Self-powered integrated sensing and decentralized distributed industrial intelligent gateway control method

Abstract

The present application relates to the technical field of industrial intelligent gateway, and particularly relates to a self-powered integrated sensing decentralized distributed industrial intelligent gateway control method. The method collects environmental energy characteristics through an integrated sensing module, calculates a real-time energy storage surplus and a predicted consumption difference, generates an energy state vector and broadcasts in a distributed state machine; when it is determined that the energy of an edge node is lower than a preset forwarding threshold, a distributed election algorithm is triggered to determine the adjacent gateway with the highest remaining energy as a takeover gateway; the takeover gateway takes over the data relay link and dynamically adjusts the encapsulation format and routing hop count of the to-be-transmitted data packet according to the protocol conversion stack capability. The present application avoids the physical disconnection of the data transmission link caused by the energy depletion of a single point, and guarantees the data integrity and consistency during the transfer of the to-be-forwarded buffer queue in the topology switching process.

Description

Technical Field

[0001] This invention relates to the field of industrial intelligent gateway technology, specifically to a self-powered integrated sensing decentralized distributed industrial intelligent gateway control method. Background Technology

[0002] Existing industrial smart gateways generally employ integrated sensing and energy harvesting technologies to maintain operation in environments without external power supply. These gateways acquire ambient energy through piezoelectric conversion, thermoelectric conversion, etc., and store it in local energy storage devices. At the data communication level, existing industrial smart gateway clusters use pre-configured static routing tables for data relay, or rely on a central controller at the core to uniformly allocate forwarding paths. When a gateway node experiences a decrease in local energy storage due to insufficient ambient energy input, the existing gateway still performs packet reception and forwarding operations according to the pre-configured static routing table. The interaction messages between gateway nodes only contain business sensor data and fixed-period status heartbeat information, lacking mechanisms for sharing and processing real-time energy storage status.

[0003] Existing industrial smart gateway clusters employing static routing mechanisms cannot adapt to drastic power fluctuations in passive nodes caused by environmental energy volatility. When the energy storage capacity of a specific gateway drops to a critical value that cannot support data relay operations, the static routing mechanism forces the gateway to continue undertaking data packet forwarding tasks. This causes the gateway's energy storage device to trigger over-discharge protection and system shutdown, resulting in a physical disconnection of the entire data transmission link passing through the gateway. Current technology lacks a mechanism to reconstruct the communication topology within the distributed gateway cluster based on the real-time energy status of nodes to avoid link disconnection. Summary of the Invention

[0004] The purpose of this invention is to provide a self-powered integrated sensing decentralized distributed industrial intelligent gateway control method, which can solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A self-powered integrated sensing decentralized distributed industrial smart gateway control method includes the following steps: the integrated sensing module of the industrial smart gateway collects environmental energy characteristics in real time; Calculate the difference between the current real-time energy storage capacity and the predicted consumption based on the environmental energy characteristics, and generate an energy state vector containing the difference between the real-time energy storage capacity and the predicted consumption. The industrial smart gateway broadcasts the energy state vector in a pre-built distributed state machine; When the energy state vector indicates that the energy of the current edge node is lower than a preset forwarding threshold, the current edge node triggers a distributed election algorithm to determine the neighboring gateway with the highest remaining energy from the neighboring gateways that received the energy state vector as the takeover gateway. The takeover gateway takes over the data relay link of the current edge node; The takeover gateway dynamically adjusts the encapsulation format and routing hop count of the data packets to be transmitted in the data relay link based on its own maintained protocol conversion stack capabilities.

[0006] Preferably, the step of the integrated sensing module of the industrial intelligent gateway collecting environmental energy characteristics in real time, calculating the difference between the current real-time energy storage capacity and the predicted consumption based on the environmental energy characteristics, and generating an energy state vector containing the difference between the real-time energy storage capacity and the predicted consumption includes: the integrated sensing module acquiring the original voltage signal and the original current signal of the multimodal sensor according to a preset sampling period; The original voltage signal and the original current signal are filtered to extract smooth voltage curves and smooth current curves; The smoothed voltage curve is input into a pre-built energy prediction model, which outputs the predicted power generation within a future preset time window. Read the current terminal voltage of the energy storage device and calculate the current real-time energy storage margin by combining it with the discharge characteristic curve of the energy storage device. The historical average power consumption of the industrial smart gateway when it is in working state is obtained, and the historical average power consumption is multiplied by the preset time window to obtain the predicted power consumption difference.

[0007] Preferably, the step of the industrial intelligent gateway broadcasting the energy state vector in a pre-built distributed state machine includes: the industrial intelligent gateway encapsulating the energy state vector into a specified field of a distributed consensus message; During the idle time slots of the distributed state machine, the industrial intelligent gateway uses a carrier sense multiple access mechanism to compete for the physical channel. After successfully occupying the physical channel, the industrial smart gateway sends the distributed consensus message encapsulated with the energy state vector to the adjacent gateway. After receiving the distributed consensus message, the neighboring gateway parses the specified field to obtain the energy state vector and updates the neighbor energy state table maintained locally, wherein the neighbor energy state table records the mapping relationship between the node identifier of the neighboring gateway and the energy state vector.

[0008] Preferably, the step of triggering a distributed election algorithm when the energy state vector indicates that the energy of the current edge node is lower than a preset forwarding threshold, and determining the neighboring gateway with the highest remaining energy from the neighboring gateways that received the energy state vector as the takeover gateway, includes: the current edge node comparing the real-time energy storage margin with the preset forwarding threshold. When the real-time energy storage capacity is less than the preset forwarding threshold, the current edge node sends an election request message containing its own node identifier to the adjacent gateway. After receiving the election request message, the adjacent gateway appends the real-time energy storage capacity stored locally to the election response message and returns it. The current edge node extracts the real-time energy storage capacity from all received election response messages, and determines the adjacent gateway corresponding to the largest real-time energy storage capacity as the takeover gateway.

[0009] Preferably, the step of the takeover gateway taking over the data relay link of the current edge node includes: the current edge node sending a link migration instruction to the takeover gateway, wherein the link migration instruction carries the routing table entry of the downlink node to be migrated and the queue head pointer of the forwarding cache queue; The takeover gateway retrieves data packets from the forwarding cache queue from the current edge node using a point-to-point transmission protocol based on the queue head pointer. The takeover gateway reconstructs the routing table entry of the downlink node to be migrated in its local memory and writes the pulled data packets into its local sending buffer. The takeover gateway sends a topology update announcement to the downlink node to be migrated, updating the uplink next-hop address of the downlink node to be migrated to the physical address of the takeover gateway.

[0010] Preferably, the step of the takeover gateway dynamically adjusting the encapsulation format and routing hop count of the data packets to be transmitted in the data relay link based on its own maintained protocol conversion stack capabilities includes: the takeover gateway parsing the local configuration file to obtain a list of currently active protocol conversion stack instances; The takeover gateway queries a preset protocol overhead mapping table based on the protocol conversion stack instance list to obtain the packet header compression ratio and maximum transmission unit corresponding to different protocol stacks; The takeover gateway compares the original length of the data packet to be transmitted with the size of the maximum transmission unit, and selects a target protocol conversion stack instance that satisfies the condition that the original length is less than the maximum transmission unit and the packet header compression ratio is the largest. The takeover gateway rewrites the header field of the data packet to be transmitted according to the encapsulation rules corresponding to the target protocol conversion stack instance, and recalculates the routing hop count based on the updated network topology path.

[0011] Preferably, the step of the integrated sensing module acquiring the original voltage signal and original current signal of the multimodal sensor according to a preset sampling period includes: the integrated sensing module synchronously activating the piezoelectric vibration sensor interface, the thermoelectric temperature difference sensor interface and the radio frequency energy harvesting sensor interface; The piezoelectric vibration sensor interface acquires the first AC voltage signal generated by the mechanical vibration of the piezoelectric material. The thermoelectric temperature difference sensor interface acquires the second DC voltage signal generated by the surface temperature difference of industrial equipment; The radio frequency energy harvesting sensor interface acquires the third pulsating voltage signal after the spatial radio frequency signal has passed through the rectifier circuit; The integrated sensing module splices the first AC voltage signal, the second DC voltage signal, and the third pulsating voltage signal into the original voltage signal, and splices the corresponding current signal into the original current signal.

[0012] Preferably, the step of the current edge node sending an election request message containing its own node identifier to the adjacent gateway further includes: the current edge node generating a random backoff timer locally; After the backoff timer expires, the current edge node checks whether the physical channel is in an idle state. If the physical channel is in the idle state, the election request message is sent, and the status flag of the local state machine is switched from the normal forwarding state to the election initiation state. While the status flag is in the election initiation state, the current edge node discards the election request messages received from other gateways, and switches the status flag to the link migration state after receiving the election response message.

[0013] Preferably, the step of the takeover gateway retrieving data packets from the forwarding buffer queue from the current edge node via a point-to-point transmission protocol based on the queue head pointer includes: the takeover gateway sending a data retrieval request to the current edge node based on the queue head pointer, wherein the data retrieval request includes the start sequence number and end sequence number of the data packets to be retrieved; The current edge node calculates the cyclic redundancy check code of the corresponding cached data fragment based on the start sequence number and the end sequence number, and appends the cyclic redundancy check code to the end of the cached data fragment before sending it. The takeover gateway receives the cached data fragment, recalculates the local cyclic redundancy check code of the cached data fragment, and writes the cached data fragment into the local transmission buffer after the comparison is consistent.

[0014] Preferably, the step of the takeover gateway rewriting the header field of the data packet to be transmitted according to the encapsulation rules corresponding to the target protocol conversion stack instance includes: the takeover gateway parsing the original header field of the data packet to be transmitted and extracting the source address field, destination address field and payload type field; The takeover gateway queries the address mapping dictionary corresponding to the target protocol conversion stack instance locally, and converts the source address field and the destination address field into a gateway identifier and a terminal identifier in the target protocol format; The takeover gateway matches the corresponding quality of service level identifier and data fragmentation flag in the protocol specification of the target protocol conversion stack instance based on the load type field. The takeover gateway writes the gateway identifier, the terminal identifier, the quality of service level identifier, and the data fragmentation flag into a new packet header according to the field arrangement order of the target protocol format, replacing the original packet header fields.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention collects environmental energy characteristics through an integrated sensing module, calculates the difference between real-time energy storage capacity and predicted consumption, generates an energy state vector, and broadcasts it in a distributed state machine. When the energy of an edge node falls below a forwarding threshold, a distributed election algorithm is triggered, transferring the data relay link to the adjacent gateway with the highest remaining energy. This control method combines the real-time energy storage status of the gateway with a distributed topology election mechanism, completing the migration of forwarding tasks before the edge node's power drops to a critical shutdown value. This allows the communication topology of the industrial gateway cluster to be dynamically reconstructed according to changes in node power, avoiding physical disconnection of the entire data transmission link caused by single-point energy depletion.

[0016] 2. After taking over the link, the takeover gateway selects the target protocol conversion stack instance based on the header compression ratio and maximum transmission unit in the protocol conversion stack instance list and protocol overhead mapping table. It then rewrites the header fields of the data packets to be transmitted and recalculates the routing hop count based on the updated network topology. During data migration, cached data fragments are extracted via data retrieval requests carrying queue head pointers and sequence numbers, and cyclic redundancy check (CRC) codes are used for data integrity comparison. These methods enable adaptive protocol format conversion in the reconstructed data relay link, ensuring data integrity and consistency during the transfer of the forwarding cache queue to the takeover gateway's local transmission buffer during topology switching. Attached Figure Description

[0017] Figure 1 Flowchart of the overall control method for a self-powered integrated sensing decentralized distributed industrial intelligent gateway; Figure 2 Flowchart for energy harvesting and energy state vector generation in a multimodal environment; Figure 3 Flowchart for distributed broadcasting of energy state vectors and updating of neighbor state tables; Figure 4 A flowchart is provided for the execution and takeover gateway of the distributed election algorithm. Figure 5 A flowchart illustrating the data relay link takeover and lossless data migration process; Figure 6 Flowchart for adaptive selection of protocol conversion stack and adjustment of data packet encapsulation. Detailed Implementation

[0018] The control method described in this specific embodiment is applied to a decentralized distributed industrial intelligent gateway cluster. The gateway cluster is deployed in industrial field environments without external power supply, including but not limited to underground mines, along oil and gas pipelines, remote wind and solar power stations, and unattended areas of large industrial plants. Each industrial intelligent gateway has the ability to collect environmental energy, store local energy, convert multiple protocols, and relay data. The gateways communicate with each other through wireless or wired industrial Ethernet, forming a distributed network topology without a central node.

[0019] Please refer to the attached document. Figure 1In one embodiment, the integrated sensing module of the industrial smart gateway collects ambient energy characteristics in real time. The integrated sensing module is electrically connected to the gateway's built-in ambient energy harvesting unit and local energy storage device. The ambient energy harvesting unit converts ambient energy in the industrial site into electrical energy, and the local energy storage device stores the converted electrical energy to provide power support for the gateway's operation. The ambient energy characteristics include the real-time output electrical signal parameters of the ambient energy harvesting unit and the real-time operating parameters of the energy storage device. Based on the collected ambient energy characteristics, the difference between the current real-time energy storage margin and the predicted consumption is calculated, generating an energy state vector containing the difference between the real-time energy storage margin and the predicted consumption. The real-time energy storage margin is the total amount of available electrical energy stored in the energy storage device at the current moment, and the predicted consumption difference is the total amount of electrical energy required for the gateway to maintain normal operation within a preset future time window. The energy state vector is a one-dimensional data sequence encoded according to a preset format, used to characterize the gateway's current energy state and future energy change trends. The industrial intelligent gateway broadcasts an energy state vector within a pre-built distributed state machine. This distributed state machine is pre-deployed in all industrial intelligent gateways within the same cluster. The distributed state machine contains multiple sets of preset state flags, each corresponding to a set of preset behavioral rules for the gateway. All gateways within the cluster follow the same state machine operating rules, achieving behavioral coordination within the distributed network. When the energy state vector indicates that the current edge node's energy is below a preset forwarding threshold, the current edge node triggers a distributed election algorithm to determine the neighboring gateway with the highest remaining energy from among those that received the energy state vector, as the takeover gateway. The preset forwarding threshold is pre-stored in the gateway's local non-volatile memory, corresponding to the minimum energy value that the energy storage device can support to complete the preset operation. Neighboring gateways are those in the same local area network segment as the current edge node and can communicate directly with it within the same cluster. The distributed election algorithm runs locally on all participating gateways, requiring no central node scheduling; the takeover gateway is determined solely through message exchange between nodes. The takeover gateway takes over the data relay link of the current edge node. The data relay link is the communication link used by the current edge node to forward data uploaded by downlink nodes and sent by uplink nodes. During the link takeover process, the current edge node synchronizes the link-related routing information and the data to be forwarded to the takeover gateway to ensure no data loss during the link switchover. Based on its maintained protocol conversion stack capabilities, the takeover gateway dynamically adjusts the encapsulation format and hop count of the data packets to be transmitted in the data relay link. The protocol conversion stack consists of pre-deployed program instances on the gateway used to convert between different industrial communication protocols. Each protocol conversion stack instance corresponds to the parsing and encapsulation rules of an industrial communication protocol. The hop count is the number of gateway nodes a data packet must pass through to travel from the source address to the destination address.

[0020] Specifically, the calculation process for real-time energy storage capacity is quantitatively described by the following formula: , in, This represents the current real-time energy storage capacity, expressed in joules. This represents the current terminal voltage of the energy storage device, measured in volts. The capacitance-voltage characteristic function of the energy storage device is obtained by fitting a pre-stored discharge characteristic curve using cubic spline interpolation, with units in farads. The discharge characteristic curve of the energy storage device is obtained by conducting full-range charge-discharge tests on the same type of energy storage device beforehand. During the test, the capacitance values ​​corresponding to different terminal voltages are recorded, forming multiple sets of discrete voltage-capacitance data. A continuously differentiable capacitance-voltage characteristic function is obtained by fitting the data using cubic spline interpolation, ensuring that the calculated real-time energy storage margin is consistent with the actual usable power of the energy storage device.

[0021] The calculation process for the predicted consumption difference is quantitatively described by the following formula: , in, The unit for predicting the consumption difference is joules; The historical average power consumption of the industrial smart gateway when it is in operation, in watts; The preset time window is in seconds. The historical average power consumption is the average power consumption of the gateway over a preset number of consecutive working cycles. The working cycle includes data acquisition cycle, data forwarding cycle, and sleep cycle. The power consumption of each cycle is obtained by integrating the product of the gateway's real-time operating voltage and operating current within that cycle. The preset time window value can be adjusted through the gateway's local configuration file, with an adjustment range of 10s to 3600s, to adapt to the energy fluctuation characteristics and data transmission requirements of different industrial sites.

[0022] The generation process of the energy state vector is quantitatively described by the following formula: , in, This is the generated energy state vector; The difference between the real-time energy storage capacity and the predicted consumption is the value that satisfies the following conditions. This is used to characterize the energy surplus or deficit status of the gateway within a preset time window; This serves as a unique identifier for the current gateway node, using a 48-bit physical address to ensure that there are no duplicate node identifiers within the same cluster. The timestamp for generating the energy state vector is generated using the gateway's local real-time clock with millisecond precision. It is used to identify the generation time of the energy state vector and provides a basis for subsequent vector validity judgment.

[0023] The industrial intelligent gateway encapsulates the generated energy state vector into a designated field of the distributed consensus message. During the idle time slots of the distributed state machine, a carrier sense multiple access (CSMA) mechanism is used to compete for the physical channel. The idle time slots are pre-defined time intervals within the distributed state machine used for broadcasting node state information. The duration of each time slot matches the gateway's communication baud rate. Within each time slot, the gateway continuously listens for the carrier signal of the physical channel. When it detects no carrier signal for a consecutive inter-frame interval, it determines that the channel is idle. After successfully occupying the physical channel, the industrial intelligent gateway sends a distributed consensus message encapsulating the energy state vector to neighboring gateways. Upon receiving the distributed consensus message, the neighboring gateway parses the designated field of the message to obtain the energy state vector and updates its locally maintained neighbor energy state table. The neighbor energy state table records the mapping relationship between the neighboring gateway's node identifier and the energy state vector, and also records the update timestamp of each vector. When the timestamp exceeds a preset aging period, the corresponding entry is deleted from the neighbor energy state table, ensuring that the locally maintained neighbor energy state remains consistent with the actual state within the network.

[0024] Table 1. Distributed Consensus Message Field Structure Definition Table Preamble 8 The clock synchronization and frame start detection at the receiving end are fixed to a preset binary sequence. Unsigned integer Start of frame delimiter 2 Used to identify the end of the preamble and the beginning of the valid message content, it is fixed to a preset hexadecimal value. Unsigned integer Source Node ID 6 The unique physical address of the gateway node that sends the message, used to identify the sender of the message. Unsigned integer Destination Node ID 6 The receiver address of the message is filled with the broadcast address in all hexadecimal FF when broadcasting the message. Unsigned integer Message type 1 Used to identify the type of message; the energy status broadcast message corresponds to a preset type value. Unsigned integer Message length 2 The total byte length used to identify the message payload, including the length of the energy state vector field. Unsigned integer Energy state vector field variable Used to store the encapsulated energy state vector, the field length is determined by the number of elements in the vector and the encoding format. byte array Frame check sequence 4 Used for error checking during message transmission, generated using a 32-bit cyclic redundancy check (CRC) code. Unsigned integer Table 1 defines the field structure of distributed consensus messages. This table clarifies the length, meaning, and data type of each field in the message. All gateways within the same cluster follow the message format defined in this table for encoding and parsing, ensuring that all gateways can correctly extract the energy state vector field from the message, thus achieving reliable sharing of node energy states within the distributed network.

[0025] The current edge node continuously compares the real-time energy storage margin in its local energy state vector with a preset forwarding threshold. When the real-time energy storage margin is less than the preset forwarding threshold, the current edge node triggers a distributed election algorithm, sending an election request message containing its own node identifier to all neighboring gateways. Upon receiving the election request message, the neighboring gateways append their locally stored real-time energy storage margin to the election response message and return it to the current edge node within a preset response time. Within a preset election timeout period, the current edge node receives election response messages from all neighboring gateways, extracts the real-time energy storage margin values ​​from all valid election response messages, sorts all extracted real-time energy storage margin values, and determines the neighboring gateway corresponding to the largest real-time energy storage margin value as the takeover gateway.

[0026] The current edge node sends a link migration command to the designated takeover gateway. This command carries the routing table entry for the downlink node to be migrated and a pointer to the head of the forwarding buffer queue. The downlink node to be migrated is an industrial field terminal device or a lower-level gateway that uses the current edge node as its uplink next hop. The routing table entry includes the downlink node's physical address, network address, quality of service level, and data forwarding cycle. The forwarding buffer queue uses a first-in, first-out (FIFO) storage structure, with the queue head pointer pointing to the starting address of the first data packet to be forwarded in the queue. Based on the received queue head pointer, the takeover gateway retrieves all data packets from the forwarding buffer queue of the current edge node using a point-to-point transmission protocol. It then reconstructs the routing table entry for the downlink node to be migrated in its local memory and writes all the retrieved data packets into its local transmit buffer. The takeover gateway sends a topology update announcement to all downlink nodes to be migrated, updating the uplink next-hop address of the downlink nodes to be migrated to the physical address of the takeover gateway. After receiving the topology update announcement, the downlink nodes to be migrated update their local routing table entries and return an acknowledgment message to the takeover gateway. After receiving the acknowledgment messages from all downlink nodes to be migrated, the takeover gateway completes the takeover operation of the data relay link.

[0027] The takeover gateway parses its local configuration file to obtain a list of currently active protocol conversion stack instances. These instances are pre-deployed on the gateway's local system, and each instance corresponds to a commonly used industrial communication protocol, including a protocol parsing module, a protocol encapsulation module, an address mapping module, and a header compression module. Only active instances can participate in the protocol selection process. Based on the obtained list of instances, the gateway queries its pre-stored protocol overhead mapping table to obtain the header compression ratio and maximum transmission unit (MTU) for different instances. The MTU is the maximum length of a single data packet supported by the communication protocol, and the header compression ratio characterizes the compression effect of the protocol on the packet header. The gateway compares the original length of the data packet to be transmitted with the MTU of each instance, filtering out instances whose original length is less than the MTU. From these filtered instances, the gateway selects the instance with the highest header compression ratio as the target instance. The takeover gateway rewrites the header fields of the data packets to be transmitted according to the encapsulation rules corresponding to the target protocol conversion stack instance. Based on the network topology path updated after the link takeover is completed, it recalculates the number of hops from the source address to the destination address of the data packet, updates the routing information of the data packet according to the recalculated number of hops, and completes the encapsulation and forwarding of the data packet.

[0028] In this embodiment, by real-time acquisition of environmental energy characteristics and generation and broadcasting of energy state vectors, distributed sharing of energy state of each node in the gateway cluster is realized. When the node energy is lower than the forwarding threshold, the link takeover is completed through distributed election. Combined with the dynamic adjustment of protocol encapsulation format and routing hop count, dynamic reconstruction of the communication topology of the industrial gateway cluster is realized, avoiding physical disconnection of the entire data transmission link caused by single-point energy depletion.

[0029] refer to Figure 2 In a preferred embodiment, the integrated sensing module acquires the raw voltage and current signals from the multimodal sensor according to a preset sampling period. The preset sampling period can be configured through a local configuration file, ranging from 10ms to 1s, to adapt to the changing characteristics of different types of environmental energy. The integrated sensing module simultaneously activates the piezoelectric vibration sensor interface, the thermoelectric temperature difference sensor interface, and the radio frequency energy harvesting sensor interface. These three interfaces are electrically connected to their respective environmental energy harvesting units. Each interface is equipped with an independent signal conditioning circuit and an analog-to-digital conversion channel to ensure that different types of electrical signals can be acquired simultaneously. The piezoelectric vibration sensor interface acquires the first AC voltage signal generated by the mechanical vibration of the piezoelectric material under the mechanical vibration of industrial equipment. The piezoelectric material is attached to the surface of the industrial equipment's outer shell. The mechanical vibration generated during the operation of the industrial equipment causes the piezoelectric material to deform, converting mechanical energy into AC electrical energy through the positive piezoelectric effect. The first AC voltage signal acquired by the piezoelectric vibration sensor interface is the unprocessed raw AC voltage signal output by the piezoelectric material. The thermoelectric temperature difference sensor interface acquires a second DC voltage signal generated by the temperature difference on the surface of industrial equipment. The hot end of the thermoelectric temperature difference sensor is attached to the heating surface of the industrial equipment, while the cold end is exposed to the ambient air. The heat generated during the operation of the industrial equipment creates a temperature difference between the hot and cold ends, converting thermal energy into DC electrical energy through the Seebeck effect. The second DC voltage signal acquired by the thermoelectric temperature difference sensor interface is the unprocessed raw DC voltage signal output by the thermoelectric temperature difference sensor. The radio frequency (RF) energy harvesting sensor interface acquires a third pulsating voltage signal after the spatial RF signal has passed through a rectifier circuit. The antenna of the RF energy harvesting sensor receives RF signals transmitted by industrial wireless devices in space. The rectifier circuit converts the RF AC signal into a pulsating DC signal. The third pulsating voltage signal acquired by the RF energy harvesting sensor interface is the unprocessed raw pulsating voltage signal output by the rectifier circuit. The integrated sensing module splices the first AC voltage signal, the second DC voltage signal, and the third pulsating voltage signal into the original voltage signal according to the preset channel order, and splices the corresponding current signals of each channel into the original current signal according to the same channel order. The voltage signal and current signal of each channel are aligned according to the sampling timestamp to ensure that the spliced ​​original voltage signal and the original current signal are synchronized in the time dimension.

[0030] Table 2 Multimodal Sensor Channel Sampling Parameter Configuration Table 0 piezoelectric vibration sensor 1kHz 2 times -5V~+5V 4th order Synchronous triggering 1 Thermoelectric temperature difference sensor 100Hz 10 times 0V~10V 2nd order Synchronous triggering 2 Radio frequency energy harvesting sensor 500Hz 5 times 0V~5V 3rd order Synchronous triggering Table 2 is a multimodal sensor channel sampling parameter configuration table. This table defines the working parameters of the sampling channels corresponding to each type of sensor. The parameters of each channel are configured according to the output signal characteristics of the corresponding sensor to ensure that different types of environmental energy signals can be accurately and synchronously collected, providing reliable raw data for subsequent signal processing and energy calculation.

[0031] The integrated sensing module filters the spliced ​​raw voltage and current signals, extracting smooth voltage and current curves. The filtering process employs a moving average filtering algorithm to eliminate high-frequency noise interference and random fluctuations in the original signals, ensuring that the processed signal curves accurately reflect the actual changing trends of environmental energy. The calculation process of the moving average filtering is quantitatively described by the following formula: , in, This represents the smoothed voltage value at the nth sampling point; Let be the original voltage value at the nk-th sampling point; N is the length of the sliding window, which is a positive integer. The length of the sliding window is configured according to the sampling frequency of the corresponding channel. The higher the sampling frequency, the larger the length of the sliding window, ensuring that the filtered signal can eliminate noise interference while retaining the actual trend of signal change. For the original current signal, the same moving average filtering algorithm is used to process it, resulting in a smooth current curve. The number of sampling points and timestamps of the smooth current curve and the smooth voltage curve are perfectly aligned.

[0032] The smoothed voltage curve obtained after processing is input into a pre-built energy prediction model, which outputs the predicted power generation within a preset time window. The energy prediction model is an autoregressive model, and the model order is determined through pre-training using historical power generation data. The input to the model is the time series data of the smoothed voltage curve, and the output is the predicted output voltage for each sampling period within the preset time window. The prediction process of the autoregressive model is quantitatively described by the following formula: , in, Let p be the predicted output voltage at the h-th time in the future; p is the order of the autoregressive model, and its optimal value is determined by the Akaike information criterion. The autoregressive coefficients of the model are obtained in advance through training with historical voltage time series data; Let be the model residual at time t, which follows a normal distribution with a mean of 0, and be used to characterize the model's prediction error. For the smooth current curve, an autoregressive model with the same structure is used for prediction to obtain the predicted output current for each sampling period within a preset time window.

[0033] Based on the predicted output voltage and output current, the predicted power generation within the preset time window is calculated. The calculation process is quantified using the following formula: , in, The predicted power generation within the preset time window is expressed in joules; H is the total number of sampling points within the preset time window, determined by the ratio of the length of the preset time window to the sampling period. The predicted output current at the h-th time in the future is expressed in amperes. The duration of the sampling period is expressed in seconds.

[0034] The integrated sensing module reads the current voltage of the energy storage device and, combined with the pre-stored discharge characteristic curve of the energy storage device, calculates the current real-time energy storage margin. The discharge characteristic curve is a curve showing the correspondence between the terminal voltage of the energy storage device and the available power obtained through pre-testing of charge and discharge. A continuous capacitance-voltage characteristic function is obtained by fitting a cubic spline interpolation. The current terminal voltage is substituted into the integral formula to calculate the current real-time energy storage margin. The integrated sensing module acquires the historical average power consumption of the industrial smart gateway when it is in working state. The historical average power consumption is the average power consumption of the gateway over the past 30 consecutive working cycles. The power consumption of each working cycle is obtained by integrating the product of the real-time operating voltage and operating current of the gateway in that cycle. The historical average power consumption is multiplied by a preset time window to obtain the predicted consumption difference. Based on the calculated real-time energy storage margin and the predicted consumption difference, an energy state vector containing corresponding parameters is generated for subsequent broadcasting and state determination.

[0035] In this embodiment, comprehensive acquisition of energy characteristics of various environmental types is achieved through synchronous acquisition of multi-modal sensor interfaces and splicing of multi-channel signals. High-frequency noise interference and random fluctuations in the original signal are eliminated through moving average filtering. The energy prediction model built based on the autoregressive model achieves accurate prediction of future power generation. Combined with the quantitative calculation of the difference between real-time energy storage capacity and predicted consumption, an accurate and reliable data foundation is provided for the generation of energy state vector.

[0036] refer to Figure 3In another preferred embodiment, the industrial intelligent gateway encapsulates the generated energy state vector into a designated field of the distributed consensus message. This designated field is a variable-length field, with its starting offset and total length identified in the message header. The energy state vector within the field is stored in a preset little-endian byte order to ensure that gateways of different architectures can correctly parse the parameters within the vector. During the idle time slots of the distributed state machine, the industrial intelligent gateway uses a carrier sense multiple access (CSMA) mechanism to compete for the physical channel. The idle time slot is a pre-defined time interval in the distributed state machine dedicated to broadcasting node state information. Each time slot is 20ms long, matching the maximum message length that can be transmitted within the time slot. Within the time slot, the gateway continuously listens for the carrier signal of the physical channel. When it detects that there is no carrier signal on the channel for a continuous inter-frame interval, it determines that the channel is in an idle state. The inter-frame interval is 96µs, consistent with the standard specifications of industrial Ethernet. After successfully securing a physical channel, the industrial intelligent gateway sends a distributed consensus message encapsulated with energy state vectors to neighboring gateways. Upon receiving the message, each neighboring gateway first verifies its integrity using a frame check sequencer. If verification fails, the message is discarded. If verification succeeds, the gateway parses the specified fields of the message to obtain the energy state vectors and updates its locally maintained neighbor energy state table. The neighbor energy state table records the mapping between the node identifiers of neighboring gateways and their energy state vectors, as well as the update timestamp of each vector and the corresponding link's communication quality parameters. When the update timestamp of an entry exceeds a preset aging period, the corresponding entry is deleted from the neighbor energy state table. The preset aging period is three times the energy state vector broadcast cycle, ensuring that the locally maintained neighbor energy state table reflects the actual energy state of neighboring nodes in the network in real time.

[0037] Table 3. Neighbor Energy State Table Structure Definition Table Neighbor Node ID 6 bytes The unique physical address of the adjacent gateway serves as the unique primary key for the table entry. Added when a new neighboring gateway message is received, and deleted when the corresponding gateway has been offline for more than the aging period. Real-time energy storage capacity 4 bytes Real-time energy storage capacity reported by adjacent gateways, in millijoules. Update in real time when receiving the energy status broadcast message from the corresponding gateway. Predicted consumption difference 4 bytes The predicted consumption difference reported by adjacent gateways, in millijoules. Update in real time when receiving the energy status broadcast message from the corresponding gateway. Timestamp 8 bytes The corresponding timestamp for the generation of the energy state vector, in milliseconds. Synchronously update when receiving the energy status broadcast message from the corresponding gateway Link quality level 1 byte The communication link quality level between the local network and the corresponding adjacent gateway is calculated and divided into levels 0-3, with level 3 being the best. Each time a packet is received from the corresponding gateway, the link packet loss rate is recalculated and the rating is updated. Aging state 1 byte Indicate whether an item is in an aging state; 0 indicates valid, 1 indicates aging. If the timestamp exceeds the aging period, set it to 1; if it is not updated within a preset period after being set to 1, delete the corresponding entry. Table 3 defines the structure of the neighbor energy state table. This table specifies the storage format and update rules for the energy state information of neighboring gateways. All gateways in the same cluster maintain their local neighbor energy state tables according to the structure of this table, ensuring that the gateways can obtain the energy state and link quality information of neighboring nodes in real time, providing accurate and comprehensive data support for subsequent distributed elections.

[0038] refer to Figure 4The current edge node continuously compares the real-time energy storage margin in its local energy state vector with a preset forwarding threshold. The preset forwarding threshold is the total energy required for the gateway to complete a full election process, link migration process, and at least 10 data forwarding operations. The threshold can be modified through a local configuration file to adapt to gateway devices with different power consumption. When the real-time energy storage margin is less than the preset forwarding threshold, the current edge node triggers a distributed election algorithm, generating a random backoff timer locally. The backoff timer's duration is randomly generated between a preset minimum backoff duration and a preset maximum backoff duration to avoid channel conflicts caused by multiple nodes initiating elections simultaneously. The calculation process of the backoff timer's duration is quantitatively described by the following formula: , in, The duration of the backoff timer, in milliseconds; This is the time slot duration, in milliseconds, and its value is consistent with the time slot duration of the distributed state machine. It is a random integer between 0 and the contention window CW. The initial value of the contention window CW is 15. When the channel contention fails, CW is doubled according to the binary exponential backoff rule, and the maximum value is doubled to 1023 to ensure that the probability of collision can be reduced when the channel load is high.

[0039] During the backoff timer's countdown, the current edge node continuously monitors the physical channel. If it detects that the channel is occupied, it pauses the backoff timer until the channel becomes idle, at which point it resumes counting. When the backoff timer expires, the current edge node checks again whether the physical channel is idle. If the physical channel is idle, it sends an election request message containing its own node identifier to the adjacent gateway and switches the status flag of its local state machine from normal forwarding to election initiation. If the physical channel is occupied, it regenerates the backoff timer and waits for the next channel contention opportunity.

[0040] While the status flag is in the election initiation state, the current edge node only processes messages related to the election process, discarding received election request messages from other gateways to avoid process conflicts caused by multiple nodes initiating elections simultaneously. It also stops receiving new downlink data and only processes the caching and forwarding of already received data. After sending an election request message, the current edge node starts an election timeout timer. Before the timeout timer expires, it continuously receives election response messages from neighboring gateways. Upon receiving an election request message, a neighboring gateway first checks its local state machine status. If the local state machine is in a normal forwarding state, it appends the locally stored real-time energy reserve to the election response message and returns it to the edge node that initiated the election within a preset response time. If the local state machine is in the election initiation state or link migration state, it discards the election request message and does not return a response, avoiding nested election processes and conflicts.

[0041] After the election timeout timer expires, the current edge node stops receiving new election response messages, extracts the real-time energy storage margin value from all valid election response messages, sorts all the extracted real-time energy storage margin values ​​in descending order, and determines the adjacent gateway corresponding to the first real-time energy storage margin value in the sorted list as the takeover gateway. If the current edge node does not receive any valid election response messages, it switches the status flag of its local state machine to the sleep state, stops power supply to all non-essential hardware modules, waits for the preset sleep duration, and then wakes up the system again to collect environmental energy characteristics, determine the energy status, and trigger the election process. After determining the takeover gateway, the current edge node switches the status flag of its local state machine from the election initiation state to the link migration state, preparing to execute subsequent link migration operations.

[0042] In this embodiment, through the standardized encapsulation of distributed consensus messages and the channel contention of the carrier sense multiple access mechanism, reliable broadcasting of energy state vectors within the gateway cluster and real-time updating of neighbor energy state tables are achieved. By setting a random backoff timer and controlling the state switching of the state machine, channel conflicts and duplicate elections during the election process are effectively avoided. The distributed election algorithm enables the rapid and accurate determination of the takeover gateway, providing reliable control logic for the seamless migration of data relay links.

[0043] refer to Figure 4In another preferred embodiment, the current edge node sends a link migration command to the determined takeover gateway. The link migration command is sent in a point-to-point unicast manner to ensure that the command is only received by the target takeover gateway. The link migration command carries the routing table entry of the downlink node to be migrated and the head pointer and tail pointer of the forwarding buffer queue. The downlink node to be migrated is an industrial field terminal device or a lower-level gateway that uses the current edge node as the uplink next hop. The routing table entry includes the physical address, network address, quality of service level, data forwarding cycle, and link keep-alive duration of the downlink node. The forwarding buffer queue is stored in the random access memory of the current edge node and adopts a first-in-first-out circular queue structure. The head pointer of the queue points to the storage start address of the first data packet to be forwarded in the queue, and the tail pointer of the queue points to the storage end address of the last data packet to be forwarded in the queue. The total length of the forwarding buffer queue and the storage location of all data packets can be determined by the head pointer and the tail pointer.

[0044] Upon receiving the link migration instruction, the takeover gateway returns an acknowledgment message to the current edge node. Simultaneously, based on the queue head and tail pointers carried in the instruction, it retrieves all data packets from the current edge node's forwarding buffer queue using a point-to-point transmission protocol. This point-to-point transmission protocol employs a stop-and-wait protocol to ensure the reliability and order of data transmission. The takeover gateway determines the start and end sequence numbers of the data packets to be retrieved based on the queue head and tail pointers and sends a data retrieval request to the current edge node. This request includes the start and end sequence numbers of the requested data packets. Upon receiving the data retrieval request, the current edge node locates the corresponding cached data fragment in its local cache based on the start and end sequence numbers, calculates the cyclic redundancy check (CRC) code for each cached data fragment, appends the CRC code to the end of the corresponding cached data fragment, and sends it sequentially to the takeover gateway according to its sequence number.

[0045] The calculation process of the Cyclic Redundancy Check (CRC) code is quantitatively described by the following formula, using the standard CRC32 generator polynomial: , in, The 32-bit cyclic redundancy check code corresponding to data fragment D; For the binary polynomial corresponding to the data slice D, each byte in the data slice is converted into a binary sequence in the order of the most significant bit first, which constitutes the coefficients of the polynomial. To generate the polynomial, a standard CRC32 generator polynomial is used, satisfying... .

[0046] The takeover gateway receives cached data fragments sent by the current edge node. For each received cached data fragment, it recalculates the corresponding local cyclic redundancy check (CRC) code and compares it with the checksum carried at the end of the data fragment. If they match, the gateway writes the cached data fragment into its local transmit buffer and sends a receive acknowledgment message for that fragment to the current edge node. If they do not match, the gateway sends a retransmission request to the current edge node, requesting retransmission of the corresponding cached data fragment, until the received checksum matches the locally calculated checksum, or the preset maximum number of retransmissions is reached. If the correct cached data fragment is not received after reaching the maximum number of retransmissions, the takeover gateway discards the data fragment, records the corresponding error log, and continues to fetch the next cached data fragment, ensuring the continuity of the data retrieval process.

[0047] After retrieving all cached data fragments, the takeover gateway rebuilds the routing table entries of the downlink nodes to be migrated in its local memory. It writes all the retrieved data packets into its local send buffer in the original sequence number order and sends a topology update announcement to all downlink nodes to be migrated. The topology update announcement is broadcast to the local area network segment where the downlink nodes to be migrated are located. The announcement carries the network address of the downlink nodes to be migrated, the new uplink next-hop physical address, the update effective timestamp, and the checksum. After receiving the topology update announcement, the downlink nodes to be migrated verify that the network address in the announcement is consistent with their own network address. After verifying that the checksum is correct, they update their local routing table entries, change the next-hop address of the uplink data to the physical address of the takeover gateway, and send an acknowledgment message to the takeover gateway. After receiving confirmation messages from all downlink nodes to be migrated, the takeover gateway completes the takeover operation of the data relay link, switches the status flag of its local state machine to the takeover running state, and sends a migration completion message to the current edge node. Upon receiving the migration completion message, the current edge node stops data forwarding, switches the status flag of its local state machine to the sleep state, and enters a low-power operation mode.

[0048] refer to Figure 6After entering the takeover operation state, the takeover gateway parses its local configuration file to obtain a list of currently active protocol conversion stack instances. These instances are pre-deployed in the gateway's local system, and each instance corresponds to a commonly used industrial communication protocol, including but not limited to Modbus-RTU, Modbus-TCP, Profinet, EtherNet / IP, and MQTT. Each instance includes a protocol parsing module, a protocol encapsulation module, an address mapping module, and a header compression module. The activation status of each instance is configured through the local configuration file; only active instances can participate in the protocol conversion selection process. Based on the obtained list of protocol conversion stack instances, the takeover gateway queries a pre-stored local protocol overhead mapping table to obtain the header length, maximum transmission unit, and header compression ratio for different instance types, providing a quantitative basis for selecting the target instance.

[0049] Table 4 Protocol Conversion Stack Overhead Mapping Table Modbus-RTU 6 bytes 256 bytes 0.75 support Not supported Modbus-TCP 20 bytes 1500 bytes 0.17 support Not supported Profinet 14 bytes 1500 bytes 0.42 support support EtherNet / IP 22 bytes 1500 bytes 0.08 support support MQTT 2 bytes 256 bytes 0.92 support support Table 4 is a protocol conversion stack overhead mapping table. This table records the transmission overhead parameters and functional support of different protocol conversion stack instances. The header compression ratio in the table is calculated based on the 40-byte header length of the standard Ethernet IP protocol. By querying this table, the takeover gateway can quickly obtain the quantitative parameters of each protocol conversion stack instance, providing a basis for selecting the target protocol conversion stack instance and ensuring that the selected protocol stack can achieve the optimal header compression effect while meeting transmission requirements.

[0050] The gateway parses the original header of the data packet to be transmitted and calculates the original length of the data packet. The calculation process is quantified by the following formula: , in, The original length of the data packet to be transmitted, in bytes; The length of the original header of the data packet to be transmitted, in bytes; The payload data length of the data packet to be transmitted, in bytes.

[0051] The takeover gateway compares the original length of the data packet to be transmitted with the size of the maximum transmission unit (MTU) of each protocol conversion stack instance, and filters out protocol conversion stack instances whose original data packet length is less than the MTU. The matching process is quantified by the following formula: , in, The maximum transmission unit matching result for the j-th protocol conversion stack instance, where 1 indicates a match and 0 indicates a non-match; This represents the maximum transmission unit (MTU) for the j-th protocol conversion stack instance, measured in bytes.

[0052] The calculation process of the compression ratio of Baotou is quantitatively described by the following formula: , in, The packet header compression ratio corresponding to the j-th protocol conversion stack instance; This represents the header length of the j-th protocol conversion stack instance, in bytes.

[0053] The takeover gateway selects the instance with the highest packet header compression ratio from the filtered matching instances as the target protocol conversion stack instance. The selection process is quantified by the following formula: , in, S represents the selected target protocol conversion stack instance; S is the set of active protocol conversion stack instances.

[0054] The takeover gateway rewrites the header fields of the data packet to be transmitted according to the encapsulation rules corresponding to the target protocol conversion stack instance. Specifically, the takeover gateway parses the original header fields of the data packet to be transmitted, extracting the source address field, destination address field, and payload type field. The takeover gateway queries the address mapping dictionary corresponding to the target protocol conversion stack instance locally, converting the extracted source address field and destination address field into the gateway identifier and terminal identifier in the target protocol format. Based on the extracted payload type field, the takeover gateway matches the corresponding quality of service level identifier and data fragmentation flag in the protocol specification of the target protocol conversion stack instance. The takeover gateway writes the converted gateway identifier, terminal identifier, matched quality of service level identifier, and data fragmentation flag into the new header according to the field arrangement order of the target protocol format, replacing the original header fields of the data packet to be transmitted.

[0055] The length of the data packet after protocol conversion is quantified using the following formula: , in, The length of the data packet after protocol conversion, in bytes; The header length corresponding to the target protocol conversion stack instance is specified in bytes; the payload data length remains unchanged during the protocol conversion process to ensure the integrity of the data content.

[0056] Based on the updated network topology after link takeover, the takeover gateway recalculates the hop count of data packets from the source address to the destination address using the shortest path first algorithm. The calculation process is quantified by the following formula: , in, The minimum number of routing hops for the updated network topology path; P is the set of all available paths from the takeover gateway to the destination address; Let be the number of hops in the i-th link of the path; n is the total number of links in the path.

[0057] The takeover gateway updates the routing information of data packets according to the recalculated minimum hop count, forwards the packets with rewritten headers according to the updated routing information, and periodically collects the transmission status of the links to calculate the availability of the links after the migration. The calculation process is quantified by the following formula: , in, To ensure the availability of the links after migration; The number of data packets successfully forwarded by the takeover gateway; The total number of data packets to be forwarded received by the gateway to take over.

[0058] In this embodiment, lossless migration of the buffer queue to be forwarded is achieved by carrying a data retrieval request with a queue head pointer and a data packet sequence number, combined with data integrity verification using a cyclic redundancy check (CRC) code. Seamless takeover of the data relay link is achieved through local reconstruction of routing table entries and the issuance of topology update announcements. Adaptive adjustment of the data packet encapsulation format and optimization of the routing hop count are achieved through querying the protocol overhead mapping table and optimal selection of the target protocol conversion stack instance, ensuring the reliability and efficiency of data transmission after topology reconstruction.

Claims

1. A self-powered integrated sensing decentralized distributed industrial intelligent gateway control method, characterized in that, Includes the following steps: The integrated sensing module of the industrial smart gateway collects environmental energy characteristics in real time. Calculate the difference between the current real-time energy storage capacity and the predicted consumption based on the environmental energy characteristics, and generate an energy state vector containing the difference between the real-time energy storage capacity and the predicted consumption. The industrial smart gateway broadcasts the energy state vector in a pre-built distributed state machine; When the energy state vector indicates that the energy of the current edge node is lower than a preset forwarding threshold, the current edge node triggers a distributed election algorithm to determine the neighboring gateway with the highest remaining energy from the neighboring gateways that received the energy state vector as the takeover gateway. The takeover gateway takes over the data relay link of the current edge node; The takeover gateway dynamically adjusts the encapsulation format and routing hop count of the data packets to be transmitted in the data relay link based on its own maintained protocol conversion stack capabilities.

2. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 1, characterized in that, The steps of the integrated sensing module of the industrial intelligent gateway to collect environmental energy characteristics in real time, calculate the difference between the current real-time energy storage capacity and the predicted consumption based on the environmental energy characteristics, and generate an energy state vector containing the difference between the real-time energy storage capacity and the predicted consumption include: the integrated sensing module acquiring the original voltage signal and the original current signal of the multimodal sensor according to a preset sampling period. The original voltage signal and the original current signal are filtered to extract smooth voltage curves and smooth current curves; The smoothed voltage curve is input into a pre-built energy prediction model, which outputs the predicted power generation within a future preset time window. Read the current terminal voltage of the energy storage device and calculate the current real-time energy storage margin by combining it with the discharge characteristic curve of the energy storage device. The historical average power consumption of the industrial smart gateway when it is in working state is obtained, and the historical average power consumption is multiplied by the preset time window to obtain the predicted power consumption difference.

3. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 1, characterized in that, The step of the industrial intelligent gateway broadcasting the energy state vector in a pre-built distributed state machine includes: the industrial intelligent gateway encapsulating the energy state vector into a specified field of a distributed consensus message; During the idle time slots of the distributed state machine, the industrial intelligent gateway uses a carrier sense multiple access mechanism to compete for the physical channel. After successfully occupying the physical channel, the industrial smart gateway sends the distributed consensus message encapsulated with the energy state vector to the adjacent gateway. After receiving the distributed consensus message, the neighboring gateway parses the specified field to obtain the energy state vector and updates the neighbor energy state table maintained locally, wherein the neighbor energy state table records the mapping relationship between the node identifier of the neighboring gateway and the energy state vector.

4. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 1, characterized in that, The step of determining that when the energy state vector indicates that the energy of the current edge node is lower than the preset forwarding threshold, the current edge node triggers a distributed election algorithm to determine the neighboring gateway with the highest remaining energy from the neighboring gateways that received the energy state vector as the takeover gateway includes: the current edge node comparing the real-time energy storage margin with the preset forwarding threshold. When the real-time energy storage capacity is less than the preset forwarding threshold, the current edge node sends an election request message containing its own node identifier to the adjacent gateway. After receiving the election request message, the adjacent gateway appends the real-time energy storage capacity stored locally to the election response message and returns it. The current edge node extracts the real-time energy storage capacity from all received election response messages, and determines the adjacent gateway corresponding to the largest real-time energy storage capacity as the takeover gateway.

5. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 1, characterized in that, The step of the takeover gateway taking over the data relay link of the current edge node includes: the current edge node sending a link migration instruction to the takeover gateway, wherein the link migration instruction carries the routing table entry of the downlink node to be migrated and the queue head pointer of the forwarding cache queue; The takeover gateway retrieves data packets from the forwarding cache queue from the current edge node using a point-to-point transmission protocol based on the queue head pointer. The takeover gateway reconstructs the routing table entry of the downlink node to be migrated in its local memory and writes the pulled data packets into its local sending buffer. The takeover gateway sends a topology update announcement to the downlink node to be migrated, updating the uplink next-hop address of the downlink node to be migrated to the physical address of the takeover gateway.

6. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 1, characterized in that, The steps by which the takeover gateway dynamically adjusts the encapsulation format and routing hop count of the data packets to be transmitted in the data relay link based on its own maintained protocol conversion stack capabilities include: the takeover gateway parses the local configuration file to obtain a list of currently active protocol conversion stack instances; The takeover gateway queries a preset protocol overhead mapping table based on the protocol conversion stack instance list to obtain the packet header compression ratio and maximum transmission unit corresponding to different protocol stacks; The takeover gateway compares the original length of the data packet to be transmitted with the size of the maximum transmission unit, and selects a target protocol conversion stack instance that satisfies the condition that the original length is less than the maximum transmission unit and the packet header compression ratio is the largest. The takeover gateway rewrites the header field of the data packet to be transmitted according to the encapsulation rules corresponding to the target protocol conversion stack instance, and recalculates the routing hop count based on the updated network topology path.

7. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 2, characterized in that, The steps of the integrated sensing module acquiring the original voltage signal and original current signal of the multimodal sensor according to the preset sampling period include: the integrated sensing module synchronously activating the piezoelectric vibration sensor interface, the thermoelectric temperature difference sensor interface and the radio frequency energy harvesting sensor interface; The piezoelectric vibration sensor interface acquires the first AC voltage signal generated by the mechanical vibration of the piezoelectric material. The thermoelectric temperature difference sensor interface acquires the second DC voltage signal generated by the surface temperature difference of industrial equipment; The radio frequency energy harvesting sensor interface acquires the third pulsating voltage signal after the spatial radio frequency signal has passed through the rectifier circuit; The integrated sensing module splices the first AC voltage signal, the second DC voltage signal, and the third pulsating voltage signal into the original voltage signal, and splices the corresponding current signal into the original current signal.

8. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 4, characterized in that, The step of the current edge node sending an election request message containing its own node identifier to the adjacent gateway further includes: the current edge node generating a random backoff timer locally; After the backoff timer expires, the current edge node checks whether the physical channel is in an idle state. If the physical channel is in the idle state, the election request message is sent, and the status flag of the local state machine is switched from the normal forwarding state to the election initiation state. While the status flag is in the election initiation state, the current edge node discards the election request messages received from other gateways, and switches the status flag to the link migration state after receiving the election response message.

9. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 5, characterized in that, The step of the takeover gateway retrieving data packets from the forwarding buffer queue of the current edge node using the point-to-point transmission protocol based on the queue head pointer includes: the takeover gateway sending a data retrieval request to the current edge node based on the queue head pointer, wherein the data retrieval request includes the start sequence number and end sequence number of the data packets to be retrieved; The current edge node calculates the cyclic redundancy check code of the corresponding cached data fragment based on the start sequence number and the end sequence number, and appends the cyclic redundancy check code to the end of the cached data fragment before sending it. The takeover gateway receives the cached data fragment, recalculates the local cyclic redundancy check code of the cached data fragment, and writes the cached data fragment into the local transmission buffer after the comparison is consistent.

10. The self-powered integrated sensing decentralized distributed industrial intelligent gateway control method according to claim 6, characterized in that, The step of the takeover gateway rewriting the header field of the data packet to be transmitted according to the encapsulation rules corresponding to the target protocol conversion stack instance includes: the takeover gateway parsing the original header field of the data packet to be transmitted and extracting the source address field, destination address field and payload type field; The takeover gateway queries the address mapping dictionary corresponding to the target protocol conversion stack instance locally, and converts the source address field and the destination address field into a gateway identifier and a terminal identifier in the target protocol format; The takeover gateway matches the corresponding quality of service level identifier and data fragmentation flag in the protocol specification of the target protocol conversion stack instance based on the load type field. The takeover gateway writes the gateway identifier, the terminal identifier, the quality of service level identifier, and the data fragmentation flag into a new packet header according to the field arrangement order of the target protocol format, replacing the original packet header fields.

Images