Industrial Internet of Things control method for deploying multiple types of working parameters

Abstract

The invention discloses an industrial Internet of Things control method for multi-type working parameter deployment, which relates to the technical field of industrial automation control, and comprises the following steps: S001, establishing a time sequence observation baseline of a multi-stage redundant path, injecting micro-amplitude identification pulses into all control nodes, generating an instruction time delay distribution curve and a resource occupation track, and establishing a time sequence observation baseline of a multi-stage redundant path; the method is used for revealing potential conflict sources and providing initial conditions. According to the method, a conflict source is revealed by establishing a time sequence observation baseline, competition nodes are identified by using a loop retrieval graph, a conflict list is formed, a single-source control right token is issued to realize link convergence, token stagnation is monitored, snapshot rollback and standby path takeover are combined to avoid interruption, and after takeover, time delay shaping is implemented to ensure stable landing of an instruction. Finally, the topological weight, the token priority and the fusing threshold value are dynamically updated, a self-evolution closed loop is constructed, and high reliability and high safety of industrial Internet of Things control are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial automation control, and in particular to an industrial Internet of Things control method for multi-type working parameter deployment. BACKGROUND

[0002] Industrial Internet of Things control for multi-type working parameter deployment refers to, in an industrial Internet of Things environment, for multiple types of working parameters (such as temperature, pressure, flow, current, voltage, speed, torque, process time, etc.) involved in production processes or device operation, collecting various parameter data in real time through distributed sensing and interconnection, and dynamically adjusting and cooperatively optimizing parameter configurations of each working link based on intelligent analysis and decision algorithms. This method not only supports joint deployment of multi-dimensional parameters, but also enables adaptive control and cooperative response across devices and processes, thereby improving the flexibility, autonomy and operating efficiency of the production system. The core lies in using the real-time interconnection and data fusion capabilities of the industrial Internet of Things to integrate multiple heterogeneous devices and sensing nodes into a unified control network, efficiently dynamically adjusting key parameters under different working conditions, and realizing intelligent optimization and safety protection of the production process.

[0003] The prior art has the following disadvantages: In the prior art, in order to improve the reliability and fault tolerance of industrial Internet of Things control, multi-level redundant control paths are usually used to realize dynamic deployment of key parameters. However, in the process of multi-type working parameter deployment, the deployment instructions are prone to loop deadlock in the multi-level redundant control paths due to improper network structure design, instruction priority conflicts or failure of real-time scheduling mechanism, etc. At this time, the control right is repeatedly contended between different control nodes, and the core deployment command cannot be finally confirmed by any node and delivered to the actual execution end, resulting in that the deployment instructions are long-term stranded in the network and cannot be smoothly implemented to the key process link. As a result, important process links are in an out-of-control state for a long time without effective adjustment, devices are always running in a high-risk interval beyond the safety limit, and serious damage to devices, process instability and even production safety accidents and other major consequences are easily caused.

[0004] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] The purpose of the present application is to provide an industrial Internet of Things control method for multi-type working parameter deployment to solve the problems in the background.

[0006] In order to achieve the above purpose, the present application provides the following technical solution: an industrial Internet of Things control method for multi-type working parameter deployment, comprising the following steps: S001, establish a timing observation baseline of multi-level redundant paths, inject a micro-identified pulse into all control nodes to generate an instruction delay distribution curve and a resource occupation trajectory, which are used to reveal potential conflict sources and provide initial conditions; S002, based on the instruction delay distribution curve and the resource occupation trajectory, construct a loop search graph, calculate the instruction backflow probability and the stagnation time, identify the closed loop structure and the competing nodes, and form a conflict list; S003, according to the conflict list, issue a single-source control token, which flows along the shortest reliable path in one direction, and the control nodes that do not hold the token are prohibited from writing to the execution end, ensuring unique control and making the control process converge to a single link; S004, monitor the token stagnation state during the token transfer process, dynamically compare the instruction delay distribution curve with the token position, and if the token stagnation exceeds the threshold, roll back to the last stable snapshot and start the standby path takeover to avoid control interruption due to path failure; S005, after the standby path takeover, implement delay shaping for the deployment instructions, dynamically adjust the sending tempo, load window and queuing weight using real-time residual information to weaken the self-excited oscillation effect caused by path switching and ensure that the deployment instructions are stably landed on the execution end; S006, after the stable execution of the deployment instructions, analyze the execution residual, the fuse record and the takeover times to dynamically update the topology weight, the token priority and the fuse threshold, build a self-evolving closed loop mechanism, and continuously reduce the risk of deadlock and realize the whole process of logical closed loop.

[0007] Preferably, step S001 comprises: In the process of initializing the multi-level redundant paths, each redundant path is divided into a continuous link composed of a head input node, an intermediate transfer node and a tail execution node; Inject a small-amplitude identification pulse that does not affect process execution to all control nodes under a unified time reference, and give a unique time stamp before the identification pulse is sent out; When the identification pulse passes through each control node, record the arrival time and the departure time, and simultaneously record the calculation time, the storage space occupation ratio and the communication bandwidth occupation ratio during the passage of the pulse; After the identification pulse is delivered, the recorded data of each node is collected and sorted, the transmission time difference of the identification pulse between adjacent nodes is calculated to form a delay distribution curve, and the calculation time, the storage space occupation ratio and the communication bandwidth occupation ratio are combined to form a resource occupation trajectory, thereby constructing a timing observation baseline that can truly reflect the node behavior and path running state.

[0008] Preferably, step S002 comprises: On the basis of the obtained instruction latency distribution curve and resource occupation trajectory, all control nodes are mapped as independent points, and each point is marked with average transmission latency and resource occupation, the connection relationship between nodes is mapped as a directed edge and marked with average transmission time, thereby forming a loop search graph covering all redundant paths; In combination with the instruction latency distribution curve and the resource occupation trajectory, each directed edge and node is analyzed to calculate instruction backflow probability and stagnation duration, thereby obtaining path-level risk parameters and node-level risk parameters; All paths in the loop search graph are traversed one by one to identify closed loop structures and calculate average backflow probability and cumulative stagnation duration, and the closed loop with a risk index exceeding a threshold value is marked as a potential conflict loop, and the node with the highest backflow probability and stagnation duration in the loop is marked as a competition node; The identification results are sorted to form a conflict list containing closed loop numbers and path descriptions, closed loop risk indexes and competition node lists, thereby providing a basis for subsequent control token issuance.

[0009] Preferably, step S003 comprises: After generating the conflict list, the node with the lowest risk index outside the closed conflict loop is selected as the token starting node based on the backflow probability, stagnation duration and resource occupation level in the conflict list, and a unique number and arbitration cycle timestamp are assigned to the token to ensure uniqueness when the token is generated; When determining the token circulation path, the shortest transmission time and the lowest risk path are selected according to the path information of the conflict list by comparing the average latency, cumulative stagnation duration and resource occupation proportion of each node, and the token is specified to circulate unidirectionally and sequentially along the path; During the token transmission process, the node receiving the token needs to verify the number, timestamp and state marker, and after verification, the node can obtain the permission to write to the execution end, and after writing is completed, the token must be transmitted to the next node within the specified time limit, and the node without the token is prohibited from writing allocation instructions to the execution end; After the token completes a round of circulation and returns to the starting node, the transmission latency, writing result and stagnation condition are confirmed, and if an abnormality is found, the corresponding node is marked as a high-risk node and the conflict list is updated to ensure that the control process converges to a single link and forms a dynamically optimized closed loop.

[0010] Preferably, step S004 comprises: When the token enters the control node, the arrival time and stay duration are recorded and compared with the pre-established instruction latency distribution curve in real time, and when the actual stay duration exceeds the normal range, an abnormality is preliminarily determined; The abnormality is confirmed in combination with the resource occupation track. When the calculation time continuously rises, the storage space usage ratio is close to full load, the communication bandwidth occupation ratio greatly increases, and the stay duration exceeds the preset threshold, it is determined that the node enters the congestion state; After confirming that the token stagnation exceeds the threshold, the current loop is immediately revoked, all related nodes are prohibited from continuing to write, and the execution end and node state are rolled back to the last successful execution stable snapshot, the key parameter value, token number and path order are restored; After the rollback is completed, a backup path with short average delay, low stagnation duration and safe resource occupation is selected as a new path according to the conflict list and loop search diagram, a new token is generated at the starting node and transmitted to the execution end along the backup path, the deployment process is ensured to be continuous and stable, and the abnormal node information is updated to the conflict list to form a dynamic closed loop.

[0011] Preferably, step S005 comprises: After the backup path takes over, residual information in the transmission process before and after the takeover is obtained, and a residual table containing node numbers, residual delays, calculation consumption durations, storage occupation ratios and bandwidth occupation ratios is formed; According to the residual table, the sending beat of the deployment instruction is dynamically adjusted. When the node delay difference value exceeds the snapshot reference value of 10 milliseconds, the sending beat interval is extended to twice the original setting value. When the overall transmission speed of the node is faster than the snapshot reference value of 5 milliseconds or more, the sending beat interval is shortened to 70% of the original setting value; While adjusting the sending beat, the load window is dynamically controlled. When the node storage occupation ratio exceeds 85% and the communication bandwidth occupation ratio exceeds 75%, the load window is reduced to half of the original setting value. When the node storage occupation ratio is less than 50% and the communication bandwidth occupation ratio is less than 40%, the load window is expanded to twice the original setting value; On the basis of the adjustment of the sending beat and the load window, the queuing weight is optimized. When the residual delay exceeds 15 milliseconds, the instruction is assigned a priority value of 10 and is transmitted in advance in the queuing order. When the residual delay is less than 5 milliseconds, the instruction is assigned a priority value of 2 and is transmitted late in the queuing order, ensuring that the deployment instruction is finally smoothly landed on the execution end, and recording the residual information after landing to update the basis for subsequent path optimization.

[0012] Preferably, step S006 comprises: After the deployment instruction is stably executed, three types of data, execution residual, fuse record and takeover number, are collected and formed. The execution residual is the difference between the actual transmission time and the stable snapshot reference time. The fuse record is the number of times the fuse is triggered, the node number and the corresponding calculation resource consumption, the storage space usage ratio and the communication bandwidth occupation ratio. The takeover number is the number of times and the duration of the backup path enabled in the current cycle; After the data collection is completed, the execution residual, the fuse record and the takeover times are analyzed, when the residual of a node is higher than 15 milliseconds and lasts for multiple rounds, it is determined that the node is a high-risk node, when the number of times of triggering the fuse of a node in ten cycles is more than five, the reliability of the node is reduced, when the number of times of taking over of a standby path in ten cycles is more than six, the priority of the path is reduced; After the data analysis is completed, the topology weight, the token priority and the fuse threshold are dynamically adjusted according to the results, when a node frequently fuses, the topology weight of the node is reduced, when a node performs stably, the topology weight of the node is increased, when a path is reliable, the token priority is increased, when a node frequently triggers the fuse by mistake, the fuse threshold is increased, when a key node is stalled, the fuse threshold is reduced. After the parameters are updated, the updated topology weight is used to recalculate the shortest reliable path, the updated token priority is used to determine the flow transfer sequence, and the updated fuse threshold is used for stall monitoring, so that in the next round of operation, the path optimization, risk avoidance and control closed loop are realized, and the control process has the ability of self-evolution and continuously reduces the risk of deadlock.

[0013] In the above technical solution, the technical effects and advantages provided by the present application are: The present application firstly realizes the accurate revelation of potential conflict sources by establishing the timing observation baseline and generating the instruction delay distribution curve and the resource occupation trajectory at all control nodes; then the closed loop structure and the competing nodes are identified by means of loop search graph to form a conflict list, which provides the basis for subsequent arbitration; then the deployment process converges to a single reliable link by issuing a single-source control token, eliminating the hidden danger of multiple nodes competing for control; the stall state of the token is monitored in real time during the token transfer, and the snapshot rollback and standby path takeover are combined to effectively avoid the global control interruption caused by path blocking; after the standby path takeover, delay shaping is implemented, the residual information is used to dynamically adjust the sending tempo, the load window and the queuing weight, eliminating the oscillation effect caused by path switching and ensuring that the deployment instruction can be smoothly landed on the execution end; finally, the topology weight, the token priority and the fuse threshold are dynamically updated through comprehensive analysis of the execution residual, the fuse record and the number of takeovers, forming a closed loop mechanism that can continuously evolve itself. Therefore, the present application not only ensures the real-time stable deployment of multiple types of working parameters, but also continuously reduces the risk of deadlock in long-term operation, realizing high reliability, high security and high robustness of industrial Internet of Things control. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0015] Figure 1 A method flowchart of the multi-type working parameter deployment industrial internet of things control method of the present application. DETAILED DESCRIPTION

[0016] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the scope of example implementations to those skilled in the art.

[0017] The present application provides a multi-type working parameter deployment industrial internet of things control method as shown in Figure 1 The multi-type working parameter deployment industrial internet of things control method comprises the following steps: S001, establishing a time sequence observation baseline of multi-level redundant paths, injecting a micro-identification pulse into all control nodes to generate an instruction time delay distribution curve and a resource occupation trajectory, for revealing potential conflict sources and providing initial conditions; In the multi-type working parameter deployment industrial internet of things control method, in order to avoid the problems of deployment instruction retention and loop deadlock under the multi-level redundant path, a time sequence observation baseline that can truly reflect the node behavior and path running state needs to be established first. This observation baseline is not only used to form a complete instruction time delay distribution curve and resource occupation trajectory, but also provides a verifiable initial condition for subsequent conflict identification, arbitration and recovery. The specific implementation is as follows: In the process of initializing the multi-level redundant path, each redundant path is divided into a continuous link composed of a first-end input node, an intermediate transmission node and a last-end execution node according to the actual arrangement. In order to capture the real propagation in the path, a small-amplitude identification pulse that does not affect the process execution needs to be injected into all control nodes one by one under a unified time reference. Each identification pulse is assigned a unique time stamp before being sent out, which is used to mark the time when it is generated and the corresponding path. When the identification pulse passes through each control node, the arrival time and the departure time are recorded by the collection device in the node, and the resource occupation of the node during the passage of the pulse is also recorded. The resource occupation here needs to include three aspects: first, the calculation time consumed by the node in processing the identification pulse, which is measured in milliseconds as the minimum unit; second, the storage space occupation ratio of the node for caching the identification pulse, which is measured in bytes as the minimum unit and converted into the percentage of the total capacity; third, the communication bandwidth occupation ratio of the node when the identification pulse passes through, which is measured in bits per second and converted into the percentage of the maximum bandwidth of the link. Through the unified injection and recording of all nodes, a complete and synchronous set of raw data can be formed in the entire multi-level redundant path.

[0018] After all the identification pulses complete the transmission, the data recorded by each node needs to be summarized and sorted. By calculating the transmission time difference of the identification pulses between adjacent nodes, the time delay distribution curve of the redundant path is formed in turn. At the same time, the calculation time, storage space occupation ratio and communication bandwidth occupation ratio of the node during the pulse transmission are combined, and these parameters are constructed into a resource occupation trajectory in chronological order. In this way, each redundant path not only has a curve about the pulse transmission speed, but also has a trajectory about the node resource load condition, thereby realizing the two-dimensional description of the path running state. When forming the time delay distribution curve, the strict correspondence between the curve and the time stamp needs to be maintained to ensure that there is no time reference deviation during subsequent comparison. When forming the resource occupation trajectory, the continuity of the three types of resource parameters needs to be maintained to avoid incomplete information caused by collecting only one type of parameter.

[0019] After obtaining the time delay distribution curve and the resource occupation trajectory, the two need to be compared point by point. Specifically, the arrival and departure time of the identification pulse at each node is combined with the resource occupation level of the node to analyze. If a node shows obvious transmission lag on the time delay distribution curve, and the occupation of computing resources, storage space and communication bandwidth in the corresponding period is high, it can be determined that the node has the risk of forming potential conflicts in actual operation. When multiple nodes show similar characteristics in the same time period, the path segment can be further locked as a potential conflict intensive area. In this process, each time delay curve and resource trajectory needs to be involved in the comparison completely to avoid the superposition effect of missing single-point problems. In this way, the potential conflict source can be concretized from the abstract "high delay" or "high load" state to the locatable node and path segment.

[0020] After completing the positioning of the potential conflict source, a complete time sequence observation baseline needs to be established based on the time delay distribution curve, resource occupation trajectory and conflict node information. The baseline not only contains the time delay data and resource data of each redundant path, but also clearly marks the nodes and path segments that may cause conflicts. The observation baseline can be used as the initial condition for conflict retrieval in subsequent use, so that the subsequent steps do not need to re-establish observation data when performing loop identification, control right arbitration and rollback. More importantly, the observation baseline can be updated in each industrial internet of things control cycle to ensure that the newly generated baseline is consistent with the current running state, thereby realizing dynamic adaptation.

[0021] The core role of establishing a timing observation baseline for multi-level redundant paths and injecting micro-amplitude identification pulses into all control nodes is to provide a comprehensive, accurate, and traceable operational reference for the entire Industrial IoT control method. By injecting identification pulses one by one under a unified time base, the transmission speed of each redundant path and the response behavior of each node can be accurately reflected, thus forming a complete instruction latency distribution curve. This is combined with the node's computation time, storage space utilization ratio, and communication bandwidth utilization ratio during the pulse passage to obtain a resource utilization trajectory. This not only clearly reveals which nodes are experiencing abnormal lags but also clearly identifies which nodes may become sources of conflict under high load. Compared to traditional methods that rely solely on a single latency metric or local monitoring data, this step, through a dual-dimensional, comprehensive approach, provides a solid data foundation for subsequent loop retrieval and conflict list generation. In other words, this step not only detects path status but also proactively reveals potential risks, ensuring that the entire control method is built on a quantifiable and verifiable data environment from the outset, thereby improving the reliability and effectiveness of subsequent arbitration, circuit breaking, and optimization steps.

[0022] S002: Based on the instruction latency distribution curve and resource occupancy trace, a loop search graph is constructed to calculate the instruction reflux probability and stall duration, identify the closed loop structure and contention nodes, and form a conflict list. After obtaining the instruction latency distribution curve and resource utilization trajectory of the multi-level redundant path, further processing is required to identify the closed loop structure and competing nodes that may form within the path, and ultimately form a conflict list. This step not only integrates and analyzes basic data, but also provides a clear basis for subsequent control arbitration and path selection. The specific implementation method is as follows: Based on the instruction delay distribution curve and resource occupancy trajectory that have been formed, all control nodes and the transmission relationship between them are expressed in the form of a graph. Specifically, each control node is mapped as an independent point in the graph, and two core parameters are marked for the point. One is the average transmission delay of the identification pulse at the node, and the other is the resource occupancy of the node when the identification pulse passes through, including the computing resource consumption time, storage space usage ratio and communication bandwidth occupancy ratio. At the same time, the connection relationship between the nodes is mapped as a directed edge, and the average transmission time of the identification pulse on the path is marked on each directed edge. In this way, a loop retrieval graph covering all redundant paths is obtained. The graph not only shows the topological relationship between the nodes, but also clearly presents the transmission performance of each path and the resource status of each node.

[0023] After the loop retrieval graph is constructed, the probability of instruction reflux and the stagnation time length of each directed edge and node need to be quantitatively analyzed by combining the instruction delay distribution curve and the resource occupation trajectory. The specific implementation manner is as follows: first, the arrival time and the departure time of the identification pulse on different paths are compared. If the round-trip time of a certain path is obviously close to or exceeds the average value of other paths, it indicates that there is a risk of reflux through the path, and the risk is expressed in the form of reflux probability in percentage. Secondly, at the node level, the difference between the residence time of the identification pulse in the node and the benchmark normal transmission time is compared. If the residence time of a certain node is significantly increased when the identification pulse passes through, and the computing resource consumption, storage occupation proportion and bandwidth occupation proportion of the node remain at a high level, it can be determined that the stagnation time length of the node is large. In this way, each path has a clear reflux probability, and each node has a clear stagnation time length. These parameters strictly correspond to the delay distribution curve and the resource trajectory obtained in the previous step, thereby ensuring the continuity and reliability of the data.

[0024] After obtaining the reflux probability and the stagnation time length, all possible paths in the loop retrieval graph need to be traversed one by one to identify the closed loop structure. The specific method is as follows: starting from an arbitrary control node, the direction indicated by the directed edge is tracked step by step, and when returning to the starting node again, it is determined that a closed loop is formed. After identifying the closed loop, the reflux probability of all nodes in the closed loop is calculated by averaging, and the stagnation time length of all nodes is summed up as the risk indicator of the closed loop. If the risk indicator exceeds the pre-set safety threshold, the closed loop is determined as a potential conflict ring. If there is a node in the same closed loop whose reflux probability is significantly higher than the average level and whose stagnation time length is also at the highest level in the ring, the node is marked as a competitive node. In this way, not only the existence of the closed loop is identified, but also the key node in the ring that may cause conflict intensification is clearly pointed out. Unlike the prior art which only identifies the closed loop based on the static topology, the present embodiment combines the delay and resource occupation characteristics into the graph, which can more truly reflect the dynamic conflict risk in operation.

[0025] After identifying all closed loop structures and competing nodes, these information needs to be sorted out to form a complete conflict list. The conflict list needs to contain three parts of content: the first part is the closed loop number and path description, which lists the control node order contained in the loop and the connection relationship between nodes in detail; the second part is the risk index of the closed loop, which clearly marks the average backflow probability and cumulative stall time of the loop to reflect its conflict severity; the third part is the competing node list, which lists the high-risk nodes marked in the loop one by one, and attaches the backflow probability and stall time data of the node. The generation process of the conflict list closely connects with the previous steps, not only uses the time delay distribution curve and resource occupation trajectory formed by the observation baseline, but also combines the traversal results of the loop search graph. The conflict list will be directly applied in the subsequent steps to guide the issuance of control tokens and path arbitration, so as to avoid multiple nodes competing for control at the same time and ensure that the deployment instructions can be smoothly sent to the execution end.

[0026] The step of constructing a loop search graph based on the instruction time delay distribution curve and resource occupation trajectory, and calculating the instruction backflow probability and stall time, so as to identify closed loop structures and competing nodes and form a conflict list, its main role is to transform the potential running risk in the distributed redundant path from abstract data state to visual, quantifiable and operable object. By mapping all control nodes and paths into a directed graph with time delay parameters and resource parameters, not only the path topology relationship can be revealed, but also the actual running state of each node can be clearly reflected. Further, by calculating the backflow probability and stall time, the risk points of backflow or long-time retention of instructions in a specific path can be identified, and these high-risk paths can be combined into closed loop structures through loop search, so as to reveal the complete path segment that may cause instruction deadlock. At the same time, the nodes with the most serious stall or the highest backflow probability are marked in the loop, which can help to lock the competing nodes. The final conflict list presents the loop number, path composition, risk index and competing node information, which provides a clear basis for subsequent token issuance, arbitration and fuse recovery. Therefore, this step not only plays a bridge role to closely connect the basic observation data with the subsequent arbitration mechanism, but also improves the accuracy and reliability of conflict identification, which is different from the traditional method of relying on single delay monitoring, and has significant creativity and engineering application value.

[0027] S003, according to the conflict list, a single-source control token is issued, the token flows along the shortest reliable path in one direction, the control node without holding the token is prohibited to write to the execution end, ensuring unique control and making the control process converge to a single link; After the generation of the conflict list, a unique arbitration method is needed to avoid the concurrent conflicts caused by multiple control nodes in the multi-level redundant path simultaneously issuing instructions to the execution end. This step ensures that only one node has the instruction writing permission at all times during the control process by issuing a single-source control token and specifying that the token flows unidirectionally along the shortest reliable path, thereby achieving the uniqueness of control and the convergence of links. The specific implementation is as follows: After generating the conflict list, a unique token starting node needs to be determined. The selection of the token starting node is based on the data in the conflict list, and the node that is not in the closed conflict ring and has a lower risk index is preferentially selected as the starting node. The risk index is assessed based on the backflow probability and the stagnation time, and reference is also made to the position of the node in the delay distribution curve, the short calculation time, the low storage space occupation ratio, and the low communication bandwidth occupation ratio in the resource occupation trajectory. When the token is generated at the starting node, it is assigned a unique number and a timestamp of the arbitration period, ensuring that there are no two or more tokens existing in parallel in the entire redundant path. This measure ensures the uniqueness of control from the generation link, eliminating the instruction conflicts that may be caused in the generation phase compared with the traditional method of relying on broadcast signals to compete for control rights by multiple nodes.

[0028] After the token is generated, the token flow path needs to be determined. The path selection is based on the path information in the conflict list, and the average delay, cumulative stagnation time, and resource occupation level of each node are compared among multiple candidate paths, and finally a path with the shortest transmission time, the lowest stagnation risk, and the resource occupation ratio of all nodes not exceeding the safety threshold is selected. When the token starts to flow, it is transmitted from the starting node to the next node. During the transmission process, each node must perform a check when it receives the token, and the check content includes whether the token number is consistent with the number generated by the starting node, whether the token timestamp is within the effective range of the current arbitration period, and whether the token state marker is intact. Only after the check passes can the node temporarily obtain the permission to write the deployment instructions to the execution end, and immediately pass the token to the next node after the writing operation is completed. In this way, the token always flows along a unique and sequentially fixed path in one direction, and there is no bifurcation, backflow, or looping phenomenon in the network, thereby ensuring the unidirectionality and reliability of the flow process.

[0029] In the process of token circulation, the nodes without the token need to be strictly limited to avoid the situation that multiple nodes write to the execution end at the same time. The specific measures are: setting a permission verification mechanism in the writing link of the execution end, only the nodes with the token can submit the deployment instructions to the execution end after passing the verification. The nodes without the token can only keep the instructions locally even if they detect that the parameters need to be adjusted, and they do not have the ability to write to the execution end. This mechanism ensures that even in the case of multiple nodes simultaneously finding abnormal working conditions, the execution end will only receive unique deployment instructions from the nodes with the token, fundamentally avoiding conflicts caused by concurrent writing. In order to prevent the token from being stagnant for a long time due to abnormal processing of individual nodes, each node must complete writing and delivery within a specified time limit when holding the token. If it is not completed within the time limit, the token will be automatically determined as invalid, and then the fuse mechanism is started to prepare for the takeover of the standby path, avoiding the blocking of the entire control link due to the stagnation of a single node.

[0030] After the token completes a complete circulation and returns to the starting node, the state of the circulation process needs to be confirmed. The starting node will check whether the transmission delay of all nodes in this circulation is within the normal range, whether all execution end writing is successfully completed, and whether the token has appeared timeout stagnation in the circulation path. If the inspection result shows that all operations are normal, it is determined that this control process has converged to a single link, and the next arbitration period can be entered. If it is found that a node is abnormal, for example, the token transmission is timeout, the instruction is not written, or the token is lost, the node is marked as a high-risk node, and its risk level is increased when updating the conflict list, so as to avoid the node as much as possible in the next path selection. Through this closed-loop confirmation mechanism, not only the convergence of the control right on a single link is guaranteed, but also the dynamic linkage with the conflict list in the previous stage is realized, so that the token issuance and path arbitration form a continuous optimization cycle.

[0031] The step of "issuing a single-source control right token according to the conflict list, unidirectionally flowing the token along the shortest reliable path, and prohibiting the write execution end of the control node that does not hold the token to ensure unique control right and make the control process converge to a single link" has the core function of solving the problems of deployment instruction conflict and deadlock caused by the simultaneous write capability of multiple control nodes in the existing multi-level redundant path. By setting a unique control right token, it is clear that only one node has the right to deliver the deployment instruction to the execution end at any time, which fundamentally eliminates the possibility of concurrent writing. The unidirectional flow mechanism of the token along the shortest reliable path makes the control right pass between nodes in a predetermined order, and does not appear backflow or loop, thereby avoiding endless competition in the closed loop. For nodes that do not hold the token, their write permission is strictly prohibited, and even if the node detects an abnormal working condition, it can only temporarily store the deployment intention and cannot directly write to the execution end, which ensures that the execution end always receives only valid instructions. At the same time, the periodic flow of the token and the confirmation mechanism back to the starting node ensure that the entire control process can continuously converge and verify the stability of the execution link. In summary, this step not only establishes a unique arbitration method for cross-node collaborative control in the industrial Internet of Things environment, but also compresses the control process to a controlled link, significantly improving the order, stability and safety of instruction issuance, and providing a reliable prerequisite for subsequent fault fusing and standby path takeover.

[0032] S004, monitoring the token stall state during the token flow process, dynamically comparing the instruction time delay distribution curve with the token position, if the token stall is detected to exceed the threshold, the current loop is revoked, rolled back to the last stable snapshot and the standby path takeover is started, to avoid control interruption caused by path failure; During the process of token completion and unidirectional flow along the shortest reliable path, the transmission state of the token needs to be continuously monitored to prevent the token from being stuck for a long time due to node failure, resource depletion or link delay anomaly, so that the control right is stuck in a certain place and cannot be passed on. Therefore, by comparing the current position of the token with the pre-established instruction time delay distribution curve, it is determined whether there is a timeout stall phenomenon. Once it is determined that the stall exceeds the threshold, the operation of revoking the current loop, rolling back to the last stable snapshot and starting the standby path takeover is immediately performed to ensure the continuity of the control process. The specific implementation is as follows: When the token enters a certain node, the arrival time needs to be recorded, and the stay duration is continuously measured while the token continues to stay in the node. When the token is passed to the next node, the departure time is recorded again, so as to obtain the actual stay duration of the node. At the same time, the data is compared with the instruction delay distribution curve obtained when the timing observation baseline is established in real time. For example, in the instruction delay distribution curve, the normal transmission duration of a certain node is 6-9 milliseconds, and the monitored actual stay duration reaches 20 milliseconds, so it can be preliminarily judged that the node has abnormal stagnation phenomenon. Through the comparison between the actual measurement and the baseline data, the unreasonable delay of the token in the path can be accurately located, so as to ensure the scientificity and objectivity of the monitoring and judgment.

[0033] After the preliminary determination of stagnation, it is necessary to comprehensively confirm whether the stagnation exceeds the threshold value in combination with the resource occupation trajectory. The threshold value is not only determined by the maximum normal value of the node in the delay distribution curve, but also considers the resource occupation of the node in the corresponding time period. If the calculation time of the node continuously rises, the storage space usage ratio is close to full load, the communication bandwidth occupation ratio is greatly increased, and the token stay duration exceeds the preset threshold value (for example, more than twice the normal maximum value) in the same period, it can be confirmed that the node has entered the blocking state. At this time, it must be immediately determined that the token cannot continue to flow effectively, otherwise it will cause the stagnation of the whole path. Unlike the traditional method of judging by fixed timeout, the actual occupation data of the calculation resource, the storage space and the communication bandwidth are introduced in the stagnation judgment in this step, so that the judgment result is more accurate, and the misjudgment or omission is avoided.

[0034] After determining that the token stagnation exceeds the threshold value, the operation of revoking the current loop and rolling back to the last stable snapshot needs to be performed immediately. The specific method is as follows: first, stop the token from continuing to pass on the path, revoke the write permission of all related nodes, and avoid the abnormal node from continuing to occupy the control right. Then, the execution end and the node state are restored to the stable snapshot saved at the last successful execution. The data saved in the stable snapshot includes the key parameter value of the execution end, the starting number and timestamp of the token in the previous round of successful flow, and the node order of the last stable path. By restoring these information, it can be ensured that the execution end still remains in the last reliable state when the current path appears abnormal, and will not enter the out-of-control operation because of the current token stagnation. This mechanism effectively provides the rollback protection, so that the control process has the recoverable ability under abnormal conditions.

[0035] After the rollback is completed, a takeover operation of the backup path needs to be started immediately. The selection of the backup path is based on the priority ranking in the previously generated conflict list and loop search map. The specific method is to select a path with shorter average latency, lower cumulative stall time, and resource occupation level within a safe range in the backup path as the new delivery path. Then a new token is regenerated at the starting node, and a new number and the timestamp of the current arbitration cycle are assigned. The new token is transmitted to the execution end along the backup path, and each node needs to verify the number and timestamp when receiving, and complete instruction writing and forwarding during the token holding period. Through the takeover of the backup path, the abnormality of the original path does not cause control interruption, but enables the instructions to continue to be delivered to the execution end, ensuring the continuity and stability of the deployment process. At the same time, the abnormal data of the original path is recorded and updated to the conflict list, and the risk level of the corresponding node is improved, so as to avoid the node in the subsequent path selection.

[0036] The step of "monitoring the token stall state in the token flow process, dynamically comparing the instruction latency distribution curve with the token position, and if the token stall exceeds the threshold, revoking the current loop, rolling back to the last stable snapshot and starting the backup path takeover" has the core role of providing a reliable abnormality detection and recovery mechanism for the entire industrial Internet of Things control process. Due to the inevitable existence of node resource overload, link delay surge or hardware failure in the multi-level redundant path, the token is prone to stall or even be stuck during transmission. Without effective monitoring and processing, the control right may be long-term detained in a node, causing the execution end to lose the deployment instructions, and thus the process link to be out of control. This step can accurately determine whether the token exceeds the normal threshold by comparing the actual residence time of the token in the node with the pre-established instruction latency distribution curve, and confirm whether it enters the blocking state in combination with the resource occupation situation. Once the stall is confirmed, the control right of the path will be revoked immediately to avoid abnormality spreading; at the same time, by rolling back to the last stable snapshot, the execution end is quickly restored to the latest safe state, ensuring that the process parameters are not affected by errors; then the backup path takeover is started again, and the new token is successfully delivered to the execution end, ensuring that the deployment process does not interrupt. Thus, this step realizes a complete closed loop from abnormality discovery to safe rollback to path switching, avoids the global shutdown caused by path failure in the prior art, and significantly improves the robustness and safety of the system.

[0037] S005, after the backup path takeover, the deployment instruction is implemented for delay shaping, and real-time residual error information is used to dynamically adjust the sending beat, load window and queuing weight, to weaken the self-excited oscillation effect caused by path switching, and ensure that the deployment instruction is stably landed to the execution end; After the standby path takeover is completed, although the new path ensures that the token and the deployment instruction can continue to be transmitted, due to the differences between the standby path and the original path in transmission delay, node load and resource usage, if not adjusted, self-excited oscillation phenomenon caused by sudden delay fluctuation and resource congestion is likely to occur, so that the deployment instruction appears to be congested in a short time or instantaneously impacts the execution end, thereby affecting the stability of the execution end. Therefore, after the standby path takeover, delay shaping needs to be implemented, by using real-time residual information, gradually adjusting the sending tempo, load window and queuing weight, to ensure that the deployment instruction can be stably landed to the execution end. The specific implementation is as follows: At the first time after the standby path takeover is completed, residual information in the transmission process before and after the takeover needs to be obtained. The residual information is the difference value obtained by comparing the actual delay of each node in the standby path with the reference delay recorded in the original path stable snapshot, and is formed by combining the current calculation time consumption, storage space usage ratio and communication bandwidth occupation ratio of the node. Specifically, when the transmission delay of a node in the standby path is higher than the snapshot reference value by more than 10 milliseconds, and the storage space occupation of the node is more than 80% and the communication bandwidth occupation is more than 70%, it can be determined that the node has obvious residual. By collecting and comparing the same data for all nodes, a complete residual table containing node number, residual delay, calculation time consumption, storage occupation ratio and bandwidth occupation ratio is formed. This residual table provides a quantitative basis for subsequent adjustment, which is different from the way of simply relying on experience to set compensation time in the prior art.

[0038] After obtaining the residual table, the sending tempo of the deployment instruction needs to be dynamically adjusted according to the residual information. The sending tempo refers to the time interval between two consecutive instructions. In the standby path, if the residual table shows that the delay difference of some nodes is much higher than the snapshot reference value, the sending tempo interval needs to be increased, for example, from the original 5 milliseconds to 10 milliseconds, to avoid the accumulation of multiple instructions at the node. If the residual table shows that the overall transmission speed of the node is better than that of the original path, the sending tempo can be shortened, for example, from the original 10 milliseconds to 7 milliseconds, thereby improving the transmission efficiency of the instruction. Each adjustment of the tempo is directly related to the residual value of the node, ensuring that the adjustment action is consistent with the actual running situation, rather than a fixed static setting. This can effectively reduce the fluctuations caused by the differences in the characteristics of the standby path.

[0039] While adjusting the sending tempo, the load window also needs to be dynamically controlled. The load window refers to the number of allocation instructions allowed to enter the standby path within a certain period of time. If the storage occupancy of a node is higher than 85% and the communication bandwidth occupancy is higher than 75% as shown in the residual table, the load window should be reduced, for example, from allowing 10 instructions to be transmitted simultaneously to 6 instructions, to reduce the instantaneous pressure of the node. If the storage occupancy of the node is lower than 50% and the bandwidth occupancy is lower than 40%, the load window can be expanded, for example, from 6 instructions to 12 instructions, to improve the overall transmission efficiency. Through the adjustment of the load window, it can be ensured that the standby path will not cause new congestion due to instantaneous instruction overload in the early takeover stage, and the balanced transmission of instructions between nodes can be ensured.

[0040] On the basis of the adjustment of the sending tempo and the load window, the queuing weight needs to be further optimized to ensure that the transmission order of different instructions in the standby path is reasonable and stable. The specific method is as follows: when multiple instructions are queued for transmission at the same time, different priorities are assigned to the instructions according to the delay difference in the residual table. For example, for instructions with a residual delay of more than 15 milliseconds, a higher queuing weight is given to make them have priority to pass through the node and avoid the residual from continuing to expand. For instructions with a residual delay of less than 5 milliseconds, a lower queuing weight is assigned so that they can be transmitted later, thereby freeing up bandwidth and resources for high-residual instructions. Through the adjustment of the priority, the transmission order of different instructions in the standby path is more reasonable, and ultimately all instructions can smoothly arrive at the execution end. When the allocation instructions are stably landed at the execution end, the actual residual information also needs to be recorded as reference data for the next round of path optimization.

[0041] The step of "after the standby path takes over, time delay shaping is implemented for the deployment instruction, real-time residual information is used to dynamically adjust the sending tempo, load window and queuing weight, the self-excited oscillation effect caused by path switching is weakened, and stable landing of the deployment instruction to the execution end is ensured" has the core role of ensuring the smooth transmission of the deployment instruction after the standby path switching and the continuous and stable operation of the execution end. When the original path is revoked due to token stagnation or resource overload, the standby path can take over the task, but there are usually significant differences in transmission delay, resource load and node performance between the original path and the standby path. If no adaptation is made, it is easy to cause excessive accumulation or burst congestion of the instruction in the new path, thereby forming a self-excited oscillation, causing the execution end to have instruction delay, burst impact or even failure. This step compares the actual operation of the standby path with the snapshot benchmark by calculating the residual data of the node in real time, and dynamically adjusts the sending tempo, load window and queuing weight based on the difference, gradually eliminating the influence of burst fluctuations. The adjustment of the sending tempo ensures reasonable instruction interval, the adjustment of the load window prevents node overload, and the optimization of the queuing weight ensures that high residual instructions are transmitted preferentially. This series of actions forms an adaptive shaping mechanism after path switching, so that the deployment instruction can enter the execution end in a balanced and orderly manner. Through this step, the limitations brought by fixed delay or static queue compensation in the traditional technology are avoided, real-time balance and vibration suppression under dynamic conditions are realized, and the continuity and reliability of the execution end in the entire industrial Internet of Things control process are ensured.

[0042] S006, after the stable execution of the deployment instruction, the execution residual, the fuse record and the takeover times are comprehensively analyzed, the topology weight, the token priority and the fuse threshold are dynamically updated, a self-evolution closed loop mechanism is constructed, and the risk of deadlock is continuously reduced and the whole process logic closed loop is realized; After the deployment instruction is stabilized and landed to the execution end after being taken over by the standby path and time delay shaping, it cannot stop at the completion of single execution, but needs to be systematically analyzed and dynamically corrected for the entire execution process, so as to continuously optimize the network operation state in the subsequent cycle. Therefore, the execution residual, the fuse record and the takeover times need to be comprehensively analyzed, and the topology weight, the token priority and the fuse threshold need to be updated accordingly, forming a closed loop mechanism that can continuously evolve. The specific implementation is as follows: After the stable execution of the deployment instructions, the running data needs to be collected and sorted completely. The running data includes three categories: first, the execution residual error, that is, the difference between the actual time of each deployment instruction from the starting node to the execution end and the reference time in the previous stable snapshot; second, the fuse record, that is, the number of times of triggering the fuse operation in the process of token flow, the node number of fuse occurrence, the calculation resource consumption, the storage space usage ratio and the communication bandwidth occupation ratio at the time of triggering; and third, the takeover times, that is, the number of times of enabling the standby path in the current cycle and the duration of each takeover. These data need to be corresponded with the node number one by one and saved in time sequence. Unlike the prior art which only records part of the error data when a serious exception occurs, the present step records completely regardless of the success or failure of the execution, thereby ensuring the comprehensiveness and continuity of the data.

[0043] After the data collection is completed, the three categories of data need to be analyzed one by one. For the execution residual error, the average residual error and the maximum residual error of each node need to be counted to determine which nodes still have sustained delay after path switching. For the fuse record, the frequency of triggering the fuse needs to be counted, and if a node triggers the fuse more than five times in ten executions, it is determined that the node is in a high-risk state. For the takeover times, the enabling ratio of the standby path and the main path needs to be compared, and if the takeover frequency of a standby path is significantly higher than that of other paths, for example, more than six times in ten cycles, it indicates that the main path has insufficient stability, and the priority thereof needs to be reduced in the next path selection. In the analysis process, the execution residual error also needs to be compared with the residual error table in the previous time delay shaping stage to verify whether the effect of the shaping measure is significant, and if the residual error is still greater than 15 milliseconds and lasts for multiple rounds, it indicates that the path selection strategy needs to be further adjusted.

[0044] After the analysis is completed, the topology weight, token priority and fuse threshold need to be adjusted according to the results. For the topology weight, when a node has high residual error and frequent fuse in multiple cycles, the weight of the node needs to be reduced, for example, from a weight value of 5 to 2, thereby reducing the probability of being included in the main path in future path selection. Conversely, for a node with a residual error of less than 5 milliseconds and almost no fuse triggering in multiple cycles, the weight thereof needs to be increased, for example, from a weight value of 3 to 6, to increase the possibility of entering the main path. For the token priority, if the nodes in a path perform stably in multiple executions, the priority of the token on the path needs to be increased to make the token pass through the reliable nodes faster in path selection and flow. For the fuse threshold, when some nodes frequently trigger the fuse due to slight fluctuations, the fuse threshold thereof needs to be increased, for example, from a stall time of 10 milliseconds to 15 milliseconds, to avoid excessive fuse; and for the key nodes that cause whole link blockage once stalled, the fuse threshold thereof needs to be reduced, for example, from 15 milliseconds to 8 milliseconds, to enable the nodes to be isolated faster.

[0045] After the adjustment of the topology weight, token priority and fuse threshold, these updates need to be applied to the next round of token distribution and path selection, thus forming a self-evolution closed-loop mechanism. The specific method is: at the beginning of the new round, the shortest reliable path is recalculated using the updated topology weight, and the node with higher weight is preferentially selected to build the path; in the token circulation, the token passing order is arranged according to the new priority order, so as to avoid high-risk nodes as much as possible; in the stagnation monitoring, the updated fuse threshold is used as the judgment standard, so that the fuse triggering is more matched with the actual running situation. In this way, the data of each round of running will directly affect the next round of running, so that the control process is gradually optimized in the continuous cycle, the weight distribution of the node tends to be reasonable, the token priority is more in line with the actual situation, and the fuse threshold is more in line with the running environment, thereby continuously reducing the risk of deadlock. Different from the way of fixed parameter setting in the prior art, the present step updates the parameters dynamically, so that the entire control method has the ability of self-learning and self-correction, ensuring the stability and high reliability of the industrial Internet of Things control in long-term operation.

[0046] The step of "after the stable execution of the deployment instruction, comprehensively analyzing the execution residual error, fuse record and takeover times, dynamically updating the topology weight, token priority and fuse threshold, and building a self-evolution closed-loop mechanism" is to make the industrial Internet of Things control process have the ability of continuous optimization and self-evolution, thereby effectively reducing the risk of deadlock in long-term operation. This step collects complete running data after each round of instruction execution, including the actual residual error of the instruction at each node, the number and conditions of fuse triggering, and the frequency and duration of standby path takeover, and compares these data with the snapshot benchmark and delay shaping results, so as to comprehensively reflect the real running state of the node and the path. Based on these analysis results, the topology weight can be dynamically adjusted, so that the nodes with high residual error and frequent fuse triggering are weakened in future path selection, and the stable and reliable nodes are strengthened; the token priority can be optimized, so that the token is preferentially transmitted in the reliable path, thereby reducing the overall delay; the fuse threshold can be corrected, so as to avoid excessive triggering to cause efficiency decline, and to isolate in time when the key node is abnormal, thereby ensuring the operation safety. Through this dynamic update, each execution result will directly affect the next round of control strategy, forming a data-driven closed loop, so that the control process is no longer dependent on static configuration, but continuously adapts to complex environment in continuous operation, and finally realizes more stable, efficient and safe industrial Internet of Things control.

[0047] The application firstly realizes accurate revelation of potential conflict sources by establishing a timing observation baseline and generating an instruction delay distribution curve and a resource occupation trajectory at all control nodes; then, with the help of a loop search diagram, a closed loop structure and a competing node are identified to form a conflict list, which provides a basis for subsequent arbitration; then, through the issuance of a single-source control token, the deployment process converges to a single reliable link, eliminating the risk of multiple nodes competing for control; in the token circulation, its stagnation state is monitored in real time, and snapshot rollback and standby path takeover are combined to effectively avoid global control interruption caused by path congestion; after the standby path takeover, delay shaping is implemented to dynamically adjust the sending tempo, load window and queuing weight using residual information, eliminating the oscillation effect caused by path switching and ensuring that the deployment instruction can be smoothly landed on the execution end; finally, through comprehensive analysis of the execution residual, the fuse record and the takeover frequency, the topology weight, the token priority and the fuse threshold are dynamically updated to form a closed loop mechanism that can continuously evolve. Thus, the application not only guarantees real-time stable deployment of multiple types of working parameters, but also continuously reduces the risk of deadlock in long-term operation, realizing high reliability, high security and high robustness of industrial Internet of Things control.

[0048] Certain exemplary embodiments of the present application have been described above by way of illustration, and it is to be understood that those skilled in the art can make various modifications to the described embodiments without departing from the spirit and scope of the present application. Therefore, the above drawings and descriptions are illustrative in nature, and should not be construed as limiting the scope of protection of the claims of the present application.

Claims

1. An industrial Internet of Things control method for adjusting multiple types of working parameters, characterized in that: The following steps are involved: S001: Establish a timing observation baseline for multi-level redundant paths, inject micro-amplitude identification pulses into all control nodes, generate instruction delay distribution curves and resource usage traces, and reveal potential conflict sources and provide initial conditions. S002: Based on the instruction latency distribution curve and resource occupancy trace, a loop search graph is constructed to calculate the instruction reflux probability and stall duration, identify the closed loop structure and competing nodes, and form a conflict list. S003: A single-source control token is issued based on the conflict list. The token flows in one direction along the shortest reliable path. Control nodes that do not hold the token are prohibited from writing to the execution end, ensuring sole control and converging the control process to a single link. S004: Monitor token stagnation during token transfer and dynamically compare the instruction latency distribution curve with the token position. If token stagnation exceeds a threshold, the current loop is canceled, the system rolls back to the last stable snapshot, and a backup path is initiated to avoid control interruption due to path failure. S005: After the backup path takes over, latency shaping is implemented for the dispatch instructions. The real-time residual information is used to dynamically adjust the sending beat, load window, and queue weight. This reduces the self-oscillation effect caused by path switching and ensures that the dispatch instructions are stably delivered to the execution end. S006, after the deployment instruction is stably executed, comprehensively analyze the execution residuals, circuit breaker records and takeover times, dynamically update the topology weight, token priority and circuit breaker threshold, and build a self-evolving closed-loop mechanism to achieve continuous reduction of deadlock risks and full-process logical closure.

2. The industrial Internet of Things control method for adjusting multiple types of working parameters according to claim 1 is characterized in that: Step S001 includes: In the process of initializing the multi-level redundant paths, each redundant path is divided into a continuous link consisting of a head-end input node, an intermediate transmission node, and a terminal execution node; Inject a small-amplitude identification pulse that does not affect process execution into all control nodes under a unified time base, and assign a unique timestamp before the identification pulse is sent; When the recognition pulse passes through each control node, the arrival time and departure time are recorded, and the node's computing time, storage space occupancy ratio, and communication bandwidth occupancy ratio during the pulse passage are also recorded synchronously; After the identification pulse transmission is completed, the recorded data of each node is summarized and sorted, and the transmission time difference of the identification pulse between adjacent nodes is calculated to form a delay distribution curve. At the same time, the calculation time, storage space occupancy ratio and communication bandwidth occupancy ratio are combined to form a resource occupancy trajectory, thereby constructing a time series observation baseline that can truly reflect the node behavior and path operation status.

3. The industrial Internet of Things control method for adjusting multiple types of working parameters according to claim 1 is characterized in that: Step S002 includes: Based on the obtained instruction latency distribution curve and resource occupancy trajectory, all control nodes are mapped as independent points, and the average transmission latency and resource occupancy of each point are marked. The connection relationship between nodes is mapped as directed edges and the average transmission time is marked, thus forming a loop search graph covering all redundant paths. Combining the instruction latency distribution curve and resource usage trajectory, we analyze each directed edge and node, calculate the instruction reflux probability and stall duration, and obtain path-level and node-level risk parameters. Traverse all paths in the loop search graph one by one, identify closed loop structures, and calculate the average backflow probability and cumulative stagnation time. Mark closed loops with risk indicators exceeding the threshold as potential conflict loops, and mark the nodes with the highest backflow probability and stagnation time within the loop as competing nodes. The identification results are sorted out to form a conflict list including closed loop number and path description, closed loop risk indicator and competing node list to provide a basis for the subsequent issuance of control tokens.

4. The industrial Internet of Things control method for adjusting multiple types of working parameters according to claim 1 is characterized in that: Step S003 includes: After generating the conflict list, based on the backflow probability, stagnation duration, and resource usage level in the conflict list, the node that is not in the closed conflict loop and has the lowest risk index is selected as the token starting node. When generating the token, a unique number and arbitration cycle timestamp are assigned to ensure uniqueness. When determining the token transfer path, the average delay, cumulative stall time, and resource occupancy ratio of each node are compared based on the path information in the conflict list. The path with the shortest transmission time and the lowest risk is selected, and the token is required to flow in a one-way sequence along this path. During the token transfer process, the node receiving the token must verify the number, timestamp, and status mark. Only after passing the verification can it obtain the permission to write to the execution end. After the writing is completed, the token must be passed to the next node within the specified time limit. Nodes that do not hold the token are prohibited from writing deployment instructions to the execution end. After the token completes a round of circulation and returns to the starting node, the transmission delay, writing results and stagnation are confirmed. If any abnormality is found, the corresponding node is marked as a high-risk node and the conflict list is updated to ensure that the control process converges to a single link and forms a dynamically optimized closed loop.

5. The industrial Internet of Things control method for adjusting multiple types of working parameters according to claim 1 is characterized in that: Step S004 includes: When the token enters the control node, the arrival time and residence time are recorded and compared with the pre-established instruction delay distribution curve in real time. When the actual residence time exceeds the normal range, it is preliminarily determined to be abnormal; The anomaly is confirmed by combining resource usage tracing. When the computing time continues to increase, the storage space usage ratio is close to full load, the communication bandwidth usage ratio increases significantly, and the stay time exceeds the preset threshold, the node is determined to have entered a blocked state. After confirming that the token stagnation exceeds the threshold, the current loop is immediately canceled, all related nodes are prohibited from continuing to write, and the execution end and node status are rolled back to the stable snapshot of the last successful execution, restoring the key parameter values, token number and path order; After the rollback is completed, an alternative path with short average delay, low stagnation time and safe resource occupancy is selected as the new path based on the conflict list and loop retrieval diagram. A new token is regenerated at the starting node and passed to the execution end along the alternative path to ensure the continuity and stability of the deployment process, and the abnormal node information is updated to the conflict list to form a dynamic closed loop.

6. The industrial Internet of Things control method for adjusting multiple types of working parameters according to claim 1 is characterized in that: Step S005 includes: After the backup path takes over, the residual information of the transmission process before and after the takeover is obtained, and a residual table containing the node number, residual delay, calculation time, storage occupancy ratio and bandwidth occupancy ratio is formed; Dynamically adjust the sending beat of the deployment instruction based on the residual table. When the node delay difference exceeds the snapshot reference value by 10 milliseconds, the sending beat interval is extended to twice the original setting value. When the overall node transmission speed is faster than the snapshot reference value by more than 5 milliseconds, the sending beat interval is shortened to 70% of the original setting value. While adjusting the sending rhythm, the load window is dynamically controlled. When the node storage occupancy rate exceeds 85% and the communication bandwidth occupancy rate exceeds 75%, the load window is reduced to half of the original setting value. When the node storage occupancy rate is less than 50% and the communication bandwidth occupancy rate is less than 40%, the load window is expanded to twice the original setting value. Based on the adjustment of the sending beat and load window, the queuing weight is optimized. When the residual delay exceeds 15 milliseconds, the instruction is assigned a priority value of 10 and transmitted in advance in the queuing order. When the residual delay is less than 5 milliseconds, the instruction is assigned a priority value of 2 and transmitted later in the queuing order. This ensures that the dispatched instructions are finally smoothly delivered to the execution end, and the residual information is recorded after delivery to update the basis for subsequent path optimization.

7. The industrial Internet of Things control method for adjusting multiple types of working parameters according to claim 1 is characterized in that: Step S006 includes: After the deployment instruction is stably executed, the running data is collected and three types of data are generated: execution residual, circuit breaking record, and takeover count. The execution residual is the difference between the actual transfer time and the stable snapshot reference time. The circuit breaking record is the number of times the circuit breaking is triggered, the node number, and the corresponding computing resource consumption, storage space usage ratio, and communication bandwidth usage ratio. The takeover count is the number of times the backup path is enabled in the current cycle and the duration of the activation. After data collection is complete, we analyze execution residuals, circuit breaker records, and takeover times. If a node's residual exceeds 15 milliseconds for multiple rounds, it is considered a high-risk node. If a node's circuit breaker is triggered more than five times in ten cycles, its reliability is reduced. If a backup path takes over more than six times in ten cycles, its priority is lowered. After completing data analysis, the topology weight, token priority, and circuit breaker threshold are dynamically adjusted based on the results. When a node frequently circuit breaks, its topology weight is reduced; when a node performs stably, its topology weight is increased; when the path is reliable, its token priority is increased; when a node frequently falsely triggers circuit breakers, the circuit breaker threshold is increased; and when a critical node stalls, blocking the entire link, the circuit breaker threshold is lowered. After the parameter update is completed, the updated topology weight is used to recalculate the shortest reliable path, the updated token priority is used to determine the flow order, and the updated circuit breaker threshold is used for stagnation monitoring, thereby achieving path optimization, risk avoidance and control closed loop in the new round of operation, ensuring that the control process has the ability to self-evolve and continuously reduce the risk of deadlock.

Images