction model according to some embodiments of this specification.

[0183] In some embodiments, the management platform is further configured to: determine the predicted bandwidth requirement 630 of the channel based on the channel's historical traffic sequence 611, load data 511, and service type 612, using a bandwidth prediction model 620; wherein the bandwidth prediction model 620 is a machine learning model; determine future adjustment parameters 640 based on the predicted bandwidth requirement 630; and generate a second adjustment instruction 650 based on the future adjustment parameters 640, the second adjustment instruction 650 including the channel's transmission bandwidth at a future time.

[0184] For more information on load data, please refer to [link / reference]. Figure 2 And related descriptions. For more information on predicted bandwidth requirements, please refer to... Figure 5 And related explanations.

[0185] Historical traffic sequence refers to the traffic sequence of a channel within a certain historical period.

[0186] In some embodiments, the historical traffic sequence can be extracted by the management platform 110 from the number of data frames of each channel in a certain historical period and arranged into a historical traffic sequence in chronological order.

[0187] Service type refers to the type of service that the channel is responsible for transmitting. For example, service type may include high-frequency feedback service of the torque sensor of the robotic arm, low-frequency reporting service of robotic arm position monitoring, and system status log transmission service, etc., which are robotic arm operation services.

[0188] In some embodiments, the management platform 110 can obtain the information directly through the business system.

[0189] In some embodiments, the management platform 110 can determine the predicted bandwidth requirement of the channel based on the channel's historical traffic sequence, load data, and service type using a bandwidth prediction model.

[0190] The bandwidth prediction model can be a machine learning model. For example, the bandwidth prediction model can be any one or a combination of Long Short-Term Memory (LSTM) networks or other custom model structures.

[0191] In some embodiments, the inputs to the bandwidth prediction model may include historical traffic sequences, load data, and service types, and the output may include predicted bandwidth requirements.

[0192] The second training sample and the second training label can be obtained based on historical data. Each second training sample may include historical traffic sequences, historical load data, and historical service types for a historical period, and the corresponding second training label may include the actual predicted bandwidth requirement corresponding to the second training sample.

[0193] In some embodiments, the bandwidth prediction model can be trained based on a large number of second training samples with second training labels. The management platform 110 can input multiple second training samples with second training labels into the initial bandwidth prediction model, construct a loss function using the second training labels and the results of the initial bandwidth prediction model, and iteratively update the parameters of the initial bandwidth prediction model based on the loss function using methods such as gradient descent. When the loss function meets preset training conditions, a trained bandwidth prediction model is obtained. These preset training conditions may include loss function convergence, the number of iterations reaching a threshold, etc.

[0194] In some embodiments, the bandwidth prediction model can be jointly trained with the performance prediction model, or it can be trained separately. More information about the performance prediction model can be found at [link to relevant documentation]. Figure 5 And its related descriptions.

[0195] In some embodiments, the management platform 110 can filter out the target period that maintains stable operation without faults within a preset time window based on the system's historical operation logs, extract the actual transmission bandwidth of each channel within the target period, and determine it as the second training label corresponding to the second training sample; at the same time, it can obtain the traffic sequence, load data and service type of all channels within a period of time before the target period (such as 1 hour before the target period), and combine these data to construct the second training sample.

[0196] Future adjustment parameters refer to bandwidth adjustment parameters at future times. For example, future adjustment parameters could be "the transmission bandwidth of channel A at T+15min (i.e., 15 minutes after the current time) is 15Mbps (this 15Mbps is the bandwidth adjustment parameter)" or "the transmission bandwidth of channel B at T+15min is 9Mbps (this 9Mbps is the bandwidth adjustment parameter)".

[0197] In some embodiments, the management platform 110 can determine future adjustment parameters based on predicted bandwidth requirements. One future adjustment parameter corresponds to all channels.

[0198] For example, the management platform 110 multiplies the predicted bandwidth requirement of the channel output by the bandwidth prediction model by a preset redundancy coefficient (such as 1.1, 1.2, etc.) for each channel to obtain the expected bandwidth quota of the channel at a future time.

[0199] Redundancy coefficients are introduced to compensate for small deviations in predicted bandwidth demand, preventing channel congestion caused by underestimating the predicted bandwidth demand. A small deviation refers to the minor difference between the predicted bandwidth demand output by the bandwidth prediction model and the actual future bandwidth demand of the channel. For example, if the bandwidth prediction model predicts a future channel demand of 10 Mbps, and the actual demand is 10.5 Mbps, the difference of 0.5 Mbps is a small deviation. Underestimating bandwidth demand means that the predicted bandwidth demand is less than the actual bandwidth demand.

[0200] Then, taking a future moment as an example, the management platform 110 calculates the sum of the expected bandwidth quotas (i.e., total demand) of all channels at that future moment, compares the total demand with the total bandwidth limit of the physical links of the edge computing devices, and if the total demand is less than or equal to the total bandwidth limit, then the expected bandwidth quota of each channel is directly determined as the future adjustment parameter of that channel.

[0201] If the total demand exceeds the total bandwidth limit, the expected bandwidth quota of the channels related to the robotic arm operation service will be prioritized according to the service type. The management platform 110 subtracts the total demand of the channels related to the robotic arm operation service from the total bandwidth limit, and the difference is the remaining available bandwidth. For the channels of other service types, the management platform 110 allocates the remaining available bandwidth according to the proportion of the expected bandwidth quota of these channels to obtain the corrected expected bandwidth quota of the channels of other service types.

[0202] The proportion of expected bandwidth quota for channels of other service types refers to the ratio of the expected bandwidth quota of a single channel of other service types to the sum of the expected bandwidth quotas of all channels of other service types.

[0203] The management platform 110 combines the expected bandwidth quota of the channel related to the robotic arm operation with the corrected expected bandwidth quota of the channels for other service types to obtain a future adjustment parameter covering all channels at a single future time. The management platform 110 then performs the above operation for other future times to obtain multiple future adjustment parameters.

[0204] The second adjustment instruction refers to an instruction used to control the channel bandwidth configuration. In some embodiments, the second adjustment instruction includes the transmission bandwidth of the channel at a future time. For example, the second adjustment instruction may include "when the current time reaches T+15min (15 minutes after the current time), immediately adjust the transmission bandwidth of channel A to 15Mbps, and at the same time adjust the transmission bandwidth of channel B to 9Mbps", etc.

[0205] In some embodiments, the management platform 110 generates a control instruction with timing triggering characteristics based on future adjustment parameters, and uses the control instruction as a second adjustment instruction.

[0206] Some embodiments in this specification analyze historical traffic sequences, load data, and service types of the channel using bandwidth prediction models (such as LSTM models) to accurately determine the predicted bandwidth demand, thereby determining future adjustment parameters and generating a second adjustment command. This achieves forward-looking bandwidth regulation, shifting from post-congestion management to pre-congestion prevention. It reserves the transmission channel at the millisecond level before the traffic peak arrives, eliminating the instantaneous packet loss caused by bandwidth adjustment lag.

[0207] The basic concepts have been described above. Obviously, for those skilled in the art, the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0208] Furthermore, this specification uses specific terms to describe embodiments thereof. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Moreover, certain features, structures, or characteristics in one or more embodiments of this specification can be appropriately combined.

[0209] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this specification are not intended to limit the order of the processes and methods described herein. Although various examples have been discussed in the foregoing disclosure of some embodiments of the invention that are currently considered useful, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the spirit and scope of the embodiments described herein. For example, while the system components described above can be implemented using hardware devices, they can also be implemented solely using software solutions, such as installing the described system on existing servers or mobile devices.

[0210] Similarly, it should be noted that, in order to simplify the description disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments in this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0211] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A sensor data collaborative transmission system based on the Internet of Things, characterized in that, The system includes a management platform and a sensing and control platform; The management platform is configured as follows: The data type of the data frame is determined based on the header parameters of the data frame; The priority of the data frame is determined based on the data type, expiration time, and information entropy, where information entropy is used to quantify the degree of deviation between the current data frame and historical data. Based on the channel load data, the priority queue backlog data, and the priority of the data frame, bandwidth adjustment parameters are determined; the management platform is further configured to: Generate candidate adjustment strategies; Based on the load data, the backlog data, the priority of the data frames, and the candidate adjustment strategies, a first control effect corresponding to the candidate adjustment strategies is determined through an effect prediction model. The first control effect refers to the control effect of the candidate adjustment strategies on each of the channels. The effect prediction model is a machine learning model. Each first control effect corresponds to one of the channels. Based on the first control effect, the second control effect of the candidate adjustment strategy is determined, wherein the second control effect refers to the comprehensive control effect of the candidate adjustment strategy on all the channels; Based on the second control effect, the bandwidth adjustment parameter is determined, and the management platform is configured to: calculate the corresponding second control effect for each of the multiple candidate adjustment strategies, filter out the candidate adjustment strategy with the highest second control effect value, and determine the candidate adjustment strategy with the highest second control effect value as the bandwidth adjustment parameter; Based on the bandwidth adjustment parameters, a first adjustment command is generated and sent to the sensing and control platform; The perception control platform includes an edge computing device, which is configured to: Based on the first adjustment instruction, the transmission bandwidth of the channel is adjusted.

2. The system according to claim 1, characterized in that, The management platform is further configured as follows: Based on the state characteristics and task characteristics of the robotic arm corresponding to the channel, the weighting coefficient of the channel is determined; and The second control effect is determined based on the weighting coefficient and the first control effect.

3. The system according to claim 1, characterized in that, The management platform is further configured as follows: Based on the historical traffic sequence of the channel, the load data, and the service type, the predicted bandwidth requirement of the channel is determined using a bandwidth prediction model; wherein, the bandwidth prediction model is a machine learning model. Based on the predicted bandwidth demand, determine the future adjustment parameters; and Based on the future adjustment parameters, a second adjustment instruction is generated, which includes the transmission bandwidth of the channel at a future time.

4. A sensor data collaborative transmission method based on the Internet of Things, characterized in that, The method is implemented based on a management platform and includes: The data type of the data frame is determined based on the header parameters of the data frame; The priority of the data frame is determined based on the data type, expiration time, and information entropy, where information entropy is used to quantify the degree of deviation between the current data frame and historical data. A bandwidth adjustment parameter is determined based on the channel load data, the priority queue backlog data, and the priority of the data frame. This determination includes: generating candidate adjustment strategies; determining a first control effect corresponding to the candidate adjustment strategy using an effect prediction model based on the load data, the backlog data, the priority of the data frame, and the candidate adjustment strategy, wherein the first control effect refers to the control effect of the candidate adjustment strategy on each of the channels; wherein the effect prediction model is a machine learning model; each first control effect corresponds to one channel; determining a second control effect of the candidate adjustment strategy based on the first control effect, wherein the second control effect refers to the comprehensive control effect of the candidate adjustment strategy on all the channels; and determining the bandwidth adjustment parameter based on the second control effect, wherein determining the bandwidth adjustment parameter based on the second control effect includes: calculating the corresponding second control effect for each of the multiple candidate adjustment strategies, selecting the candidate adjustment strategy with the highest second control effect value, and determining the candidate adjustment strategy with the highest second control effect value as the bandwidth adjustment parameter. Based on the bandwidth adjustment parameters, a first adjustment command is generated and sent to the sensing control platform, which includes an edge computing device; and The edge computing device adjusts the transmission bandwidth of the channel based on the first adjustment instruction.

5. The method according to claim 4, characterized in that, The step of determining the second regulatory effect of the candidate adjustment strategy based on the first regulatory effect includes: Based on the state characteristics and task characteristics of the robotic arm corresponding to the channel, the weighting coefficient of the channel is determined; and The second control effect is determined based on the weighting coefficient and the first control effect.

6. The method according to claim 4, characterized in that, The method further includes: Based on the historical traffic sequence of the channel, the load data, and the service type, the predicted bandwidth requirement of the channel is determined using a bandwidth prediction model; wherein, the bandwidth prediction model is a machine learning model. Based on the predicted bandwidth demand, determine the future adjustment parameters; and Based on the future adjustment parameters, a second adjustment instruction is generated, which includes the transmission bandwidth of the channel at a future time.

7. A computer-readable storage medium, characterized in that, The storage medium stores computer instructions. When the computer reads the computer instructions from the storage medium, the computer executes the sensor data collaborative transmission method as described in claim 4.

Images