Industrial internet of things real-time scheduling method and device based on two-stage multi-population genetic algorithm

Abstract

The invention relates to an industrial Internet of Things scheduling technology, in particular to an industrial Internet of Things real-time scheduling method and device based on a two-stage multi-population genetic algorithm. The real-time scheduling device can efficiently process data streams with different priorities in an industrial network and accurately adapt to a complex and changeable network environment. The system comprises a data stream processing subunit used for collecting and processing various data streams in an industrial Internet of Things network. And the criticality classification subunit is used for dividing the data streams processed by the data stream processing subunit into two types, namely high-criticality data streams and low-criticality data streams. And the two-stage multi-population genetic algorithm scheduling subunit is used for improving the scheduling effect through a two-stage optimization process. And the path selection subunit is used for selecting the path with the highest reliability for the high-criticality data flow and selecting the path with the lowest resource consumption for the low-criticality data flow. And the transmission optimization subunit is used for dynamically adjusting transmission parameters according to the actual network state.

Description

Technical Field

[0001] The present invention relates to industrial Internet of Things scheduling technology, and in particular to an industrial Internet of Things real-time scheduling method and device based on a two-stage multi-population genetic algorithm. Background Art

[0002] With the rapid advancement of industrial automation, industrial Internet of Things technology has become increasingly important in the field of intelligent manufacturing. The Industrial Internet of Things creates a complex heterogeneous environment by connecting various types of industrial equipment, realizing data collection, equipment control, and communication between devices. However, with the continuous increase of equipment information and the surge in data volume, the existing data transmission scheduling methods face challenges in dealing with the delay sensitivity and real-time performance of data. In this case, the real-time transmission of equipment emergency information is limited, resulting in a decrease in the real-time performance of delay-sensitive data and an increase in equipment failure and damage rates. In this complex industrial field environment, the existing communication scheduling strategies often cannot effectively guarantee communication efficiency. Therefore, improving the efficiency of data classification services and ensuring the reliable real-time transmission of delay-sensitive data have become urgent issues to be solved in the field of industrial Internet of Things communications.

[0003] Generally, the transmission of industrial equipment data must follow industrial data exchange standards to support communication between devices and vertical data transmission from field devices to the advanced control layer. There are many challenges in the current data scheduling methods, especially in wireless network environments. The existing mixed transmission strategy of real-time and non-real-time data is mainly applicable to wired network environments. When applied to wireless networks, its functional compatibility and data recognition capabilities are insufficient, resulting in increased network packet loss rate and latency, making it difficult to meet the real-time requirements of data transmission of various devices in industrial wireless networks. Summary of the invention

[0004] The present invention aims at the defects of the prior art and provides an industrial Internet of Things real-time scheduling method and device based on a two-stage multi-population genetic algorithm.

[0005] To achieve the above purpose, the present invention adopts the following technical solution, a real-time scheduling device for industrial Internet of Things based on a two-stage multi-population genetic algorithm, comprising:

[0006] The data stream processing subunit is used to collect and process various data streams in the industrial Internet of Things network, and provide basic data for the classification of the criticality classification subunit and the scheduling of the two-stage multi-population genetic algorithm scheduling subunit.

[0007] The criticality classification subunit is used to classify the data streams processed by the data stream processing subunit into two categories: high criticality data streams (critical data streams) and low criticality data streams (non-critical data streams), to ensure that different types of data streams are reasonably distinguished and prioritized during the scheduling process.

[0008] The two-stage multi-population genetic algorithm scheduling subunit is used to receive and process high-criticality data streams and low-criticality data streams, and improve the scheduling effect through a two-stage optimization process.

[0009] In the first stage, a multi-population genetic algorithm is used to explore different data flow scheduling schemes in parallel. Each population evolves independently, and a preliminary routing allocation scheme and data flow scheduling sequence C = {1, 0, 0, 1..., 1; f3, f7, f2, f3, f7 ... l ..., f8}, where the first half is used to indicate the routing selection of the data flow (0 indicates source routing, 1 indicates graph routing), and the second half is used to indicate the scheduling order of the data flow; the solution space is extensively searched to ensure that the transmission requirements of the data flow are met globally.

[0010] The second stage is based on the candidate solutions screened out in the first stage. Through information exchange and co-evolution among populations, the comfort function is used to select the scheduling plan with the highest comfort value, and finally the optimal scheduling plan that adapts to the dynamic changes of the network is output.

[0011] Throughout the entire process, the comfort function plays a core role in evaluating the pros and cons of scheduling plans, comprehensively considering key indicators such as transmission success rate and the matching degree of routing selection and data flow criticality, and then guiding the algorithm to select excellent plans and eliminate poor plans, ensuring that the scheduling results are continuously optimized and gradually approach the global optimum.

[0012] The path selection subunit is used to select the most reliable path for high-criticality data flows and the path with the least resource consumption for low-criticality data flows based on the scheduling results of the two-stage multi-population genetic algorithm scheduling subunit and the current network topology information, ensuring that data can be transmitted efficiently in complex industrial environments and reducing delays and packet loss rates.

[0013] The transmission optimization subunit monitors the transmission process of the data stream in real time and dynamically adjusts the transmission parameters according to the actual network status.

[0014] Furthermore, the various data streams include periodic data streams, event-driven data streams, control instruction data streams, real-time control data streams, regular status data streams, and multimedia data streams.

[0015] Furthermore, the data stream processing subunit also includes a preprocessing module for performing data cleaning, encoding and encryption on the collected data stream to ensure the security and integrity of the data.

[0016] Furthermore, the criticality classification subunit is based on the real-time R of the data stream. i , Task Importance i and transmission demand T i To calculate the data flow Fi Criticality C i , based on the formula:

[0017] C i =αR i +βH i +γT i

[0018] Among them, α, β, and γ are weight coefficients, corresponding to the real-time nature of the data flow, the importance of the task, and the transmission requirements, respectively;

[0019] Real-time i The score is: The real-time score is used to measure the processing time requirements of the data flow; the score range is 1 to 3 points, and the specific definitions are as follows:

[0020] 1: Able to tolerate delays of 10ms or more;

[0021] 2: Able to tolerate a delay of 5ms to 10ms;

[0022] 3: The processing must be completed within 5ms, and there is almost no tolerance for delay;

[0023] Task Importance i The scores are: The task importance score is used to evaluate the criticality of the data flow in the business process; the higher the bandwidth requirement, the greater the importance of the task; the score range is 1 to 3 points, and the specific definitions are as follows:

[0024] 1: Low bandwidth requirement (less than 20Mbps);

[0025] 2: Medium bandwidth requirement (50Mbps to 100Mbps);

[0026] 3: High bandwidth requirements (greater than 100Mbps);

[0027] Transmission requirement T i The score is: The transmission requirement score is used to measure the quality requirements of the data flow during the transmission process; the score range is 1 to 3 points, and the specific definitions are as follows:

[0028] 1: 0.05% data loss or error is allowed;

[0029] 2: 0.01% data loss or error is allowed;

[0030] 3: 0.001% data loss or error is allowed;

[0031] α evaluates the real-time nature of data flows by measuring performance indicators (including end-to-end delay and strict deadline requirements of data flows), β determines the importance of data flows based on their bandwidth requirements, and γ evaluates transmission requirements by analyzing the transmission reliability of data flows. The weight coefficients are dynamically adjusted based on the priority of business requirements:

[0032] Real-time performance: When the end-to-end delay of the data stream is required to be in the range of 1ms to 5ms, the weight coefficient α is set to 0.6 to 0.7;

[0033] When the end-to-end delay requirement is in the range of 10ms to 50ms, the weight coefficient α is set to 0.3 to 0.4;

[0034] Task importance: When the bandwidth requirement is greater than 100 Mbps, the weight coefficient β is set to 0.6 to 0.7;

[0035] When the bandwidth demand is greater than 10Mbps to 50Mbps, the weight coefficient β is set to 0.3 to 0.4;

[0036] Transmission requirements: When the data loss rate is required to be less than 0.001%, the weight coefficient γ is set to 0.6 to 0.7;

[0037] When the data loss rate is required to be between 0.1% and 0.01%, the weight coefficient γ is set to 0.3 to 0.4;

[0038] The comprehensive criticality score C is calculated based on the scores of each dimension and the corresponding weight coefficients i , divide the data stream into high criticality and low criticality data streams;

[0039] If C i 3.5-6.0: high criticality (needs priority treatment);

[0040] If C i 1.0 to 2.9: low criticality (can be processed later).

[0041] Furthermore, in the first stage, the algorithm encodes chromosomes according to high-criticality and low-criticality data flows. Assuming there are I data flows, a chromosome represents a scheduling scheme. The first half of the chromosome is used to represent the routing of the data flow, and the second half is used to represent the scheduling order of the flow. Each individual (also called chromosome) c = {f1, f2, ..., f I ;f I+1 , f I+2 , ..., f 2l} contains 2I genes; {f1, f2, ..., f l} represents the routing option assigned to the data flow (0 for source routing, 1 for graph routing), {f I+1 , fI+2 , ..., f 2I} represents the scheduling order of data flow. Specifically, f i ∈{1, ..., N} (i = 1, ..., I) indicates the number of cells assigned to data stream f. i The index of the routing type, where f1=0 means that the data flow f1 selects the source routing, and f2=1 means that the data flow f2 selects the graph routing; l+i (i=1, ...I) represents the index of the i-th scheduled data stream; for example, if f l+1 =2, it means that data stream f2 is the first data stream to be scheduled in the scheduling order; the preliminary scheduling scheme is generated by the crossover and mutation operation of the genetic algorithm, and the preliminary scheme is C = {1, 0, 0, ..., 1; f5, f2, f1, f I ..., f3}, the first half is used to indicate the routing selection of data flows, and the second half is used to indicate the scheduling order of data flows. For high-criticality data flows, graph routing with multiple paths will be preferentially allocated to ensure high-reliability transmission of real-time data.

[0042] In the second stage, the preliminary scheduling plan of the first stage is crossover and mutation operation. After multiple generations of crossover and mutation operation, the scheduling plan with the highest comfort value is selected by using the comfort function, that is, the global optimal scheduling plan C = {1, 0, 1, ..., 1; f8, f2, f5, f I ..., f3}, this scheme includes the global optimal routing allocation and data flow scheduling order.

[0043] Furthermore, the two-stage multi-population genetic algorithm scheduling subunit also includes evaluating the fitness of the scheduling scheme through a fitness function;

[0044] The fitness function comprehensively considers the transmission success rate, the matching degree of routing selection and data flow criticality; the core indicator of the fitness function is the transmission success rate, that is, the ratio of the number of successfully transmitted data flows to the total number of data flows; the higher the transmission success rate, the better the scheduling scheme corresponding to the chromosome; in addition, the fitness function will also be weighted according to whether the routing selection meets the criticality requirements of the data flow; therefore, by selecting chromosomes with higher fitness values, the transmission efficiency and overall reliability of the system can be optimized;

[0045] The following formula is used to define the Fitness function:

[0046] Fitness=Accep tanceRate+Misma tch_Penalty

[0047]

[0048] Among them, AcceptanceRate is the acceptance rate of data flow, SuccessfulTransmissions is the number of successfully transmitted data flows, TotalTransmissions is the total number of data flows, and Mismatch_Penalty is the weighted matching degree between routing selection and data flow criticality. If in a certain scheduling scheme, high-criticality data flows are assigned graph routing and low-criticality data flows are assigned source routing, then this scheme is assigned a higher weighted score;

[0049] The details are as follows:

[0050] When 20% of the data flows are assigned matching routes, that is, high-criticality data flows are assigned graph routes and low-criticality data flows are assigned source routes, then Mismatch_Penalty is 1 point;

[0051] When 20%-50% of the data flows are assigned matching routes, the Mismatch_Penalty is 2 points;

[0052] When 50%-80% of the data flows are assigned matching routes, the Mismatch_Penalty is 3 points;

[0053] When 80%-100% of the data flows are assigned matching routes, the Mismatch_Penalty is 4 points;

[0054] In the two-stage multi-population genetic algorithm, the fitness function uses the acceptance rate of the data stream as the evaluation index;

[0055] The acceptance rate is the ratio of the number of successfully transmitted data flows to the total number of data flows, which reflects the effectiveness of the scheduling scheme;

[0056] The fitness function evaluates and selects the best scheduling scheme by optimizing the acceptance rate, thereby ensuring that the selected scheme can maximize the successful transmission rate of the data flow and improve the overall scheduling efficiency of the system.

[0057] Furthermore, the path selection subunit also considers network topology, node load and link status to ensure that the data stream can be transmitted through the optimal path.

[0058] Furthermore, the path selection subunit comprises the steps of:

[0059] When a high-criticality data stream fails to transmit on the primary path of the route, the shared path is enabled as a backup transmission path to ensure that the data stream can still be transmitted reliably under high priority and avoid delays or data loss caused by the failure of the primary path; at the same time, low-criticality data streams continue to be transmitted through the source route if resources permit, so as to reduce overall resource consumption.

[0060] Furthermore, the transmission optimization subunit includes: continuously monitoring the transmission status of each data flow during operation, and adjusting the transmission strategy in real time according to the network conditions, to ensure that high-criticality data flows always occupy network resources first, and low-criticality data flows obtain transmission opportunities when the network load is low, and ultimately realize dynamic allocation and optimization of resources, thereby further improving the transmission efficiency and reliability of the overall system.

[0061] The real-time scheduling method of industrial Internet of Things based on the scheduling device comprises the following steps:

[0062] S1 Data stream collection and preprocessing: The data stream processing subunit collects various data from industrial IoT devices in real time and preprocesses these data streams, including data cleaning, encoding and encryption of the collected data streams to ensure data security and integrity;

[0063] S2 Data Stream Criticality Classification: The criticality classification subunit analyzes the collected data streams and classifies the data streams into high criticality data streams and low criticality data streams according to the real-time nature of the data, task importance, and transmission requirements;

[0064] S3 Scheduling Scheme Generation and Optimization: The two-stage multi-population genetic algorithm scheduling subunit first generates a preliminary scheduling scheme through a multi-population genetic algorithm, and further optimizes the scheduling scheme based on this;

[0065] S4 Optimal Path Selection: The path selection subunit selects the optimal transmission path for each data stream based on the optimized scheduling scheme;

[0066] S5 Dynamic optimization of transmission process: The transmission optimization subunit monitors the transmission process of data stream in real time and dynamically adjusts the transmission parameters according to the actual network status.

[0067] Further, S2 includes the steps of:

[0068] According to the real-time nature of data flow i , Task Importance i and transmission demand T i To calculate the data flow F i Criticality C i , based on the formula:

[0069] C i =αR i +βH i +γT i

[0070] Among them, α, β, and γ are weight coefficients, corresponding to the real-time nature of the data flow, the importance of the task, and the transmission requirements, respectively;

[0071] Real-timei The score is: The real-time score is used to measure the processing time requirements of the data flow; the score range is 1 to 3 points, and the specific definitions are as follows:

[0072] 1: Able to tolerate delays of 10ms or more;

[0073] 2: Able to tolerate a delay of 5ms to 10ms;

[0074] 3: The processing must be completed within 5ms, and there is almost no tolerance for delay;

[0075] Task Importance i The scores are: The task importance score is used to evaluate the criticality of the data flow in the business process; the higher the bandwidth requirement, the greater the importance of the task; the score range is 1 to 3 points, and the specific definitions are as follows:

[0076] 1: Low bandwidth requirement, less than 20Mbps;

[0077] 2: Medium bandwidth requirement, 50Mbps to 100Mbps;

[0078] 3: High bandwidth requirement, greater than 100Mbps;

[0079] Transmission requirement T i The score is: The transmission requirement score is used to measure the quality requirements of the data flow during the transmission process; the score range is 1 to 3 points, and the specific definitions are as follows:

[0080] 1: 0.05% data loss or error is allowed;

[0081] 2: 0.01% data loss or error is allowed;

[0082] 3: 0.001% data loss or error is allowed;

[0083] α evaluates the real-time performance of data streams by measuring performance indicators, β determines the importance of data streams based on their bandwidth requirements, and γ evaluates transmission requirements by analyzing the transmission reliability of data streams. The weight coefficient is dynamically adjusted based on the priority of business requirements:

[0084] Real-time performance: When the end-to-end delay of the data stream is required to be in the range of 1ms to 5ms, the weight coefficient α is set to 0.6 to 0.7;

[0085] When the end-to-end delay requirement is in the range of 10ms to 50ms, the weight coefficient α is set to 0.3 to 0.4;

[0086] Task importance: When the bandwidth requirement is greater than 100 Mbps, the weight coefficient β is set to 0.6 to 0.7;

[0087] When the bandwidth demand is greater than 10Mbps to 50Mbps, the weight coefficient β is set to 0.3 to 0.4;

[0088] Transmission requirements: When the data loss rate is required to be less than 0.001%, the weight coefficient γ is set to 0.6 to 0.7;

[0089] When the data loss rate is required to be between 0.1% and 0.01%, the weight coefficient γ is set to 0.3 to 0.4;

[0090] The comprehensive criticality score C is calculated based on the scores of each dimension and the corresponding weight coefficients i , divide the data stream into high criticality and low criticality data streams;

[0091] If C i 3.5~δ.0: high criticality data flow;

[0092] If C i 1.0 to 2.9: low criticality data flow.

[0093] Further, in S3, the improving the scheduling effect through the two-stage optimization process includes:

[0094] In the first stage, chromosomes are encoded according to high-criticality and low-criticality data flows. Assuming there are I data flows, a chromosome represents a scheduling scheme. The first half of the chromosome is used to represent the routing of the data flow, and the second half is used to represent the scheduling order of the flow. Each individual C = {f1, f2, ..., f I ;f I+1 , f I+2 , ..., f 2I} contains 2I genes.

[0095] {f1, f2, ..., f I} represents the routing selection assigned to the data flow, where 0 represents source routing, 1 represents graph routing, and {f l+1 , f l+2 , ..., f 2I} indicates the scheduling order of data streams.

[0096] Specifically, f i ∈{1,...,N}(i=1,....I) indicates the number of cells assigned to data stream f i The index of the routing type, where f1=0 means that the data flow f1 selects the source routing, and f2=1 means that the data flow f2 selects the graph routing; I+i (i=1, . . . I) represents the index of the i-th scheduled data stream.

[0097] For example, if f I+1=2, it means that data stream f2 is the first data stream to be scheduled in the scheduling order; the preliminary scheduling scheme is generated by the crossover and mutation operation of the genetic algorithm, and the preliminary scheme is C = {1, 0, 0, ..., 1; f5, f2, f1, f I ..., f3}, the first half is used to indicate the routing selection of data flows, and the second half is used to indicate the scheduling order of data flows. For high-criticality data flows, graph routing with multiple paths will be preferentially allocated to ensure high-reliability transmission of real-time data.

[0098] In the second stage, the preliminary scheduling scheme generated in the first stage is subjected to crossover and mutation operations. After multiple generations of crossover and mutation operations, the scheduling scheme with the highest comfort value is selected using the comfort function, i.e., the global optimal scheduling scheme C = {1, 0, 1, ..., 1; f8, f2, f5, f I ..., f3}, the scheme includes the globally optimal routing allocation and data flow scheduling order.

[0099] Compared with the prior art, the present invention has beneficial effects.

[0100] 1. This invention proposes an industrial Internet of Things real-time scheduling device based on a two-stage multi-population genetic algorithm, which can efficiently process data streams of different priorities in industrial networks and accurately adapt to complex and changing network environments. The device effectively distinguishes between critical data streams and non-critical data streams through advanced classification and scheduling strategies, ensuring high reliability and low latency in the transmission of critical data, while optimizing the transmission efficiency of non-critical data, significantly improving the overall performance of the system.

[0101] 2. The present invention improves the existing scheduling method and establishes an industrial Internet of Things real-time scheduling method and device based on a two-stage multi-population genetic algorithm, giving full play to the utilization potential of network resources. This method not only ensures the transmission success rate of time-sensitive data, but also improves the transmission efficiency of non-critical data streams through a two-stage collaborative optimization strategy, thereby effectively improving the scheduling accuracy and stability of the industrial Internet of Things system in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. The protection scope of the present invention is not limited to the following description.

[0103] Figure 1 It is a schematic diagram of the system structure of the present invention.

[0104] Figure 2 It is a schematic diagram of the device structure of the present invention.

[0105] Figure 3 It is a schematic diagram of the two-stage multi-population genetic algorithm in the device of the present invention.

[0106] Figure 4 It is a schematic diagram of the adaptive heterogeneous routing structure in the device of the present invention. DETAILED DESCRIPTION

[0107] In order to make the purpose, technical scheme and beneficial effects of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0108] Embodiment 1, as Figure 1 As shown, the industrial Internet of Things real-time scheduling device based on the two-stage multi-population genetic algorithm is based on the service platform system, and its structure includes an equipment data acquisition device, a scheduling device based on an intelligent algorithm, and a data display device.

[0109] Among them, the equipment data acquisition device consists of a sensor network and a control module. The sensor network is used to collect the operating data of various equipment in the industrial Internet of Things field environment, while the control module is responsible for managing the working status of the sensor nodes and uploading the collected and processed data to the intelligent scheduling module.

[0110] The scheduling communication device based on intelligent algorithm is composed of a heterogeneous multi-core processor system and an integrated wireless communication module. It is used to receive and classify device data and perform scheduling tasks. It outputs device data to the network wirelessly according to a multi-priority real-time data transmission scheduling method. It mainly includes five parts: data stream processing subunit, criticality classification subunit, two-stage multi-population genetic algorithm scheduling subunit, path selection subunit and transmission optimization subunit.

[0111] The display device adopts a human-computer interaction mode to show users the operating status, scheduling information, early warning prompts and other related information of industrial Internet of Things equipment in real time. Users can use this interface to monitor the system and optimize scheduling.

[0112] Embodiment 2, as Figure 2 As shown, the industrial Internet of Things real-time scheduling device based on a two-stage multi-population genetic algorithm includes: a data stream processing subunit, a criticality classification subunit, a two-stage multi-population genetic algorithm scheduling subunit, a path selection subunit and a transmission optimization subunit.

[0113] Among them, the data stream processing subunit is responsible for collecting and processing various data streams in the industrial Internet of Things network, providing basic data for subsequent classification and scheduling.

[0114] The criticality classification subunit classifies the data flows according to their real-time nature and importance, and divides the data flows into critical data flows and non-critical data flows, ensuring that different types of data flows are reasonably distinguished and prioritized during the scheduling process.

[0115] The two-stage multi-population genetic algorithm scheduling subunit optimizes the scheduling of data streams through a two-stage multi-population genetic algorithm, generates a preliminary scheduling scheme in the first stage, and further optimizes the scheme in the second stage to improve the overall transmission efficiency and reliability.

[0116] The path selection subunit selects the optimal transmission path for each data stream based on the scheduling results, ensuring that data can be efficiently transmitted in complex industrial environments and reducing delays and packet loss rates.

[0117] The transmission optimization subunit further optimizes the selected transmission path, dynamically adjusts the transmission path and strategy according to the network conditions, improves the transmission quality of the data stream, and maximizes the use of network resources.

[0118] The real-time scheduling method of the industrial Internet of Things based on the device includes the following steps:

[0119] S1 Data stream collection and preprocessing: The data stream processing subunit collects various data from industrial IoT devices in real time and preprocesses these data streams, including data cleaning, formatting, and preliminary analysis and processing, to ensure the integrity and accuracy of the data stream in subsequent processing steps.

[0120] S2 Data Stream Criticality Classification: The criticality classification subunit analyzes the collected data streams and classifies them into critical data streams and non-critical data streams based on the data's real-time nature, task importance and other attributes. Critical data streams are given higher priority to ensure that they are given priority in subsequent scheduling and transmission. i , Task Importance i and transmission demand T i To calculate the data flow F i The criticality of the classification is to ensure the scientificity and rationality of the classification. The criticality calculation formula is as follows:

[0121] C i =αR i +βH i +γT i

[0122] Among them, α, β, and γ are weight coefficients, corresponding to the real-time nature of the data flow, task importance, and transmission requirements, respectively.

[0123] Among them, critical data streams usually contain urgent data that needs to be processed first, such as alarm signals, real-time monitoring data, etc. This type of data requires high transmission reliability and low latency, so it has the highest priority. Non-critical data streams include regular status data, historical data or other data that does not have real-time requirements.

[0124] Critical data flows (highly critical data flows) give priority to graph routing paths. Specifically, we find a path with the highest reliability and the lowest latency for critical data flows in the network topology. First, we use the Dijkstra algorithm to calculate the shortest path between nodes in the network, and evaluate the transmission delay and reliability of each candidate path. Combining the two factors of delay and reliability, we select the optimal path for data transmission. The transmission path of critical data flows will be selected based on this, ensuring its high reliability and low latency in the network.

[0125] Non-critical data flows (low-criticality data flows) give priority to source routing paths. Specifically, according to the network load and bandwidth, a path with the least resource consumption is selected for non-critical data flows. By evaluating the load and bandwidth of each path, the path with the least resource consumption is selected for the transmission of non-critical data flows. This path selection strategy is adjusted according to the dynamic changes in network load to ensure that the transmission efficiency of non-critical data flows is improved without affecting the transmission of critical data flows.

[0126] S3 Scheduling Scheme Generation and Optimization: Two-stage Multi-population Genetic Algorithm The scheduling subunit first generates a preliminary scheduling scheme through a multi-population genetic algorithm. In this stage, multiple populations evolve in parallel to generate diverse and initially adaptable schemes, laying the foundation for subsequent optimization. Subsequently, based on the preliminary scheme, the scheduling subunit further optimizes the scheduling scheme by introducing a local search mechanism and a refined adjustment strategy. This optimization process significantly improves the transmission efficiency of key data streams and the utilization of network resources, ensuring that the transmission process of the entire network is more efficient and reliable. The optimization objective function of its scheduling algorithm is:

[0127]

[0128] Among them, AcceptanceRate is the acceptance rate of the data flow, SuccessfulTransmissions is the number of successfully transmitted data flows, and TotalTransmissions is the total number of data flows.

[0129] S4 Optimal Path Selection: The path selection subunit selects the optimal transmission path for each data stream based on the optimized scheduling scheme. Taking into account multiple factors such as network topology, node load, and link status, this step ensures that the data stream can be transmitted through the optimal path, thereby reducing delays, packet loss rates, and improving the overall transmission success rate.

[0130] S5 Dynamic optimization of transmission process: The transmission optimization subunit monitors the transmission process of data streams in real time and dynamically adjusts transmission parameters according to the actual network status, including bandwidth allocation, transmission rate and retransmission mechanism, to ensure the transmission quality of data streams and the optimal use of network resources. At the same time, the transmission optimization subunit also handles emergencies during transmission, and adjusts the path and transmission strategy in time to ensure the stable transmission of data streams.

[0131] Embodiment 3, as Figure 3 As shown in the figure, the two-stage multi-population genetic algorithm includes five parts: population initialization, global optimization, local optimization, experimental verification, and result output. First, the basis of the algorithm is initialized by randomly generating multiple populations and designing a comfort function. Next, the first stage performs global optimization, using multi-population co-evolution and migration strategies to improve global search capabilities and prevent premature convergence. In the second stage, local optimization is performed to further improve the accuracy and efficiency of the algorithm through deep search and adaptive parameter adjustment. Subsequently, the effect of the comfort function and optimization parameters is experimentally verified to ensure the effectiveness of the algorithm. Finally, the optimal solution obtained by optimization is output, and the parameter configuration and experimental data during the optimization process are recorded.

[0132] Embodiment 4, as Figure 4 As shown in the figure, the structure of the adaptive heterogeneous routing model is shown. Based on the criticality of data flow, the model combines the advantages of source routing and graph routing to dynamically select the optimal routing strategy. For critical data flows, the model gives priority to graph routing, which ensures high reliability and low latency through multi-path transmission to prevent interruption or delay; while for non-critical data flows, source routing is selected to reduce the computing overhead and the transmission amount of control information, which is more suitable for data transmission that tolerates a certain delay. Especially in the industrial Internet of Things environment, this adaptive routing selection mechanism significantly improves the transmission success rate of critical data flows and the overall stability of the network, while effectively reducing transmission conflicts and data packet losses caused by improper routing selection. Among them, the solid line represents the main link and the dotted line represents the shared link. Source routing and graph routing have the same selection on the main path, but in graph routing, data is first tried to be transmitted on the main link. If the transmission fails, it is switched to the shared path for transmission. By analyzing the criticality of data flows in real time, the mechanism can dynamically adjust the routing strategy to achieve optimal resource allocation and maximize network performance.

[0133] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features therein may be replaced by equivalents. Therefore, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A real-time scheduling device for industrial Internet of Things based on a two-stage multi-population genetic algorithm, characterized in that: include: The data stream processing subunit is used to collect and process various data streams in the industrial Internet of Things network, and provide basic data for the classification of the criticality classification subunit and the scheduling of the two-stage multi-population genetic algorithm scheduling subunit; The criticality classification subunit is used to classify the data streams processed by the data stream processing subunit into two categories: high criticality data streams and low criticality data streams, so as to ensure reasonable distinction and priority setting of different types of data streams during the scheduling process; A two-stage multi-population genetic algorithm scheduling subunit is used to receive and process high-criticality data streams and low-criticality data streams, and improve the scheduling effect through a two-stage optimization process; The path selection subunit is used to select the path with the highest reliability for high-criticality data flows and the path with the lowest resource consumption for low-criticality data flows based on the scheduling results of the two-stage multi-population genetic algorithm scheduling subunit and the current network topology information; ensuring that data can be transmitted efficiently in complex industrial environments and reducing delays and packet loss rates; The transmission optimization subunit monitors the transmission process of the data stream in real time and dynamically adjusts the transmission parameters according to the actual network status.

2. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: The various data streams include periodic data streams, event-driven data streams, control instruction data streams, real-time control data streams, regular status data streams, and multimedia data streams.

3. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: The data stream processing subunit also includes a preprocessing module for performing data cleaning, encoding and encryption on the collected data stream to ensure the security and integrity of the data.

4. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: The criticality classification subunit is based on the real-time nature of the data stream. i , Task Importance i and transmission demand T i To calculate the data flow F i Criticality C i , based on the formula: C i =αR i +βH i +γT i Among them, α, β, and γ are weight coefficients, corresponding to the real-time nature of the data flow, task importance, and transmission requirements, respectively; Real-time i The score is: The real-time score is used to measure the processing time requirements of the data flow; the score range is 1 to 3 points, and the specific definitions are as follows: 1: Able to tolerate delays of 10ms or more; 2: Able to tolerate a delay of 5ms to 10ms; 3: The processing must be completed within 5ms, and delays are not tolerated; Task Importance i The scores are: The task importance score is used to evaluate the criticality of the data flow in the business process; the higher the bandwidth requirement, the greater the importance of the task; the score range is 1 to 3 points, and the specific definitions are as follows: 1: Low bandwidth requirement, less than 20Mbps; 2: Medium bandwidth requirement, 50Mbps to 100Mbps; 3: High bandwidth requirement, greater than 100Mbps; Transmission requirement T i The score is: The transmission requirement score is used to measure the quality requirements of the data flow during the transmission process; the score range is 1 to 3 points, and the specific definitions are as follows: 1: 0.05% data loss or error is allowed; 2: 0.01% data loss or error is allowed; 3: 0.001% data loss or error is allowed; α evaluates the real-time performance of data streams by measuring performance indicators, β determines the importance of data streams based on their bandwidth requirements, and γ evaluates transmission requirements by analyzing the transmission reliability of data streams. The weight coefficient is dynamically adjusted based on the priority of business requirements: Real-time performance: When the end-to-end delay of the data stream is required to be in the range of 1ms to 5ms, the weight coefficient α is set to 0.6 to 0.7; When the end-to-end delay requirement is in the range of 10ms to 50ms, the weight coefficient α is set to 0.3 to 0.4; Task importance: When the bandwidth demand is greater than 100 Mbps, the weight coefficient β is set to 0.6 to 0.7; when the bandwidth demand is greater than 10 Mbps to 50 Mbps, the weight coefficient β is set to 0.3 to 0.4; Transmission requirements: When the data loss rate is required to be less than 0.001%, the weight coefficient γ is set to 0.6 to 0.7; When the data loss rate is required to be between 0.1% and 0.01%, the weight coefficient γ is set to 0.3 to 0.4; The comprehensive criticality score C is calculated based on the scores of each dimension and the corresponding weight coefficients i , divide the data stream into high criticality and low criticality data streams; If C i 3.5 to 6.0: high criticality data flow; If C i 1.0 to 2.9: low criticality data flow.

5. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: In the two-stage multi-population genetic algorithm scheduling subunit, the two-stage optimization process to improve the scheduling effect includes: In the first stage, chromosomes are encoded according to high-criticality and low-criticality data flows. Assuming there are I data flows, a chromosome represents a scheduling scheme. The first half of the chromosome is used to represent the routing of the data flow, and the second half is used to represent the scheduling order of the flow. Each individual C = {f1, f2, ..., f I ;f l+1 , f l+2 , ..., f 2I } contains 2I genes; {f1, f2, ..., f I } represents the routing selection assigned to the data flow, where 0 represents source routing, 1 represents graph routing, and {f l+1 , f l+2 , ..., f 2I } indicates the scheduling order of data streams; Specifically, f i ∈{1,...,N} (i=1,...,I) indicates the number of cells assigned to data stream f. i The index of the routing type, where f1=0 means that the data flow f1 selects the source routing, and f2=1 means that the data flow f2 selects the graph routing; I+i (i=1, ...I) represents the index of the i-th scheduled data stream; For example, if f I+1 =2, it means that data stream f2 is the first data stream to be scheduled in the scheduling order; the preliminary scheduling scheme is generated by the crossover and mutation operation of the genetic algorithm, and the preliminary scheme is C = {1, 0, 0, ..., 1; f5, f2, f1, f I ..., f3}, the first half is used to indicate the routing selection of data flows, and the second half is used to indicate the scheduling order of data flows. For high-criticality data flows, graph routing with multiple paths will be preferentially allocated to ensure high-reliability transmission of real-time data; In the second stage, the preliminary scheduling scheme generated in the first stage is subjected to crossover and mutation operations. After multiple generations of crossover and mutation operations, the scheduling scheme with the highest comfort value is selected using the comfort function, i.e., the global optimal scheduling scheme C = {1, 0, 1, ..., 1; f8, f2, f5, f I ..., f3}, the scheme includes the globally optimal routing allocation and data flow scheduling order.

6. The real-time scheduling device for industrial Internet of Things according to claim 5, characterized in that: The two-stage multi-population genetic algorithm scheduling subunit also includes evaluating the fitness of the scheduling scheme through a fitness function; The core indicator of the fitness function is the transmission success rate, which is the ratio of the number of successfully transmitted data flows to the total number of data flows. The higher the transmission success rate, the better the scheduling scheme corresponding to the chromosome. The fitness function also weights the routing selection according to whether it meets the criticality requirements of the data flow. Therefore, by selecting chromosomes with higher fitness values, the transmission efficiency and overall reliability of the system can be optimized. The following formula is used to define the Fitness function: Fitness=AcceptanceRate+Mismatch_Penalty Among them, AcceptanceRate is the acceptance rate of the data flow, SuccessfulTransmissions is the number of successfully transmitted data flows, TotalTransmissions is the total number of data flows, Mismatch_Penalt y Weight the matching degree between routing selection and data flow criticality. If the scheduling scheme assigns graph routing to high criticality data flows and source routing to low criticality data flows, then this scheme is assigned a higher weighted score. The details are as follows: When 20% of the data flows are assigned matching routes, that is, high-criticality data flows are assigned graph routes and low-criticality data flows are assigned source routes, then Mismatch_Penalty is 1 point; When 20%-50% of the data flows are assigned matching routes, the Mismatch_Penalty is 2 points; When 50%-80% of the data flows are assigned matching routes, the Mismatch_Penalty is 3 points; When 80%-100% of the data flows are assigned matching routes, the Mismatch_Penalty is 4 points; In the two-stage multi-population genetic algorithm, the fitness function uses the acceptance rate of the data stream as the evaluation index; The acceptance rate is the ratio of the number of successfully transmitted data flows to the total number of data flows, which reflects the effectiveness of the scheduling scheme; The fitness function evaluates and selects the best scheduling scheme by optimizing the acceptance rate, thereby ensuring that the selected scheme can maximize the successful transmission rate of the data flow and improve the overall scheduling efficiency of the system.

7. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: The path selection subunit also considers network topology, node load and link status to ensure that the data stream can be transmitted through the optimal path.

8. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: The path selection subunit comprises the steps of: When a high-criticality data stream fails to transmit on the primary path of the router, the shared path is enabled as a backup transmission path to ensure that the data stream can still be transmitted reliably under high priority and avoid delays or data loss caused by the failure of the primary path; At the same time, low-criticality data flows continue to be transmitted via source routing if resources permit, to reduce overall resource consumption.

9. The real-time scheduling device for industrial Internet of Things according to claim 1, characterized in that: The transmission optimization subunit includes: continuously monitoring the transmission status of each data stream during operation, and adjusting the transmission strategy in real time according to the network status, ensuring that high-criticality data streams always occupy network resources first, and low-criticality data streams get transmission opportunities when the network load is low, and ultimately realizing dynamic allocation and optimization of resources, thereby further improving the transmission efficiency and reliability of the overall system.

10. A real-time scheduling method for industrial Internet of Things based on a two-stage multi-population genetic algorithm, characterized in that: The scheduling method is based on the industrial Internet of Things real-time scheduling device according to any one of claims 1 to 9, and the scheduling method specifically comprises the following steps: S1 Data stream collection and preprocessing: The data stream processing subunit collects various data from industrial IoT devices in real time and preprocesses these data streams, including data cleaning, formatting, and preliminary analysis and processing; S2 Data Stream Criticality Classification: The criticality classification subunit analyzes the collected data streams and classifies the data streams into high criticality data streams and low criticality data streams according to the real-time nature of the data, task importance, and transmission requirements; S3 Scheduling Scheme Generation and Optimization: The two-stage multi-population genetic algorithm scheduling subunit first generates a preliminary scheduling scheme through a multi-population genetic algorithm, and further optimizes the scheduling scheme based on this; S4 Optimal Path Selection: The path selection subunit selects the optimal transmission path for each data stream based on the optimized scheduling scheme; S5 Dynamic optimization of transmission process: The transmission optimization subunit monitors the transmission process of data stream in real time and dynamically adjusts the transmission parameters according to the actual network status.

Images