A Distributed Humidity Coordination Control Method and System for Multiple Dehumidifier Clusters
By using a distributed self-organizing network and inter-node collaborative control method, the problems of uneven humidity control and high energy consumption in multi-moisture box clusters are solved, achieving efficient and reliable humidity management, which is suitable for low-cost applications of small and medium-sized moisture box clusters.
Patent Information
- Application Number
- CN202611461425.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies cannot achieve humidity information sharing, balance calculation, collaborative decision-making, and closed-loop feedback correction in a multi-humidifier cluster without a central server. This results in uneven humidity control, high energy consumption, poor reliability, and sensor drift and communication failures affecting control accuracy.
A distributed self-organizing network protocol is adopted to form a decentralized or quasi-centralized self-organizing network of dehumidifier clusters. The nodes exchange humidity status messages through communication, calculate the cluster humidity balance index, and drive the humidity control actuator based on local collaborative control decisions. Combined with sensor drift correction and link redundancy mechanisms, closed-loop feedback and fault-tolerant operation are achieved.
It achieves rapid humidity convergence, high uniformity, low energy consumption, and high reliability within the dehumidification box cluster, reducing operation and maintenance costs while ensuring the accuracy of humidity data and the continuity of control.
Smart Images

Figure CN123114104A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of distributed control technology, specifically to a distributed humidity collaborative control method and system for multiple dehumidifier clusters. Background Technology
[0002] Dehumidifiers (electronic dehumidifiers, drying cabinets, etc.) are key equipment in the storage of humidity-sensitive items such as optical instruments, precision electronic components, semiconductor chips, medical reagents, and archival artifacts. They use built-in humidity sensors to collect real-time humidity data and drive dehumidification modules (such as molecular sieve rotors, semiconductor condensers, nitrogen purging, etc.) or humidification modules to maintain the humidity within a set target range, thus inhibiting mold, oxidation, corrosion, and performance degradation of stored items. With the rapid development of industries such as integrated circuits, optoelectronic manufacturing, biomedicine, and scientific research, a single dehumidifier is no longer sufficient to meet the needs of large-scale storage. In actual production and research sites, dozens or even hundreds of dehumidifiers are typically deployed in the same workshop, warehouse, or laboratory to form a dehumidifier cluster for unified management.
[0003] In existing technologies, humidity management for clusters of multiple desiccant boxes mainly involves the following three types of technical solutions and their shortcomings:
[0004] The first type of solution involves independent closed-loop control for each dehumidifier. This means that each dehumidifier relies solely on its own humidity sensor and local controller to independently collect and regulate its own humidity, with no information exchange between units. This solution is simple in structure and low in cost, but it has obvious drawbacks: First, the moisture-proof boxes are not aware of each other. When moisture circulates within the cluster due to opening and closing doors, environmental seepage, or dehumidification module discharge, a "tug-of-war" phenomenon of humidity between adjacent boxes can easily occur. That is, the moisture discharged by box A causes the humidity in box B to rise, and box B then starts dehumidifying, which in turn affects box A. This causes frequent start-stop of the actuators, a significant increase in energy consumption, and the humidity of each box in the cluster is difficult to converge to its target range for a long time. Second, some boxes with strong dehumidification capacity are idle after completing their own control, while adjacent boxes with higher humidity are running at full capacity, and the overall control capacity of the cluster cannot be used evenly. Third, the single-box control lacks a comprehensive consideration of differentiated factors such as the humidity sensitivity level of the stored items, the box volume, and the historical response rate, making it difficult to achieve on-demand allocation and key protection.
[0005] The second type of solution is centralized monitoring and scheduling based on a central server. This involves connecting multiple dehumidifiers to a host computer or cloud platform via RS-485 bus, Ethernet, or wireless means. The central server then aggregates the humidity data from each cabinet, calculates control strategies, and issues control commands. While this solution addresses the information silo problem within the cluster to some extent, it still suffers from the following shortcomings: First, the system architecture heavily relies on the central node. If the central server, gateway, or main communication link fails, the entire cluster's coordinated control becomes paralyzed, posing a single point of failure risk and insufficient reliability. Second, all humidity data must be uploaded to the center, and all control commands must be issued from the center, resulting in heavy communication link load and high control latency. This makes it difficult to meet the need for rapid response in scenarios of sudden humidity changes, and the computational and communication pressure on the central node increases dramatically as the cluster size expands, limiting scalability. Third, the centralized solution requires dedicated server deployment, configuration software, and continuous network maintenance, leading to high hardware investment and maintenance costs, making it less economical for small to medium-sized dehumidifier clusters. Fourth, the centralized solution typically remains at a coarse-grained control level of "data acquisition + threshold alarm + remote switching," failing to achieve true collaborative decision-making based on cluster humidity balance among the individual dehumidifiers, resulting in a relatively primitive control strategy.
[0006] The third type of solution is a simple network data acquisition scheme. This involves networking multiple dehumidifier cabinets using dedicated software, with a single computer monitoring the operating status of each cabinet, plotting curves, and reading historical records. However, it is essentially still a data aggregation and visualization tool and lacks the ability for peer-to-peer communication between nodes, calculation of cluster humidity balance, and generation and execution of collaborative control commands between nodes. Therefore, it cannot fundamentally solve the problems of humidity coupling and energy peak shifting among multiple cabinets.
[0007] In addition, existing dehumidifiers also face problems such as reading distortion caused by humidity sensor drift, data interruption caused by single communication link failure, and control blind spots caused by node failure during long-term operation, which further restrict the ability of dehumidifier clusters to operate stably, accurately, and with low energy consumption in the long term.
[0008] In summary, current technologies lack a distributed humidity collaborative control method and system that can achieve humidity information sharing, balance calculation, collaborative decision-making, and closed-loop feedback correction at the cluster level of dehumidifiers without relying on a central server, while also possessing node fault tolerance and link redundancy capabilities. Therefore, a new technical solution is urgently needed to overcome the shortcomings of the existing technologies. Summary of the Invention
[0009] To overcome the shortcomings of existing technologies, the present invention aims to provide a distributed humidity collaborative control method and system for multiple dehumidifier clusters, so as to realize humidity information sharing, cluster humidity balance calculation, inter-node collaborative decision-making, closed-loop feedback correction and fault-tolerant operation of the dehumidifier cluster under the condition of no central server, thereby achieving the control effect of rapid cluster humidity convergence, balanced and accurate, peak-shifting energy saving, high reliability and low cost.
[0010] To solve the above problems, the technical solution adopted by the present invention is as follows:
[0011] A first aspect of the present invention provides a distributed humidity collaborative control method for a cluster of multiple dehumidifier boxes, comprising the following steps:
[0012] S1. Each moisture-proof box node completes node registration and identity assignment through a distributed self-organizing network protocol, forming a decentralized or quasi-centralized self-organizing network of moisture-proof box clusters.
[0013] S2. Each moisture-proof box node periodically collects its own humidity data based on a unified clock reference, and broadcasts and / or sends humidity status messages containing node identifiers, humidity data and timestamps to adjacent nodes through inter-node communication links.
[0014] S3. Each moisture-proof box node calculates the cluster humidity balance index based on its own humidity data and the humidity status messages received from neighboring nodes, and generates a local collaborative control decision based on the cluster humidity balance index and the humidity deviation of its own box.
[0015] S4. Each moisture-proof box node drives its humidity control actuator to perform corresponding humidity control actions according to the control instructions in the local collaborative control decision, and synchronizes the control execution results as feedback information to adjacent nodes.
[0016] S5. Each dehumidifier node dynamically corrects the control parameters in the local collaborative control decision based on the feedback information, forming a closed-loop feedback distributed humidity collaborative control, until the humidity values of each dehumidifier in the cluster converge to their respective target humidity range.
[0017] Preferably, the unified clock reference in step S2 is established through a distributed clock synchronization mechanism, which includes at least one of master node election, clock offset compensation, and periodic calibration; when the number of dehumidifier nodes in the cluster changes, network topology reconstruction and clock reference resynchronization are triggered.
[0018] Preferably, the humidity data in step S2 undergoes data compensation processing before transmission, including sensor drift correction, temperature-humidity cross-sensitivity compensation, and historical data interpolation compensation under communication packet loss conditions.
[0019] Preferably, the cluster humidity balance index in step S3 is calculated iteratively using a weighted average or consensus algorithm based on the humidity value of the current box, the humidity values of adjacent nodes, and the priority weight of each node; the priority weight is dynamically allocated based on the humidity sensitivity level of the items stored in the dehumidifying box, the box volume, and / or the historical control response rate.
[0020] Preferably, the generation of local collaborative control decision in step S3 specifically includes: when the humidity deviation of this box is greater than the first preset threshold, generating an independent rapid control instruction for this box; when the humidity deviation of this box is less than or equal to the first preset threshold and the cluster humidity balance index exceeds the second preset threshold, generating a node collaborative linkage control instruction and sending it to the scheduled adjacent nodes, so that the adjacent nodes can perform collaborative dehumidification or humidification operations accordingly.
[0021] Preferably, the humidity control action performed in step S4 includes:
[0022] S41. Obtain the control instructions to be executed and determine their type; the control instructions include independent rapid control instructions for this box and inter-node collaborative control instructions received from adjacent nodes.
[0023] S42. Match the target execution mode according to the instruction type: the independent rapid control instruction of this box matches the rapid control mode; the inter-node coordinated control instruction matches the coordinated control mode. Its start and stop sequence is determined according to the coordinated timing information carried by the instruction, so that the control actions of this node and the scheduled adjacent node are coordinated with each other in timing.
[0024] S43. Output the corresponding drive signal to the humidity control actuator of this box according to the target execution mode, and perform humidity control action with the corresponding start-stop sequence and power level.
[0025] S44. After the humidity value of this chamber enters the target humidity range, switch to maintenance mode and periodically output fine-tuning drive signals with a low duty cycle to prevent the humidity value of this chamber from exceeding the target humidity range.
[0026] Preferably, the above method further includes a node failure tolerance step: when any dehumidifier node neither receives a humidity status message from the adjacent node nor receives a link layer response from the adjacent node through any candidate communication link within a preset time, the adjacent node is determined to be faulty, the node switches to local independent control mode, and reports the faulty node information, triggering network topology update and route reconstruction; wherein, the node failure tolerance step and the fault switching of a single communication link are executed independently of each other.
[0027] Preferably, the dynamic correction of the control parameters in the local collaborative control decision-making in step S5 is based on the energy consumption optimization objective function. The energy consumption optimization objective function takes the total energy consumption of the cluster and / or the control convergence time as the optimization target. By adjusting the start-stop sequence and power level of the humidity control actuators of each node at the cluster level, the control actions of multiple nodes are staggered.
[0028] Preferably, sending a humidity status message in step S2 includes:
[0029] S21. Evaluate the link quality of each candidate communication link to obtain link quality parameters including at least one of received signal strength, packet loss rate and transmission delay.
[0030] S22. Determine the target communication link based on the link quality parameters and the message priority of the message to be sent; wherein the message priority of the humidity status message carrying abnormal alarm information is higher than that of the regular periodic humidity status message.
[0031] S23. Encapsulate the humidity status message in a unified data frame format that is independent of the underlying communication protocol of each candidate link and send it so that the same message does not need to be reframed when switching between different candidate links.
[0032] S24. Monitor the transmission status. If no acknowledgment is received within a preset time or the link quality parameters deteriorate to below a preset threshold, switch to the backup link to retransmit the encapsulated data frame, and update the link quality parameters of each candidate communication link based on the transmission result.
[0033] A second aspect of the present invention provides a distributed humidity collaborative control system for a cluster of multiple dehumidifier boxes, comprising:
[0034] Multiple dehumidifier cabinet nodes, each node including cabinet body, humidity sensor installed inside the cabinet body, local controller electrically connected to humidity sensor, and humidity control actuator electrically connected to local controller;
[0035] The communication network allows each dehumidifier node to access the network through a communication unit located within its node, forming a distributed and interconnected dehumidifier cluster network.
[0036] The distributed collaborative control module is deployed in each local controller. It is used to enable each dehumidifier node to execute the above-mentioned distributed humidity collaborative control method based on the humidity status information exchanged between the humidity data of its own chamber and the humidity status information of the adjacent nodes, without the need for a central server.
[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0038] Eliminating single-point-of-failure risks significantly improves system reliability. This invention employs a decentralized or quasi-centralized distributed self-organizing network architecture. The collaborative control module is deployed in the local controller of each dehumidifier node, enabling cluster-wide humidity collaborative control without a central server. The master node is elected and can be automatically re-elected; the failure of any node will not paralyze the entire cluster control system, overcoming the single-point-of-failure defect of centralized solutions where a central server failure leads to the paralysis of the entire cluster control. Simultaneously, node failure tolerance steps and single-communication link failure switching are executed independently, forming a dual fault-tolerance system of link redundancy and node redundancy. Under abnormal operating conditions, the cluster can automatically degrade to a local independent control mode and trigger topology reconfiguration, ensuring control continuity.
[0039] This invention eliminates the "tug-of-war" of humidity between chambers, resulting in rapid humidity convergence and high uniformity in the cluster. It involves the periodic exchange of humidity status messages between nodes, containing node identifiers, humidity data, and timestamps. Each node iteratively calculates the cluster humidity uniformity index based on its own humidity and that of neighboring nodes. A dual-threshold decision mechanism (deviating above the first threshold triggers independent rapid control; deviation below the first threshold but uniformity exceeding the second threshold triggers coordinated inter-node control) coordinates the dehumidification / humidification actions and start / stop sequences of each chamber at the cluster level. This ensures that the control actions of adjacent nodes are synchronized in time, avoiding frequent start / stop of actuators caused by mutual humidity interference between adjacent chambers. This significantly accelerates the convergence of cluster humidity towards their respective target ranges, improving the overall humidity uniformity of the cluster.
[0040] This invention achieves differentiated on-demand control and cluster-based peak-shaving energy saving, significantly reducing energy consumption. The priority weights of this invention are dynamically allocated based on the humidity sensitivity level of stored items, cabinet volume, and / or historical control response rate, giving greater authority to cabinets containing humidity-sensitive items, thus ensuring key items receive priority protection. Simultaneously, based on an objective function that optimizes total cluster energy consumption and / or control convergence time, the start-stop sequence and power levels of each node's actuators are adjusted at the cluster level, enabling peak-shaving operations of multiple nodes, smoothing out power supply peaks, reducing total cluster energy consumption, and achieving both economic efficiency and grid friendliness.
[0041] This invention eliminates the need for a central server, resulting in low deployment costs and high scalability. It eliminates the need for dedicated server deployment, configuration software, and continuous network maintenance. Each node can be plugged into and automatically connects to the self-organizing network, completing registration, clock synchronization, and topology construction. When the number of dehumidifier nodes within the cluster changes, network topology reconstruction and clock resynchronization are automatically triggered. Adding or removing dehumidifiers does not require redesigning the system architecture, and the cluster size can be flexibly scaled, making it particularly suitable for low-cost networking applications of small to medium-sized dehumidifier clusters.
[0042] The humidity data is accurate and reliable, providing a solid foundation for regulation. Before data transmission, this invention performs sensor drift correction, temperature-humidity cross-sensitivity compensation, and historical data interpolation compensation under communication packet loss conditions on the humidity data. Combined with the humidity sensor drift self-calibration unit, which performs zero-point and range correction based on a preset period or the readings of a reference node within the cluster, this effectively overcomes the reading distortion problems caused by long-term sensor drift and temperature cross-sensitivity. This ensures the comparability and consistency of humidity readings across multiple nodes within the cluster, providing a reliable data foundation for balance calculations and collaborative decision-making. The local data caching unit temporarily stores humidity data and regulation execution records during communication interruptions and synchronizes them in batches after communication is restored, ensuring the temporal continuity of the cluster's humidity status data.
[0043] The communication links are flexible and reliable, and message transmission is efficient and seamless. This invention supports hybrid networking of multiple communication links such as LoRa, Ethernet, CAN, and RS-485. It dynamically selects the target communication link by combining link quality assessment (received signal strength, packet loss rate, and transmission delay) with message priority, and prioritizes the transmission of abnormal alarm messages. The humidity status message is encapsulated in a unified data frame format independent of the underlying communication protocol. There is no need to reframe when switching links, realizing seamless switching and fault retransmission between multiple links, and balancing real-time performance, reliability, and network capacity.
[0044] The system offers precise control and maintains low power consumption and low noise during operation. This invention employs multiple execution modes, including rapid control, coordinated linkage, and maintenance, based on the command type and deviation magnitude. It supports independent start / stop and multi-level power adjustment. Upon entering the target humidity range, it switches to maintenance mode and periodically outputs fine-tuning drive signals with a low duty cycle. This ensures humidity control accuracy while minimizing energy consumption and operating noise, thus extending the lifespan of the actuator.
[0045] Facilitates cluster-level operation and maintenance management. The distributed collaborative control module of this invention is embedded in the local controller in firmware form, supporting remote firmware upgrades and batch distribution of control strategy parameters via communication networks. Operation and maintenance personnel can complete the software upgrade and parameter adjustment of the entire cluster without performing on-site operations one by one, significantly reducing the operation and maintenance costs of large-scale dehumidifier clusters.
[0046] In summary, this invention overcomes the inherent defects of existing independent control schemes and centralized schemes by organically combining distributed self-organizing networks, collaborative decision-making on cluster humidity balance, priority-differentiated allocation, energy consumption optimization and off-peak operation, and multiple fault-tolerant mechanisms. It achieves high balance, fast convergence, low energy consumption, high reliability and low cost in humidity control of dehumidifier clusters, and has good practical value and promotion prospects.
[0047] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. Attached Figure Description
[0048] Figure 1 This is an overall flowchart of the distributed humidity collaborative control method provided in an embodiment of the present invention;
[0049] Figure 2 A flowchart of the control execution process provided in this embodiment of the invention;
[0050] Figure 3 This is a timing diagram of inter-node collaborative peak-shaving provided in an embodiment of the present invention;
[0051] Figure 4 This is a schematic diagram of the maintenance mode fine-tuning provided in an embodiment of the present invention;
[0052] Figure 5 This is a structural block diagram of a distributed humidity collaborative control system provided in an embodiment of the present invention.
[0053] Reference numerals: 100, Dry box node; 110, Box body; 120, Humidity sensor; 121, Drift self-calibration unit; 130, Local controller; 300, Distributed collaborative control module; 310, Humidity equalization calculation submodule; 320, Collaborative instruction generation submodule; 131, Local data cache unit; 140, Humidity control actuator; 141, Dehumidification module; 142, Humidification module; 150, Communication unit; 200, Communication network. Detailed Implementation
[0054] The technical solution of the present invention will be further described clearly and completely below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0055] Example 1: System Overall Architecture
[0056] like Figure 5 As shown, this embodiment provides a distributed humidity collaborative control system for multiple dehumidifier clusters, including multiple dehumidifier nodes 100, a communication network 200, and a distributed collaborative control module 300.
[0057] Each moisture-proof cabinet node 100 includes a cabinet body 110, a humidity sensor 120 disposed within the cabinet body 110, a local controller 130 electrically connected to the humidity sensor 120, and a humidity control actuator 140 electrically connected to the local controller 130. Wherein:
[0058] The humidity sensor 120 employs a capacitive humidity sensor (such as SHT30, HTU21D, etc., with a measurement range of 0–100%RH and an accuracy of ±2%RH) or a resistive humidity sensor. To eliminate zero-point drift and range drift caused by long-term operation, the humidity sensor 120 is equipped with a drift self-calibration unit 121. This drift self-calibration unit 121 performs zero-point and range calibration once based on a preset period (e.g., every 720 hours). Simultaneously, a stable, high-accuracy dehumidifier node 100 within the cluster can be designated as a reference node. The remaining nodes periodically receive humidity readings from the reference node and perform mutual calibration based on the relative deviation, thereby ensuring the comparability and consistency of humidity readings from multiple nodes within the cluster without manual intervention.
[0059] The humidity control actuator 140 includes a dehumidification module 141 and a humidification module 142. The dehumidification module 141 can be one or more of a molecular sieve rotary dehumidification unit, a semiconductor condensation dehumidification unit (based on the Peltier effect), or a nitrogen purge valve assembly; the humidification module 142 can be an ultrasonic humidification unit or an evaporative humidification unit. The dehumidification module 141 and the humidification module 142 are each electrically connected to the local controller 130 via independent drive circuits, enabling the local controller 130 to independently control the start / stop of the dehumidification and humidification branches and adjust multiple power levels (e.g., low, medium, and high). The type of humidity control actuator 140 configured in the multiple dehumidification box nodes 100 within the cluster is determined according to their respective target humidity ranges and is allowed to be different from each other. For example, a dehumidifying cabinet for storing optical lenses has a target humidity range of 40% to 50% RH, and can be equipped with a semiconductor condensation dehumidification unit; a dehumidifying cabinet for storing precision components has a target humidity range of 10% to 30% RH, and can be equipped with a molecular sieve rotor dehumidification unit to obtain a lower humidity limit; while a dehumidifying cabinet used in arid regions can be equipped with only a humidification module 142 or a humidification module 142 with a larger power ratio.
[0060] The local controller 130 is an embedded microcontroller (such as the STM32F4 series or ESP32 series) or an edge computing processor (such as the RK3568). The local controller 130 has a built-in local data cache unit 131 (which can be implemented using FRAM or EEPROM) to temporarily store the humidity data and control execution records of the box during communication interruption, and to synchronize the temporarily stored data to adjacent nodes in batches after communication is restored, so as to ensure the time continuity of the cluster humidity status data.
[0061] The communication network 200 includes a wireless ad-hoc network subnet and / or a wired bus subnet. Each dehumidifier node 100 connects to the communication network 200 through a communication unit 150 located within that node. The communication unit 150 integrates at least one of a LoRa RF front-end, an Ethernet physical layer interface, a CAN transceiver, or an RS-485 transceiver. This embodiment preferably employs a dual-link redundancy scheme with a LoRa wireless ad-hoc network as the primary link and an RS-485 wired bus as the backup link: LoRa links offer flexible cabling and are suitable for cross-rack networking within a data center; RS-485 links have strong anti-interference capabilities and low latency. The communication unit 150 is used to automatically switch to the backup communication link and perform data retransmission when a failure of the current communication link is detected.
[0062] The distributed collaborative control module 300 is deployed in each local controller 130 and is fixed in the storage medium of the local controller 130 in the form of firmware. It is used to enable each dehumidifier node 100 to execute the distributed humidity collaborative control method described in this invention based on the humidity status information exchanged between the humidity data of its own chamber and the adjacent nodes, without the need for a central server.
[0063] Furthermore, the distributed collaborative control module 300 includes a humidity balance calculation submodule 310 and a collaborative instruction generation submodule 320. The humidity balance calculation submodule 310 iteratively calculates the cluster humidity balance index based on a weighted average or consensus algorithm; the collaborative instruction generation submodule 320 generates independent rapid control instructions for the local unit or inter-node collaborative control instructions based on the humidity deviation between the cluster humidity balance index and the local unit's humidity, and sends the inter-node collaborative control instructions to adjacent nodes. In addition, the local controller 130 supports remote firmware upgrades and batch distribution of control strategy parameters via the communication network 200. For example, the maintenance terminal sends a firmware upgrade package to any node in the cluster. After receiving the package, the node verifies and writes it block by block, and then forwards it to the remaining nodes level by level through the internal cluster link to achieve firmware upgrades and parameter updates for the entire cluster without the need for on-site operation.
[0064] Example 2: Distributed Humidity Coordinated Control Method
[0065] like Figure 1 As shown, this embodiment provides a distributed humidity collaborative control method for multiple dehumidifier clusters, applied to the system described in Embodiment 1, and includes the following steps:
[0066] Step S1: Node Registration and Identity Assignment
[0067] Each of the 100 moisture-proof box nodes in the cluster completes node registration and identity assignment through a distributed self-organizing network protocol, forming a decentralized or quasi-centralized self-organizing network for the moisture-proof box cluster.
[0068] Specifically, after each dehumidifier node 100 is powered on, it enters the network discovery state and sends a node discovery broadcast frame via a LoRa RF front-end on a preset channel using a random backoff method. The frame carries the node's MAC address, hardware version number, initial priority weight, and target humidity range. Nodes receiving the discovery broadcast frame record neighbor information and reply with a response frame. After a preset discovery window (e.g., 60 seconds) ends, each node elects a quasi-central coordinating node (i.e., the master node) based on the number of neighbor responses and link quality score. The master node is responsible for assigning unique node identifiers (e.g., 0x01~0xFF) to each node in the cluster and maintaining the cluster member table. The master node election uses a weighted scoring rule of "link quality score + remaining node resources," with the highest scorer elected. If the master node is out of contact for more than a preset duration, the remaining nodes automatically re-initiate the election, ensuring the network's self-healing capability under decentralized conditions. The cluster member table is periodically flooded to all nodes, allowing each node to determine its own set of neighboring nodes.
[0069] Step S2: Periodic acquisition of humidity data and exchange of status messages
[0070] Each moisture-proof box node 100 periodically collects its own humidity data based on a unified clock reference, and broadcasts and / or sends humidity status messages containing node identifiers, humidity data and timestamps to adjacent nodes through inter-node communication links.
[0071] S21: Establishment of a unified clock reference
[0072] The unified clock reference is established through a distributed clock synchronization mechanism, which includes at least one of master node election, clock offset compensation, and periodic calibration. This embodiment specifically employs the following method: the master node periodically (e.g., every 10 seconds) broadcasts a clock synchronization frame, carrying the master node's current timestamp Tm within the frame; slave nodes record their local clock at the local time Tl upon receiving the frame, calculating the clock offset ΔT = Tm − Tl − Tprop, where Tprop is the propagation delay estimated based on the LoRa air transmission time; slave nodes then compensate for the offset of their local clocks accordingly. Furthermore, each node periodically performs drift rate fitting (least squares method) based on the arrival time sequence of the synchronization frame, periodically calibrating the frequency drift of its local crystal oscillator, ensuring that the clock deviation between any two nodes within the cluster is controlled within ±10ms.
[0073] When the number of 100 dehumidifier nodes in the cluster changes (e.g., a new node is added, an old node is powered off or fails), network topology reconstruction and clock base resynchronization are triggered: the master node updates the cluster member table and re-floods the broadcast; if the master node itself fails, a new master node is elected and the clock base is re-established according to the rules described in step S1.
[0074] S22: Humidity data compensation processing
[0075] Humidity data undergoes data compensation processing before transmission. This data compensation processing includes:
[0076] (a) Sensor drift correction: Based on the most recent zero-point / range correction coefficient recorded by the drift self-calibration unit 121, the original humidity reading is linearly corrected, i.e., H_correct = a·H_raw + b, where a and b are the proportional coefficient and offset obtained from the most recent correction, H_raw is the original humidity reading, and H_correct is the humidity reading after linear correction.
[0077] (b) Temperature-humidity cross-sensitivity compensation: Due to the temperature dependence of the capacitive humidity sensor, the temperature sensor reading T inside the chamber (which can be integrated with the humidity sensor 120 on the same chip) is corrected according to the predetermined compensation polynomial H_comp = H_correct + f(T), where H_comp is the humidity reading after compensation correction, and f(T) is calibrated by the sensor manufacturer's datasheet or obtained by fitting the factory calibration experiment.
[0078] (c) Historical data interpolation compensation under communication packet loss conditions: When this node fails to receive the humidity status message of a neighboring node in time, it uses linear interpolation or exponential weighted moving average extrapolation to estimate the current humidity value based on the humidity data sequence that the neighboring node has successfully received in the most recent N times (e.g., N=5), to fill the data gap and avoid the distortion of the balance calculation due to a single packet loss.
[0079] S23: Link Quality Assessment and Target Communication Link Determination
[0080] Specifically, the following steps are included:
[0081] Link quality is evaluated for all available candidate communication links (LoRa link, RS-485 link, Ethernet link, etc.) at this node to obtain link quality parameters for each candidate communication link. These parameters include at least one of Received Signal Strength Indicator (RSSI), packet loss rate, and transmission delay. In this embodiment, a comprehensive score S = w1·Score_RSSI + w2·Score_loss + w3·Score_delay is used to quantize and rank each link (where w1+w2+w3=1, preferably w1=0.3, w2=0.4, w3=0.3).
[0082] Based on the link quality parameters and the message priority of the humidity status message to be sent, the target communication link for this transmission is determined from among the candidate communication links. Humidity status messages carrying abnormal alarm information (e.g., humidity over-limit alarms, sensor fault alarms) have a higher priority than regular periodic humidity status messages. For high-priority alarm messages, the link with the shortest transmission delay is selected first, and cross-link concurrent redundant transmission is allowed. For regular periodic messages, the link with the highest comprehensive score and sufficient remaining channel occupancy is selected to balance real-time performance and network capacity.
[0083] Humidity status messages are encapsulated in a unified data frame format and sent to adjacent nodes via the target communication link. This unified data frame format is independent of the underlying communication protocol of each candidate communication link and includes: a frame header (2-byte synchronization word), frame type (1 byte), source node ID (1 byte), timestamp (4 bytes), humidity payload (4 bytes, fixed-point representation), temperature payload (2 bytes), flag bits (1 byte, used to mark alarms / coordination requests / feedback, etc.), and CRC checksum (2 bytes). Because a unified intermediate encapsulation format is used, the same humidity status message does not need to be reframed when switching between different candidate communication links; only the underlying physical encapsulation needs to be changed for seamless forwarding, significantly reducing the software complexity of link switching.
[0084] The system monitors the transmission status of the target communication link. If no acknowledgment (ACK) response is received from the adjacent node within a preset time (e.g., 500ms), or if the link quality parameters of the target communication link deteriorate to below a preset threshold (e.g., packet loss rate exceeds 20% or RSSI is below -110dBm), the system switches to the backup communication link to retransmit the encapsulated data frame. The system also updates the link quality parameters of each candidate communication link based on the transmission results (e.g., using exponential moving average to update packet loss statistics) for subsequent transmissions to determine the target communication link, thus forming a continuous optimization closed loop for link selection.
[0085] Step S3: Cluster humidity balance calculation and local collaborative control decision-making
[0086] Each moisture-proof box node 100 calculates the cluster humidity balance index based on its own humidity data and the humidity status messages received from neighboring nodes, and generates local collaborative control decisions based on the cluster humidity balance index and the humidity deviation of its own box.
[0087] (1) Calculation of the cluster humidity uniformity index
[0088] Each node iteratively calculates the cluster humidity balance using a weighted average or consensus algorithm based on its own humidity value, the humidity values of neighboring nodes, and the priority weight of each node. This embodiment provides two optional implementations:
[0089] Method 1 (Weighted Average Method): When node i is calculated in the kth round, it is weighted according to the priority weight wi of each node based on its own humidity Hi(k) and the humidity set {Hj(k)} of the neighboring nodes that has been received, and the estimated value of the cluster average humidity is calculated.
[0090] The cluster humidity balance index Q is then calculated, which is the sum of the weighted absolute deviations of each node's humidity relative to the estimated average humidity of the cluster. A smaller Q indicates a more balanced overall humidity in the cluster. The priority weights are dynamically allocated based on the humidity sensitivity level of the items stored in the dehumidifier, the cabinet volume, and / or the historical control response rate. For example, the weight of nodes storing the most humidity-sensitive items (such as special optical components) is set to 3.0, the weight of nodes storing ordinary electronic components is set to 1.5, and the weight of nodes storing general items is set to 1.0; the weight of nodes with larger cabinet volumes and higher thermal inertia is appropriately increased; the weight of nodes with fast historical control response rates (large humidity changes per unit time) is also appropriately increased to give them greater weight in the balance calculation. The weights are recalculated according to a preset period (e.g., 24 hours) based on item replacement and historical statistical data.
[0091] Method 2 (Iterative Consensus Algorithm): Each node iteratively updates based on a distributed consensus protocol (such as an average consensus protocol):
[0092] Hi(k+1) = Hi(k) + ε·Σ aij·(Hj(k) − Hi(k));
[0093] Where aij is the communication weight between node i and node j (determined by link quality and priority), Hj(k) is the humidity of node j in round k, Hi(k) is the humidity of node i in round k, Hi(k+1) is the humidity of node i in round k+1, and ε is the convergence step size (0 < ε < 1 / maximum node degree). After a finite number of iterations, the humidity estimates of all nodes converge to a consistent value, based on which the cluster humidity balance index Q is calculated. This method does not rely on any central node and can guarantee convergence even if the topology changes dynamically.
[0094] (2) Generation of local coordinated regulation and control decisions
[0095] When the humidity deviation of this chamber (i.e., the difference between the actual humidity of this chamber and the target humidity value / target humidity range boundary) is greater than the first preset threshold (e.g., deviation > 10%RH), an independent rapid control command for this chamber is generated.
[0096] When the humidity deviation of this box is less than or equal to the first preset threshold and the humidity balance index of the cluster exceeds the second preset threshold (e.g., Q>8%RH), a node collaborative linkage control command is generated and sent to the scheduled neighboring nodes so that the neighboring nodes can perform collaborative dehumidification or humidification operations accordingly.
[0097] For example: Suppose node A (target humidity 45%RH) currently has a measured humidity of 41%RH, a deviation of 4%RH, which does not exceed the first preset threshold of 10%RH, and therefore does not require independent rapid adjustment. However, its neighboring nodes B and C have already started dehumidification, and the cluster balance index is high, indicating that the dehumidification capacity distribution within the cluster is uneven. Based on this, node A generates a coordinated control command, scheduling the neighboring node D with higher humidity to increase the dehumidification level, while appropriately reducing its own dehumidification power to achieve "peak-shifting complementarity." This ensures that the humidity of each chamber enters the target range while avoiding mutual interference and energy waste between adjacent chambers due to air circulation.
[0098] Step S4: Control Execution
[0099] Each moisture-proof cabinet node 100, based on the control instructions in the local coordinated control decision, drives its humidity control actuator 140 to perform corresponding humidity control actions, and synchronizes the control execution results as feedback information to adjacent nodes. The feedback information includes the execution mode, power level, start and stop times, and the humidity change trend after execution.
[0100] like Figure 2 As shown, the specific steps involved in the implementation of the control measures are as follows:
[0101] S41. Obtain the control instructions to be executed and determine the instruction type. The control instructions include the local independent rapid control instructions generated by this node in step S3 and the inter-node collaborative control instructions received from adjacent nodes. The node will queue the two types of instructions according to the time of receipt and priority, and the rapid control instructions triggered by alarms will be executed first.
[0102] S42. Based on the instruction type of the control command, match the target execution mode from a variety of preset control execution modes. Wherein:
[0103] When the control command is a rapid control command for this box, the rapid control mode is matched: it operates at the highest power level and in a continuous start-stop mode to bring the humidity back to the target range as soon as possible.
[0104] When the control command is an inter-node coordinated control command, the coordinated control mode is matched. The start and stop sequence of the coordinated control mode is determined based on the coordinated timing information carried in the inter-node coordinated control command, so that the humidity control action of this node is coordinated with the humidity control action of the scheduled adjacent nodes in terms of timing. The coordinated timing information includes the scheduling time slot offset and duration, such as... Figure 3As shown, for example, node A specifies in the instruction that node D starts dehumidification in the second time slot (30 seconds per time slot) and continues for 5 time slots, while node A itself runs at a low speed in the 0th and 4th time slots, thereby avoiding the superposition of power supply peaks caused by multiple nodes dehumidifying at high power at the same time and the "tug-of-war" of moisture between adjacent cabinets 110.
[0105] S43. According to the matched target execution mode, output the corresponding drive signal to the humidity control actuator 140 of this box to control the dehumidification module 141 and / or humidification module 142 to perform humidity control actions according to the start-stop sequence and power level corresponding to the target execution mode. The drive signal adopts a combination of PWM power adjustment and relay leveling: the power level is switched by relays (e.g., series and parallel combination of semiconductor cooling chips), and fine-grained adjustment is achieved by PWM duty cycle.
[0106] S44. After the humidity value of this chamber enters the target humidity range, switch to maintenance mode and periodically output fine-tuning drive signals with a low duty cycle to prevent the humidity value of this chamber from exceeding the target humidity range. Figure 4 As shown, for example, after entering the target range of 40% to 50% RH, the dehumidification module 141 outputs a fine-tuning pulse with a duty cycle of 10% and a duration of 10 seconds every 120 seconds to control the humidity fluctuation within ±2% RH in the middle of the target range, taking into account humidity control accuracy, energy consumption, and noise.
[0107] Step S5: Closed-loop feedback correction and convergence
[0108] Each moisture-proof box node 100 dynamically corrects the control parameters in the local collaborative control decision based on feedback information, forming a closed-loop feedback distributed humidity collaborative control until the humidity values of each moisture-proof box in the cluster converge to their respective target humidity range.
[0109] Specifically, each node continuously monitors feedback information (execution results, humidity change rate) from neighboring nodes and adjusts the following control parameters accordingly: adaptive relaxation or tightening of the first / second preset thresholds, adjustment of the collaborative timing offset, adjustment of the power level, and updating of priority weights. For example, if it is found that the humidity decrease rate of the scheduled node D is lower than expected after execution in the specified time slot, its scheduling time slot will be advanced or its power level will be increased in the next round of decision-making.
[0110] Furthermore, the dynamic correction is based on an energy consumption optimization objective function. This energy consumption optimization objective function aims to optimize the total energy consumption of the cluster and / or the control convergence time. It achieves this by adjusting the start-stop timing and power levels of the humidity control actuators 140 at the cluster level, allowing the control actions of multiple nodes to operate in staggered shifts. This embodiment uses the following objective function:
[0111] min J = α·ΣPi·ti + β·Tconv + γ·Q;
[0112] Where ΣPi·ti represents the total energy consumption of the cluster (the sum of the products of the power of each node and its runtime), Tconv represents the cluster humidity convergence time, Q represents the cluster humidity balance index, and α, β, and γ are weighting coefficients (which can be adjusted according to operational strategies, such as reducing α and increasing β during off-peak hours to accelerate convergence). The solution adopts a distributed greedy algorithm with local negotiation: each node locally evaluates the energy consumption of candidate (tier, time slot) combinations and negotiates time slot occupancy with neighboring nodes through limited instruction exchange to avoid conflicts, achieving "peak-shifting and valley-filling" cluster operation and reducing peak power demand.
[0113] Node failure fault tolerance steps
[0114] This method also includes a node failure fault tolerance step: when any moisture-proof box node 100 neither receives a humidity status message from the adjacent node nor receives a link layer response from the adjacent node through any candidate communication link within a preset time (e.g., 30 seconds, equal to 3 message sending cycles plus margin), the adjacent node is determined to be faulty. The node switches to local independent control mode (i.e., it only performs the control described in S42 to S44 based on the humidity of the box and the target humidity, and does not participate in coordinated linkage), and reports the faulty node information, triggering network topology update and route reconstruction (updating the cluster member table, recalculating the set of adjacent nodes, and re-electing the master node if necessary).
[0115] It should be noted that the node failure fault tolerance step and the fault switching of a single communication link are executed independently: the link-level failure (the degradation of a candidate link) is handled by the S24 sub-step of step S23, and the node-level failure is handled by this fault tolerance step. The two are decoupled from each other in terms of judgment conditions, processing level and recovery mechanism, thereby achieving dual fault tolerance of link redundancy and node redundancy. Failure at any level will not cause the collapse of the entire cluster control system.
[0116] Example of the overall workflow of the method
[0117] Taking a cluster of 6 dehumidifiers in a computer room of a certain laboratory building as an example: After power-on, the 6 nodes complete registration through self-organizing network (node IDs are 0x01 to 0x06). Node 0x01 is selected as the master node due to its highest link quality and remaining resource score and is responsible for clock synchronization broadcasting. Each node collects temperature-compensated humidity data every 5 seconds, encapsulates it into a unified frame format, and broadcasts it via LoRa link. The wireless signal of the cabinet where node 0x03 is located is blocked by the metal cabinet, and its LoRa link packet loss rate rises to 25%, triggering link switching. It then switches to RS-485 bus for transmission and reception, and historical missing data is filled in through interpolation compensation. During operation, the humidity deviation of node 0x04 reached 12%RH, triggering independent rapid regulation. After continuous dehumidification at a high level for 18 minutes, it returned to the target range. Subsequently, the cluster balance index Q=9.5%RH exceeded the second threshold of 8%RH. Nodes 0x02 and 0x05 established a coordinated scheduling mechanism to perform dehumidification / humidification at off-peak times. With the help of the energy consumption optimization objective function, the humidity of the entire cluster converged to their respective target ranges within 15 minutes and entered the maintenance mode.
[0118] The above embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention shall fall within the scope of protection claimed by the present invention.
Claims
1. A distributed humidity collaborative regulation method for a multi-hygrostat box cluster, characterized in that, Includes the following steps: S1. Each moisture-proof box node completes node registration and identity assignment through a distributed self-organizing network protocol, forming a decentralized or quasi-centralized self-organizing network of moisture-proof box clusters; S2. Each moisture-proof box node periodically collects its own humidity data based on a unified clock reference, and broadcasts and / or sends humidity status messages containing node identifiers, humidity data and timestamps to adjacent nodes through inter-node communication links. S3. Each moisture-proof box node calculates the cluster humidity balance index based on its own humidity data and the humidity status messages received from neighboring nodes, and generates a local collaborative control decision based on the cluster humidity balance index and the humidity deviation of its own box. S4. Each moisture-proof box node drives its humidity control actuator to perform corresponding humidity control actions according to the control instructions in the local collaborative control decision, and synchronizes the control execution results as feedback information to adjacent nodes. S5. Each dehumidifier node dynamically corrects the control parameters in the local collaborative control decision based on the feedback information, forming a closed-loop feedback distributed humidity collaborative control, until the humidity values of each dehumidifier in the cluster converge to their respective target humidity range.
2. The method of claim 1, wherein, The unified clock reference mentioned in step S2 is established through a distributed clock synchronization mechanism, which includes at least one of master node election, clock offset compensation, and periodic calibration; when the number of dehumidifier nodes in the cluster changes, network topology reconstruction and clock reference resynchronization are triggered.
3. The method of claim 1, wherein, The humidity data mentioned in step S2 undergoes data compensation processing before transmission, including sensor drift correction, temperature-humidity cross-sensitivity compensation, and historical data interpolation compensation under communication packet loss conditions.
4. The method of claim 1, wherein, The cluster humidity balance index mentioned in step S3 is calculated iteratively using a weighted average or consensus algorithm based on the humidity value of this box, the humidity values of adjacent nodes, and the priority weight of each node; the priority weight is dynamically allocated based on the humidity sensitivity level of the items stored in the dehumidifying box, the box volume, and / or the historical control response rate.
5. The method according to claim 1, characterized in that, The generation of local collaborative control decisions in step S3 specifically includes: when the humidity deviation of this box is greater than the first preset threshold, generating an independent rapid control command for this box; when the humidity deviation of this box is less than or equal to the first preset threshold and the cluster humidity balance index exceeds the second preset threshold, generating a collaborative control command between nodes and sending it to the scheduled adjacent nodes, so that the adjacent nodes can perform collaborative dehumidification or humidification operations accordingly.
6. The method according to claim 5, characterized in that, Step S4 involves performing humidity control actions, including: S41. Obtain the control instructions to be executed and determine their type; the control instructions include independent rapid control instructions for this box and inter-node collaborative control instructions received from adjacent nodes. S42. Match the target execution mode according to the instruction type: the independent rapid control instruction of this box matches the rapid control mode; the inter-node coordinated control instruction matches the coordinated control mode. Its start and stop sequence is determined according to the coordinated timing information carried by the instruction, so that the control actions of this node and the scheduled adjacent node are coordinated with each other in timing. S43. Output the corresponding drive signal to the humidity control actuator of this box according to the target execution mode, and perform humidity control action with the corresponding start-stop sequence and power level. S44. After the humidity value of this chamber enters the target humidity range, switch to maintenance mode and periodically output fine-tuning drive signals with a low duty cycle to prevent the humidity value of this chamber from exceeding the target humidity range.
7. The method according to claim 1, characterized in that, It also includes a node failure tolerance step: when any dehumidifier node neither receives a humidity status message from the adjacent node nor receives a link layer response from the adjacent node through any candidate communication link within a preset time, the adjacent node is determined to be faulty, the node switches to local independent control mode, and reports the faulty node information, triggering network topology update and route reconstruction; wherein, the node failure tolerance step and the fault switching of a single communication link are executed independently of each other.
8. The method according to claim 1, characterized in that, The dynamic correction of the control parameters in the local collaborative control decision-making in step S5 is based on the energy consumption optimization objective function. The energy consumption optimization objective function takes the total energy consumption of the cluster and / or the control convergence time as the optimization target. By adjusting the start-stop sequence and power level of the humidity control actuators of each node at the cluster level, the control actions of multiple nodes are staggered.
9. The method according to claim 1, characterized in that, Step S2 involves sending a humidity status message, including: S21. Evaluate the link quality of each candidate communication link to obtain link quality parameters including at least one of received signal strength, packet loss rate and transmission delay. S22. Determine the target communication link based on the link quality parameters and the message priority of the message to be sent; wherein the message priority of the humidity status message carrying abnormal alarm information is higher than that of the regular periodic humidity status message. S23. Encapsulate the humidity status message in a unified data frame format that is independent of the underlying communication protocol of each candidate link and send it so that the same message does not need to be reframed when switching between different candidate links. S24. Monitor the transmission status. If no acknowledgment is received within a preset time or the link quality parameters deteriorate to below a preset threshold, switch to the backup link to retransmit the encapsulated data frame, and update the link quality parameters of each candidate communication link based on the transmission result.
10. A distributed humidity collaborative control system for multiple dehumidifier clusters, characterized in that, include: Multiple dehumidifier cabinet nodes, each node including cabinet body, humidity sensor installed inside the cabinet body, local controller electrically connected to humidity sensor, and humidity control actuator electrically connected to local controller; The communication network allows each dehumidifier node to access the network through a communication unit located within its node, forming a distributed and interconnected dehumidifier cluster network. A distributed collaborative control module is deployed in each local controller to enable each dehumidifier node to execute the distributed humidity collaborative control method as described in any one of claims 1 to 9, based on the humidity status information exchanged between the dehumidifier node and its neighboring nodes, without the need for a central server.
