An intelligent door and window remote control system based on the Internet of Things

Abstract

The application relates to the field of intelligent door and window control, and discloses a remote control system for intelligent doors and windows based on the Internet of Things, which comprises a field bus, a plurality of execution nodes and a master controller in communication with the field bus. The master controller obtains a target opening degree, an actual opening degree of the nodes and external environmental parameters, comprehensively generates an initial scheduling matrix and sends the same to each execution node to drive corresponding actions. The system determines whether the current state is a local non-steady state or a global critical state based on the obtained environmental parameters and the running state data fed back by the execution nodes, and sends a scheduling maintenance instruction to dynamically intervene in response to the abnormal state. During the execution of the scheduling maintenance instruction, if it is determined that the system recovery condition is met, the current actual opening degree is obtained, a reconstructed scheduling matrix is generated, and the execution node action is continuously driven. The application feeds forward the mechanical action of the door and window and the state of the air conditioner flow field, prevents indoor air pressure imbalance caused by concurrent action, and realizes control closed-loop self-recovery after resistance abnormality occurs.

Description

Technical Field

[0001] This invention relates to the field of intelligent door and window control, specifically to an intelligent door and window remote control system based on the Internet of Things. Background Technology

[0002] With the development of IoT technology, intelligent remote control systems for doors and windows based on cloud-based interaction with field controllers have been widely applied. For example, patent document CN105003154B discloses a control architecture including an intelligent control terminal, a server, and an MCU controller, establishing the basic link for remote control of doors and windows; document CN110206439A introduces narrowband IoT communication technology to optimize data transmission. At the status monitoring level, documents CN115988314A and CN111459085A propose security monitoring mechanisms for all-time status monitoring and abnormal trigger video feedback, respectively. Furthermore, document CN116876950B achieves batch management and environmental sensing linkage for multiple door and window devices; document CN104389475B further proposes a collaborative control scheme integrating door / window and air conditioning switches.

[0003] However, the above-mentioned technical solutions still have the following shortcomings when dealing with complex building environments and multi-node collaborative control: First, when executing batch remote control commands, the existing system does not consider the feedforward coupling relationship between door and window actions and the HVAC system. Concurrent actions from multiple nodes can cause sudden changes in the effective opening area of ​​the building, leading to fluctuations in indoor air pressure, which in turn interferes with the indoor air conditioning operation logic and undermines the stability of indoor environmental control.

[0004] Secondly, in terms of monitoring equipment operating resistance, existing technologies mostly rely on fixed current thresholds to determine whether the motor is obstructed. In actual operation, the motor current is greatly affected by fluctuations in the supply voltage and the temperature of the motor itself. Furthermore, existing systems cannot distinguish whether the resistance originates from localized mechanical jamming of individual doors and windows or from wind pressure imbalance caused by gusts of wind. When encountering wind load fluctuations in normal environments, they tend to take a uniform shutdown protection action, increasing the rate of erroneous shutdown of on-site equipment.

[0005] Finally, when an external environmental anomaly is encountered and the shutdown protection is triggered, the existing control system directly terminates and invalidates the current remote dispatching commands. After the external strong wind resistance or abnormal interference dissipates, the system lacks a monitoring and self-recovery mechanism for intermediate states, causing the equipment to remain in an interrupted position. It is necessary to rely on manual re-issuance of commands, which cannot guarantee the final closed loop of the remote dispatching task. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an IoT-based intelligent door and window remote control system. This system solves the problems in existing intelligent door and window remote control systems, such as abnormal fluctuations in indoor air pressure caused by concurrent actions of multiple nodes, coupling with air conditioning and heating systems affecting indoor environmental regulation, and the lack of dynamic scheduling and safety self-recovery mechanisms when communication is congested on-site or when encountering abnormal external wind resistance.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent door and window remote control system based on the Internet of Things, comprising a fieldbus, multiple execution nodes, and a master controller communicating with the execution nodes via the fieldbus; the master controller includes: The status acquisition module is used to acquire the target opening degree, the actual opening degree of the execution node, and external environment parameters; The scheduling generation module is used to generate an initial scheduling matrix based on the actual opening degree, the target opening degree, and the external environment parameters. The instruction issuing module is used to issue the initial scheduling matrix to the execution node to drive the corresponding execution node action; The state determination module is used to determine whether a local unsteady state or a global critical state has been entered based on the external environment parameters and the running state data fed back by the execution node. An anomaly control module is used to intervene by issuing a scheduling and holding command in response to the local unsteady state or the global critical state. The closed-loop reconfiguration module is used to generate a reconfiguration scheduling matrix and continue to drive the execution node actions when the system recovery conditions are met under the execution state of the scheduling hold instruction.

[0008] In a preferred embodiment of the present invention, the external environmental parameters include building area pressure difference, outdoor wind speed and direction, air conditioning static pressure data, and inverter status; the system further includes a cloud platform communicatively connected to the main controller and used for issuing remote dispatch commands, and an air conditioning controller providing the air conditioning static pressure data and the inverter status. The status acquisition module extracts the target opening degree by parsing the remote dispatch command and collects the actual opening degree via the fieldbus; furthermore, the status acquisition module is also used to verify whether the current control time is within the effective time range of the remote dispatch command. The external environmental parameters are timestamped and invalid data exceeding the maximum allowable data age are removed through a global synchronization time window to ensure the timeliness of the external environmental parameters. A sliding statistical time window is used to monitor the current message load rate of the fieldbus. Furthermore, the status acquisition module stops sending the initial scheduling matrix when it detects a fault in a sensor used to collect the external environmental parameters, an offline status of a preset key execution node, or when the current message load rate exceeds a preset communication congestion threshold. When the air conditioning controller communication is interrupted but the building area pressure difference continues to update, the status acquisition module also executes pressure difference constraint degradation control logic, forcibly reducing the upper limit of the allowed concurrent door and window reference number and calculating the initial scheduling matrix solely based on the building area pressure difference.

[0009] Furthermore, the scheduling generation module is further used to divide the multiple execution nodes into discrete action batches based on preset constraints; determine the predicted displacement increment based on the preset initial prediction time step and the calibration motion speed and motion direction coefficients corresponding to the execution nodes, thereby generating the expected opening degree and the corresponding predicted running time limited by the target opening degree, and obtaining the effective opening increment based on the expected opening degree; and use a multivariate autoregressive model to combine the predicted pressure difference of the previous action batch, the effective opening increment, the expected change of the inverter output frequency extracted based on the inverter status, and the air conditioning static pressure deviation obtained based on the air conditioning static pressure data to comprehensively generate the predicted pressure difference of the controlled area.

[0010] Furthermore, the initial scheduling matrix includes at least a start delay parameter and a segmented target opening; the scheduling generation module sorts and pushes multiple execution nodes into the current action batch for verification based on the pressure difference influence weight; if adding a new node causes the preset constraint condition to be broken, the scheduling generation module will postpone the execution node that violates the preset constraint condition to the next action batch, or re-verify and estimate by halving the prediction time step of the corresponding execution node and generate a smaller segmented target opening; if the prediction time step has reached the preset minimum prediction time step and still does not meet the preset constraint condition, the scheduling generation module will mark the corresponding execution node as unschedulable under the current boundary conditions.

[0011] Furthermore, the instruction issuing module is also used to broadcast lease renewal frames according to a set period; and is also used to determine the start delay parameters of subsequent action batches based on the sum of the start delay parameters of the previous action batch, the longest predicted running time in the previous batch, the regional pressure response observation time, and the fieldbus communication margin time; the execution node controls the action timing according to the start delay parameters and the set lease duration, and stops outputting the drive pulse signal when the actual opening degree enters the allowable error range of the segmented target opening degree or when the lease renewal frame is not received.

[0012] In a preferred embodiment of the present invention, the operating status data includes motor load residual; the execution node includes a motor and a node controller, the node controller is used to acquire the actual opening degree, actual power supply voltage, operating temperature and real-time operating current; the node controller dynamically compensates the real-time operating current based on the actual power supply voltage and the operating temperature to generate a normalized operating current; and generates the motor load residual based on the difference between the normalized operating current and the basic operating current without external wind resistance and feeds it back to the fieldbus.

[0013] Furthermore, the state determination module is further used to calculate a normalized regional residual index based on the motor load residual of the execution node running in the controlled area and the rated operating current of the corresponding motor; when the regional load abnormality condition is met or the absolute value of the air conditioning static pressure deviation exceeds the set allowable deviation and the inverter is in a limited saturation state, the state determination module determines that the system has entered the local unsteady state; when the building area pressure difference or its pressure difference change rate exceeds the preset safety boundary, the system determines that the system has entered the global critical state.

[0014] Furthermore, the specific logic for the intervention executed by the anomaly control module is as follows: when forcibly abolishing the current scheduling version, the execution nodes that have not yet started are forcibly stopped, and the running execution nodes are controlled to perform blocking and recovery control under the global critical state; when performing the blocking and recovery control, the anomaly control module evaluates the impact of continued action on the predicted pressure difference in the controlled area. If it is determined that continued action is beneficial to the pressure difference recovering to the pressure target range and not exceeding the limit, the output of the drive pulse signal is maintained, allowing it to continue running to the safe set opening within the authorized duration; if it is determined to be in a deteriorating direction, the current position is maintained by braking; and the control logic for continued action is executed under the local unsteady state to continue to complete only the action corresponding to the current segmented target opening.

[0015] In a preferred embodiment of the present invention, the reconfiguration scheduling matrix is ​​generated based on the current actual opening degree and the restored building area pressure difference; wherein, the closed-loop reconfiguration module, under the scheduling hold command execution state, continuously monitors the air conditioning static pressure deviation and the inverter status to determine whether the system recovery conditions are met; specifically, the closed-loop reconfiguration module comprehensively determines the system settling time based on the air conditioning static pressure deviation, the inverter status, and the normalized regional residual index; The criteria for determining whether the system recovery conditions are met include: the accumulated waiting time since the issuance of the scheduling hold command reaches the system stabilization time, the updated building area pressure difference enters the pressure target range, and the air conditioning static pressure deviation is continuously maintained within the preset steady-state error bandwidth within the preset confirmation period; when the accumulated waiting time exceeds the preset system circuit breaker threshold and the system recovery conditions are not met, the closed-loop reconfiguration module terminates the current scheduling task and reports the reason for recovery failure and environmental out-of-control alarm. After the system recovery conditions are met, the closed-loop reconfiguration module initiates position polling to the execution node to obtain the actual opening degree, treats the unfinished actions as the initial disturbance state, recalculates and generates the reconfiguration scheduling matrix, and issues it to continue the remaining control actions until the instructions are executed.

[0016] This invention provides an intelligent door and window remote control system based on the Internet of Things, which has the following beneficial effects: This invention divides execution nodes into discrete action batches through a scheduling generation module. Combining the action opening increment, the expected change in inverter output frequency, and the air conditioning static pressure deviation, a multivariate autoregressive model is used to calculate the predicted pressure difference of the controlled area to dynamically adjust the scheduling matrix. This method feeds forward the mechanical displacement of doors and windows with the operating state of the air conditioning and HVAC system, avoiding sudden changes in indoor air pressure and interference with the air conditioning operating logic caused by concurrent actions of multiple doors and windows, thus maintaining the stability of indoor environmental control.

[0017] This invention calculates the motor load residual reflecting the physical load by combining the node controller with the actual voltage and operating temperature. The state determination module then uses the residual index and differential pressure data to accurately distinguish between local unsteady states and global critical states. The anomaly control module then executes differentiated interventions, either completing only the current segment's action or forcibly braking, respectively. This mechanism eliminates the interference of electrical and temperature fluctuations on load detection, effectively distinguishing between local mechanical resistance at the node and system-level flow field imbalances, thus reducing the rate of false shutdowns while ensuring the safety of field equipment.

[0018] This invention monitors the static pressure of the air conditioner and the status of the frequency converter during the execution of the scheduling hold command through a closed-loop reconfiguration module. After the building pressure difference recovers to the target range and the system settling time is met, the current actual opening degree of the node is reacquired, and this is used as a new calculation starting point to generate a reconfigured scheduling matrix. This mechanism solves the problem of the control system directly terminating and abolishing the original command when encountering strong external wind resistance or environmental anomalies triggering protection. It enables the system to continue driving the node to complete the remaining actions after the abnormal interference dissipates, ensuring the final closed loop of the remote scheduling task. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall system architecture of the present invention; Figure 2 This is a flowchart of the IoT-based smart door and window remote control method of the present invention; Figure 3 This is a timing diagram of the cross-domain multi-source data synchronous acquisition and instruction parsing mechanism of the present invention; Figure 4 This is a schematic diagram illustrating the principle of generating the initial scheduling matrix based on pressure gradient constraint according to the present invention. Figure 5 This is a schematic diagram of the soft real-time execution mechanism of the present invention; Figure 6 This is a schematic diagram illustrating the node feature extraction and non-steady-state determination principle of the present invention. Figure 7 This is a schematic diagram of the hierarchical unsteady truncation and distributed cooperative maintenance principle of the present invention. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, this embodiment of the invention provides an intelligent door and window remote control system based on the Internet of Things (IoT). As a typical IoT control system, it includes a cloud platform, a main controller, a fieldbus, a differential pressure unit, a meteorological node, an air conditioning controller, and multiple execution nodes. The building environment controlled by the system is divided into multiple controlled areas.

[0022] The differential pressure unit includes one or more differential pressure measurement channels. Each controlled area is configured with at least one corresponding differential pressure measurement channel, or multiple controlled areas can share a pre-calibrated differential pressure measurement channel that can characterize the pressure state of the corresponding area. The main controller internally stores an area configuration table, which records at least the correspondence between controlled areas, execution nodes, differential pressure measurement channels, air conditioning control objects, and air conditioning static pressure feedback variables.

[0023] The main controller is an industrial-grade controller, such as a PLC (Programmable Logic Controller), which is responsible for global scheduling and instruction parsing. It establishes a communication link with the cloud platform to receive remote scheduling commands and simultaneously connects to the differential pressure unit, meteorological node, and air conditioning controller to synchronously acquire external environmental parameters. These parameters include building area differential pressure, outdoor wind speed and direction, air conditioning static pressure data (including air conditioning static pressure setpoint and feedback), and inverter status. To achieve distributed control and time-series scheduling of distributed hardware, the main controller sends initial or reconstructed scheduling matrices to multiple execution nodes distributed across various controlled areas via a fieldbus. The fieldbus is an industrial bus, specifically a CAN bus or Modbus RTU bus.

[0024] The main controller is equipped with multiple functional modules, including a status acquisition module, a schedule generation module, an instruction issuance module, a status determination module, an anomaly control module, and a closed-loop reconstruction module.

[0025] The status acquisition module is used to suspend remote scheduling commands and synchronously collect the actual opening degree of the node, communication status, external environmental parameters, and message load rate of the fieldbus. The scheduling generation module is used to calculate the effective opening increment and predicted pressure difference based on physical operating parameters, and divide the action batches according to preset constraints to generate the initial scheduling matrix. The command issuance module is used to issue the initial scheduling matrix and unified activation command to the fieldbus, and broadcast lease renewal frames periodically to maintain the soft real-time operation of the node.

[0026] The status determination module is used to perform integrated calculations based on external environmental parameters and motor load residuals fed back as operating status data from each execution node. It also considers the exceeding limits of air conditioning static pressure deviation and building area pressure difference to determine whether the system has entered a local unsteady state or a global critical state. The anomaly control module is used to forcibly abolish the current scheduling version when an anomaly is triggered and issue scheduling hold instructions to implement control logic for continued action or asymmetric blocking control, respectively. The closed-loop reconstruction module is used to calculate the system settling time and generate a reconstruction scheduling matrix when the system recovery conditions are met. It reports closed-loop feedback data to the cloud platform after the instructions are executed.

[0027] Each execution node is installed on a corresponding door or window to control its opening degree. Each execution node includes a node controller, drive circuit, position unit, current unit, and temperature unit. The node controller, as the microprocessor core of the node, parses the initial scheduling matrix, reconstructs the scheduling matrix, and unifies activation or hold instructions from the fieldbus. It then outputs drive pulse signals to the drive circuit to control motor operation and monitors the motor's power supply parameters via the drive circuit. During this execution process, the position unit provides real-time feedback on the actual opening degree and operating speed of the door or window to the node controller, while the current and temperature units extract the real-time operating current and operating temperature of the motor for subsequent calculations of the motor load residual.

[0028] The system establishes and relies on various boundary parameters during operation. The building zone differential pressure is defined as the difference between the indoor pressure of the corresponding controlled zone and the outdoor reference static pressure used in that controlled zone, with measurement data provided by the differential pressure measurement channel specified in the zone configuration table. The air conditioning static pressure deviation is defined as the difference between the air conditioning static pressure setpoint and the air conditioning static pressure feedback value. The air conditioning static pressure feedback value corresponds to a specific closed-loop feedback variable actually used by the air conditioning controller for the corresponding controlled zone. This closed-loop feedback variable is one of the following: supply air duct static pressure, return air duct static pressure, or room pressure, and is used in pairs with the air conditioning static pressure setpoint of the same type.

[0029] Different types of feedback variables are configured with separate allowable deviations and steady-state error bandwidths, and do not share the same threshold. A scheduling lease is defined as the effective duration for which an execution node is authorized to start a motor according to a specified scheduling version. Scheduling hold is defined as the control state that stops the countdown of execution nodes that have not yet started and prohibits them from starting motors according to the current scheduling version, and controls the actions of running execution nodes. Motor load residual is defined as the difference between the normalized operating current obtained by normalizing the real-time operating current to standard supply voltage and standard temperature conditions, and the base operating current under the corresponding actual opening degree and operating speed. The pressure target range is defined as the upper and lower limits of the pressure expected to be maintained in each controlled area.

[0030] Reference Figure 2 This invention provides a method for remote control of smart doors and windows based on the Internet of Things, comprising the following steps: S10. Remote Command Suspension and Cross-Domain Status Acquisition. The cloud platform issues remote scheduling commands containing command numbers, controlled areas, and target opening degrees. After receiving the remote scheduling commands through the status acquisition module, the main controller sets them to a pending scheduling state, temporarily suspending the generation and issuance of scheduling matrices for that command. In the pending scheduling state, the main controller synchronously collects the actual opening degree and communication status of each execution node through the status acquisition module via the fieldbus, and coordinates with the differential pressure unit, meteorological node, and air conditioning controller to acquire building area differential pressure, outdoor wind speed and direction, air conditioning static pressure data including air conditioning static pressure setpoint and air conditioning static pressure feedback value, inverter status, and fieldbus message load rate; and when it detects abnormal acquisition of multi-source operating data or offline of key nodes, it executes command suspension or differential pressure constraint degradation control logic.

[0031] S20. Initial scheduling generation under differential pressure constraints. Based on the acquired physical operating parameters and the target opening degree in the remote scheduling command, the main controller calculates the effective opening increment generated by the execution node actions through the scheduling generation module, and combines the regional differential pressure mapping with the inverter status of the air conditioning controller to calculate the predicted differential pressure of the controlled area according to the effective opening increment. The physical operating parameters include the actual opening degree, building area differential pressure, outdoor wind speed and direction, and air conditioning static pressure data.

[0032] Next, based on the acquired communication status, multiple execution nodes are divided into discrete action batches. During the division process, it is ensured that the divided action batches meet preset constraints, including the upper limit of absolute pressure difference and the upper limit of pressure difference gradient for the predicted pressure difference, and the upper limit of message load rate for the fieldbus. Finally, by assigning start delay parameters to different batches, an initial scheduling matrix is ​​generated, which is configured with node identifier, start delay parameters, action batch number, controlled area, command number, lease duration, scheduling version number, segment target opening degree, action direction, target opening degree allowable error, and maximum allowable running time.

[0033] S30. Scheduling Data Activation and Node Execution. The master controller sends the initial scheduling matrix to the corresponding execution nodes via the command sending module for pre-loading of the initial scheduling matrix. After the parameters are fixed, a unified activation command is broadcast downwards. Each execution node eliminates communication delay differences and uses the moment of receiving the unified activation command as the starting point for timing, starting a local delay countdown based on the start delay parameters. The master controller continues to broadcast lease renewal frames at a set period during system operation via the command sending module. When the countdown reaches zero and is within the valid duration of the scheduling lease, the execution node outputs a drive pulse signal to drive the local motor to perform actions, and stops driving when the actual opening degree enters the allowable error range of the segmented target opening degree. At the same time, if a valid lease renewal frame is not received, causing the scheduling lease to expire, the local motor control authority is forcibly locked to enter a safe suspension mode.

[0034] S40. Load Residual Calculation and Unsteady-State Classification. During motor operation, the execution node acquires node operating parameters and calculates and feeds back the motor load residual as operating status data to the fieldbus based on the node operating parameters and the real-time operating current of the motor. The node operating parameters include its actual opening degree, operating speed, power supply parameters, and motor operating temperature. After receiving the node feedback through the status determination module, the main controller performs integrated calculation on the motor load residual uploaded by each execution node according to the controlled area; and when the area load abnormality condition is met, or the absolute value of the air conditioning static pressure deviation exceeds the allowable deviation and the frequency converter is in the limit saturation state in the corresponding adjustment direction, the system is determined to enter a local unsteady state; when the building area pressure difference or pressure difference change rate exceeds the preset safety boundary, the system is determined to enter a global critical state. When both the local unsteady state and the global critical state are met, the global critical state is taken as the current risk level.

[0035] S50. Scheduling Hold and Physical Cut-off in Abnormal State Response. In response to the triggering of a local unsteady state or a global critical state, the master controller, through the abnormal control module, forcibly cancels the current scheduling version based on the scheduling version number and issues a scheduling hold command with high communication priority via the fieldbus. Execution nodes that have not yet started are forcibly stopped from counting down and are prohibited from starting motors according to the current scheduling version. Running execution nodes execute the control logic for continued operation under local unsteady states; under global critical states, they execute blocking and recovery control based on differential pressure prediction, i.e., cutting off and locking motors causing pressure deterioration, while allowing motors that contribute to pressure recovery to run to a safe set opening degree, thereby suppressing system load dispersion and accelerating environmental steady-state recovery.

[0036] S60. System Stabilization Assessment and Closed-Loop Scheduling Reconstruction. While maintaining the scheduled operation, the main controller continuously monitors the air conditioning static pressure deviation and inverter status of the air conditioning controller via the closed-loop reconstruction module to calculate the system stabilization time. Upon determining that the system meets the system recovery conditions, a reconstructed scheduling matrix is ​​calculated and generated based on the current actual opening degree and the recovered building area pressure difference. System recovery conditions include waiting time and boundary pressure difference conditions. The reconstructed scheduling matrix is ​​then issued, and a unified activation command is issued again to continue the remaining control actions. The corresponding process is iteratively executed until all remote scheduling commands are completed. After final completion, the closed-loop reconstruction module reports the final actual opening degree of each execution node and the system operation log to the cloud platform, achieving global closed-loop feedback for remote control.

[0037] To more clearly demonstrate the operating mechanism of the embodiments of the present invention, the control logic of each core module and the execution algorithm of each step are described in detail below.

[0038] like Figure 3As shown, during the execution of step S10, remote command suspension and cross-domain status acquisition, the main controller specifically uses the status acquisition module to perform command buffering and physical quantity time window alignment acquisition to obtain basic operating status data before the system executes actions, including the following sub-steps: S11. Command Reduction and Safety Buffer Suspension. The cloud platform issues remote dispatch commands based on building ventilation requirements. The data packet of the remote dispatch command includes the command number, controlled area, target opening degree, generation time, and command validity timestamp. The status acquisition module parses the received remote dispatch command, pushes the parsed data into the main controller's execution queue, and simultaneously sets the remote dispatch command to a pending dispatch state, temporarily suspending the triggering of the control link of the underlying nodes and delaying the generation of the initial dispatch matrix for the remote dispatch command.

[0039] Before a remote scheduling command enters the execution queue and before each generation of the scheduling matrix, the status acquisition module verifies whether the current control time is within the command's valid time range. If a command exceeds its valid time during the suspension period, the status acquisition module removes it from the execution queue, prohibits further generation or activation of the scheduling matrix, and reports the command's expiration status to the cloud platform. Through the command suspension mechanism, the system reserves a time window for calculating the scheduling before the action is executed, preventing uncontrollable impacts on the building's ambient air pressure caused by concurrent start-up and shutdown of multiple nodes.

[0040] S12. Cross-protocol multi-source physical parameter synchronous time window acquisition. During the time period when the remote scheduling command is in the waiting-for-scheduling state, the status acquisition module starts multi-channel concurrent acquisition. The first data channel sends status polling messages to each execution node in the controlled area via the fieldbus. The execution nodes respond to the status polling messages and upload the actual opening degree, movement direction, and hardware fault identifier obtained by the absolute encoder or incremental encoder; at the same time, the status acquisition module updates the online communication status of each execution node according to the response timeout.

[0041] The second data channel acquires boundary constraint parameters for environmental and HVAC equipment. The status acquisition module communicates with the differential pressure unit to extract the building area pressure difference and with the meteorological node to extract outdoor wind speed and direction. The outdoor wind speed and direction are only used to calculate the wind pressure correction coefficient and must not replace the actual measured value of the building area pressure difference. It also uses the industrial Ethernet protocol to read the internal operating buffer of the air conditioning controller to extract the air conditioning static pressure data, which includes the air conditioning static pressure setpoint and the air conditioning static pressure feedback value. Simultaneously, it captures the inverter status, which includes the current output frequency and limiting status flag of the air conditioning inverter.

[0042] The status acquisition module establishes a fixed-duration global synchronization time window. The duration of the global synchronization time window is set based on the longest polling cycle of the fieldbus, for example, 500ms. The status acquisition module latches each physical operating parameter falling within the same global synchronization time window and adds a unified control timestamp to eliminate phase errors introduced by asynchronous sampling in subsequent differential pressure calculations. A maximum allowable data age is configured for each type of physical operating parameter. Data falling within the same global synchronization time window but whose sampling time differs from the unified control timestamp by more than the corresponding maximum allowable data age is marked as invalid data and will not be included in subsequent prediction and judgment.

[0043] The current message load rate of the fieldbus is calculated using a sliding statistical time window. The master controller counts the total number of bits of messages actually sent and received within this time window. The total number of bits includes the address field, function field, data field, check field, frame interval converted bits as specified by the protocol, and the actual number of retransmitted messages. The current message load rate is obtained by dividing this total number of bits by the product of the fieldbus communication baud rate and the duration of the statistical time window. The expected message load rate of the candidate action batch is calculated based on the current message load rate, plus the expected occupancy of scheduling matrix distribution messages, loading response messages, unified activation instructions, lease renewal frames, status polling messages, and running residual feedback messages per unit time. The worst-case communication occupancy is calculated based on the configured maximum number of retransmissions.

[0044] S13. Data Validity Verification and Abnormal State Response Degradation. The status acquisition module performs integrity assessment on the latched multi-source operational data. The benchmarks for data integrity assessment include the range overflow flags fed back by sensors, the verification error statistics of the communication protocol layer, and the node offline rate. Based on the assessment results, the status acquisition module implements different levels of command interception or degradation flow.

[0045] The status acquisition module determines that the system has lost its control loop under any of the following conditions: First, the differential pressure unit corresponding to the building area pressure difference reports a fault frame, indicating a sensor failure used to collect external environmental parameters; second, based on the acquired outdoor wind direction, the preset execution node located on the windward side of the building and having the largest proportion of its window area to the total exhaust area of ​​the controlled area is offline, indicating a preset critical execution node is offline; third, the current message load rate of the fieldbus exceeds the preset communication congestion threshold. The preset communication congestion threshold is a fixed limit preset based on the fieldbus communication baud rate and message polling cycle, for example, set to 80% of the bus bandwidth.

[0046] At this time, the status acquisition module maintains the pending scheduling status bit of the remote scheduling command, stops sending new scheduling matrices, reduces the polling frequency of non-critical statuses, and suspends the sending of non-control messages such as system logs. Scheduling is only resumed when the message load rate is lower than the preset communication recovery threshold within several consecutive statistical time windows. The communication recovery threshold is lower than the communication congestion threshold to form a recovery hysteresis and avoid repeated entry and exit from the congestion state until the communication link is restored or the sensor fault is cleared.

[0047] When an interruption in the air conditioning controller communication causes the static pressure data and inverter status to become invalid, but the differential pressure unit continuously updates the building area differential pressure within the effective measurement range, the status acquisition module determines that the system has entered a partial operational data blind zone. Under partial operational data blind zone conditions, the status acquisition module executes differential pressure constraint degradation control logic, forcibly modifying the upper limit of the number of doors and windows that the system is allowed to simultaneously perform opening and closing actions to a preset low-concurrency capacity parameter, thereby limiting the concurrent scale of door and window actions. The low-concurrency capacity parameter is a value calibrated based on the natural ventilation rate of the controlled area of ​​the building and historical safe operation data; for example, it is set to allow only a single execution node to perform an action in a single scheduling. When using degradation control logic, the scheduling generation module only performs scheduling calculations based on the building area differential pressure, and the system disables the early recovery function based on the air conditioning controller's stabilization status during subsequent execution, mainly utilizing the natural balance of indoor and outdoor pressure differences to smooth out air pressure fluctuations caused by door and window opening and closing.

[0048] When the meteorological node data fails but the differential pressure measurement is still valid, the scheduling generation module uses the most unfavorable wind pressure correction parameter for the current candidate action from the pre-stored wind direction intervals and limits the concurrent capacity to a low concurrent capacity parameter; if the differential pressure constraint still cannot be met under the most unfavorable wind pressure correction parameter, the command is suspended.

[0049] When the location data of an execution node fails, the node shall not be included in the action batch; when the current, temperature, power supply voltage or location data of a running node fails, the node controller stops the corresponding motor and sets a sensor abnormality fault flag, and the master controller excludes the node from subsequent reconfiguration scheduling.

[0050] See Figure 4 During the initial scheduling generation process under the pressure difference constraint in step S20, the master controller maps the air pressure changes in the physical environment into a calculation model through the scheduling generation module to limit the number of concurrent actions of the execution nodes and perform timing orchestration. This includes the following sub-steps: S21. Calculation of Node Expected Opening and Effective Opening Increment. The scheduling generation module calculates the expected opening based on the received target opening, the actual opening of the execution node included in the physical operation parameters, the pre-stored calibration speed of the execution node, and the preset initial prediction time step. The initial prediction time step is set based on the control cycle period of the microprocessor inside the main controller and the corresponding prediction running time. To handle the physical boundary truncation problem during the bidirectional movement of doors and windows, the scheduling generation module calculates the expected opening, which is equal to the current actual opening plus the predicted displacement increment in the direction of the node's movement. The expected opening is limited to the closed interval defined by the current actual opening and the target opening. The predicted displacement increment is equal to the product of the calibration speed of the door / window, the prediction time step, and the movement direction coefficient. When the target opening is greater than the current actual opening, the movement direction coefficient is positive; when the target opening is less than the current actual opening, the movement direction coefficient is negative. When the difference between the target opening and the current actual opening is within the allowable error range of the target opening, no action item is generated for the corresponding node.

[0051] After obtaining the expected opening degree, the scheduling generation module combines the aerodynamic impact of outdoor wind speed and direction on the opening, including physical operation parameters, to calculate the effective opening increment of the corresponding doors and windows within the action batch, in order to correct for the differences in disturbance to airflow exchange caused by different wind conditions. The formula for calculating the effective opening increment is as follows: ; in, For the first The first door and window Effective opening increment within a batch of actions; In the first Outdoor wind speed within each action batch; In the first Outdoor wind direction within each batch of actions; For the first Normal orientation parameters of the building facade where each door and window is located; For the first The first door and window The expected opening degree after the execution of each batch of actions; For the first The first door and window The actual opening at the start of each action batch; The dynamic facade wind direction correction coefficient is obtained based on building aerodynamic experiments. Its value range is set between 0 and 1.5. It is used to characterize the ventilation gain or attenuation of wind pressure relative to the static pressure environment. For the first The expected effective ventilation area of ​​each door and window at its expected opening degree; For the first The current effective ventilation area of ​​each door and window at its actual opening degree.

[0052] The calibration process for the nonlinear calibration function can be achieved by extracting data from wind tunnel testing or fluid dynamics simulation models and then performing curve fitting. Before the system is put into operation, the opening degree and area are calibrated for each execution node. During calibration, multiple opening degree sampling points are selected between the fully closed and fully open positions of doors and windows, and the corresponding geometric ventilation area or the equivalent ventilation area obtained by air volume conversion is recorded. These are then stored in an opening degree and area calibration table in ascending order of opening degree. The scheduling generation module calculates the values ​​of adjacent calibration points using piecewise linear interpolation or a preset monotonic fitting function. It also restricts inputs outside the calibration range to fully closed and fully open endpoint values.

[0053] A two-dimensional calibration table is established for facade wind direction correction parameters based on outdoor wind speed and wind direction ranges. For wind direction ranges that cause changes in the direction of pressure differential action, the sign of the corresponding direction is retained by the regional pressure differential response parameters, recording not only the magnitude of the gain. When the wind speed exceeds the calibrated range, the most unfavorable boundary value is used, without unbounded extrapolation.

[0054] S22. Predicted pressure difference generation based on regional pressure difference mapping. The scheduling generation module introduces a multivariate autoregressive model based on the effective opening increment, combining the regional pressure difference mapping with the inverter status of the air conditioning controller, and comprehensively considering historical pressure difference inertia, disturbances from newly added physical openings, and the stabilization compensation provided by the inverter to calculate the predicted pressure difference within the controlled area. In this calculation process, the basic data of the regional pressure difference mapping is deeply integrated with the building area pressure difference and air conditioning static pressure data included in the physical operating parameters. For candidate action batches... The scheduling generation module calculates the predicted pressure difference for each affected and controlled area in this batch. The calculation expression for the predicted pressure difference is as follows: ; in, Controlled area Introducing the first Predicted pressure difference after each batch of actions; Controlled area Introducing the first Predicted pressure difference before each batch of actions; Controlled area The pressure difference history inertial response parameters; For the proposed selection of the first A set of execution nodes for each action batch; For the first Each door and window corresponds to a controlled area. The opening pressure difference response parameters; For the first The first door and window Effective opening increment within a batch of actions; Controlled area The frequency converter's calming response parameters; For the corresponding to the first The expected change in the inverter output frequency of the air conditioner controller within a batch of actions; These are the static pressure deviation response parameters for air conditioning. For the corresponding to the first The deviation between the air conditioning static pressure setpoint and the corresponding air conditioning static pressure feedback value within a batch of actions; This represents the inertial lag effect of historical pressure differentials in the current forecast period; For the first The effective increase in the opening of each door and window affects the controlled area. The resulting single-node pressure differential disturbance; The total physical opening pressure difference disturbance caused by the actions of all doors and windows within the batch set; The differential pressure is used to compensate for the environmental stabilization caused by the adjustment of the inverter output.

[0055] In the recursive calculation of predicted differential pressure, the initial predicted differential pressure of the first candidate action batch is equal to the measured differential pressure of the building area latched within the current global synchronization time window. Subsequent candidate action batches all use the predicted differential pressure of the previous batch as the starting point for recursion. The expected change in inverter output frequency is obtained by multiplying the rate of change of output frequency within the two most recent effective global synchronization time windows by the prediction time interval of this batch, and is limited to the inverter's allowed rate of increase / decrease and the upper and lower limits of output frequency. When the inverter is in a limited saturation state in the corresponding direction, it is no longer predicted to continue changing in that direction. When the controlled area is not associated with inverter air conditioning equipment or the air conditioning data is invalid, the inverter stabilization response parameter and the air conditioning static pressure deviation response parameter are both set to zero, and a degraded scheduling rule is adopted.

[0056] Regional differential pressure prediction parameters, including historical inertial response parameters, opening differential pressure response parameters, and inverter stabilization response parameters, are all obtained through multiple linear regression identification using batch action calibration data. Each calibration data point includes at least the regional differential pressure before action, the regional differential pressure after action, the effective opening increment of each door and window, the static pressure deviation of the air conditioner, the change in inverter output frequency, and the outdoor wind speed and direction range. The system uses the regional differential pressure after action as the regression output and the remaining parameters as regression inputs for parameter identification. The multiple linear regression algorithm can be solved using the least squares method. The identified parameters are written to the main controller's parameter storage area only if the prediction error on independent verification data is less than the preset allowable error; otherwise, the system will degrade to low-concurrency degraded scheduling.

[0057] S23. Constrained Boundary Verification and Discrete Action Batch Division. The scheduling generation module divides the execution nodes into action batches based on preset constraints. These preset constraints include three safety threshold boundaries: The first is the absolute pressure difference upper limit; the absolute value of the calculated predicted pressure difference must not exceed the configured maximum static pressure extreme value, which is set according to the physical pressure bearing standards of doors and windows. The second is the pressure difference gradient upper limit, designed to prevent drastic changes in indoor air pressure. Its constraint is: the absolute value of the difference between the predicted pressure difference and the current building area pressure difference, divided by the prediction time step, must result in a pressure difference change rate per unit time that is less than or equal to that of the controlled area. The upper limit of the differential pressure gradient is configured and is experimentally calibrated based on the human comfort tolerance limit. The third item is the upper limit of the message load rate. The expected message load rate corresponding to the selected action batch must be less than or equal to the set upper limit of the message load rate. The upper limit of the message load rate is calculated based on the maximum data throughput capacity of the fieldbus.

[0058] During the partitioning phase, the scheduling generation module, based on the acquired communication status, directly removes nodes that have gone offline abnormally, and then performs the following actions sequentially: Each execution node in the set is independently substituted into the differential pressure prediction model to calculate the absolute value of the predicted differential pressure increment as its impact weight, and then sorted in descending order according to the impact weight; next, the sorted execution nodes are pushed into the current action batch one by one, and the calculation expressions and preset constraints are mathematically verified simultaneously. If adding a new node causes any preset constraint in any controlled area affected by that node to be broken, the scheduling generation module immediately truncates the current action batch, and the execution nodes that violate the preset constraints and the unprocessed execution nodes are carried over to the next action batch.

[0059] To prevent a single node with an excessively large opening from failing to be included in the batch and causing an infinite loop, when a node is detected to violate preset constraints when checked alone, the scheduling generation module halves the predicted time step of the corresponding execution node and re-estimates the opening. The predicted opening corresponding to the halved predicted time step is then determined as the segmented target opening for that node's current action. This ensures that the execution node stops after actually reaching the segmented target opening, rather than running all the way to the final target opening in the remote scheduling command. The scheduling generation module pre-sets the minimum predicted time step and the maximum number of segments. The minimum predicted time step is determined based on the shortest control cycle of the master controller and the minimum refresh rate limit of the fieldbus; the maximum number of segments is calibrated based on the ratio of the total travel time of the door / window to the minimum predicted time step and the set upper limit of control overhead.

[0060] When the prediction time step has reached the minimum prediction time step or the number of segments has reached the maximum number of segments, and the node's individual action still cannot meet the upper limit of absolute pressure difference or the upper limit of pressure difference gradient, the scheduling generation module stops further reducing the prediction time step, marks the node as unschedulable under the current boundary conditions, maintains its current opening, and reports the constraint unsatisfiability status to the cloud platform.

[0061] S24. Time Parameter Allocation and Scheduling Matrix Encapsulation. After completing all groupings, the scheduling generation module assigns independent start-up delay parameters to each action batch. The start-up delay of the first action batch is set to zero or a preset base delay, which is set based on the inherent delay of the node controller's hardware wake-up and instruction parsing, for example, set to 50ms to 100ms. The start-up delay of the subsequent action batch is at least equal to the start-up delay of the previous action batch, the longest predicted running time within the previous batch, the area pressure response observation time, and the fieldbus communication margin time. The longest predicted running time within the previous batch is calculated based on the segmented displacement of each execution node in that batch and the lowest available movement speed after load correction. The area pressure response observation time is determined based on the time required for the building area pressure difference change rate to enter a stable range from the time the doors and windows reach the segmented target opening during the controlled area batch action test. The running times of adjacent batches are allowed to overlap only when the scheduling generation module merges the simultaneously running nodes of multiple batches for prediction and confirms that the merged actions still meet all pressure difference and message load rate constraints.

[0062] The scheduling generation module packages and merges the node identifier, start delay parameters, action batch number, controlled area, command number, auto-incrementing scheduling version number, and lease duration of the node's running period, along with the segmented target opening degree, action direction, target opening degree allowable error, and maximum allowable running time of this action, to generate the initial scheduling matrix of one-dimensional array structure data.

[0063] Reference Figure 5 During the execution of step S30, which involves scheduling data activation and node execution, the system establishes a closed-loop soft real-time control system through asynchronous activation via master-slave collaboration. This includes the following sub-steps: S31. Clock-independent scheduling preloading and unified activation. The master controller sends the initial scheduling matrix to the corresponding execution nodes through the instruction delivery module. The node controller of the execution node parses the initial scheduling matrix and stores the start delay parameters, scheduling version number, lease duration, segment target opening, action direction, target opening allowable error, and maximum allowable running time contained in the initial scheduling matrix into its local memory to complete the data preloading. It then sends a successful loading response frame back to the master controller. During this preloading phase, the node controller does not output drive pulse signals.

[0064] Once the master controller confirms that all online execution nodes included in the current scheduling matrix have returned a successful loading response frame with consistent command number, scheduling version number, and data integrity verification result, the instruction issuing module directly broadcasts a unified activation instruction containing the command number and scheduling version number to the fieldbus. If any execution node fails to return a successful response within the loading confirmation period, the master controller must not activate the scheduling matrix containing that node, but should mark that node as having failed to load and regenerate the scheduling matrix, or terminate the scheduling task of the corresponding node.

[0065] Each execution node eliminates communication delay differences, using the moment of receiving the unified activation command as the timing start point, and starts its internal timer for local delay countdown based on the startup delay parameters. Through a mechanism that separates data preloading from unified broadcast triggering, the network queuing transmission delay error caused by serial data transmission based on node addresses on the fieldbus is effectively reduced.

[0066] S32. Network outage deadlock protection based on dynamic leases. When the internal delay countdown reaches zero, the node controller determines the status of the internally maintained scheduling lease. During system operation, the master controller broadcasts lease renewal frames to the fieldbus at set intervals via the command issuing module.

[0067] The set period is a fixed time interval, which is set based on the total time constraint of the master controller performing global state polling. The total time of global state polling is calculated and calibrated jointly by the system communication baud rate and the total number of network nodes. The fixed time interval is set to be less than half of the lease duration to provide communication retransmission redundancy. The lease renewal frame carries the currently globally effective scheduling version number. After receiving the lease renewal frame, the node controller compares the scheduling version number carried in it with the locally pre-loaded scheduling version number. If they match, the local lease countdown timer is refreshed.

[0068] When the local delay countdown reaches zero and the scheduling lease is within its effective duration, the execution node outputs a drive pulse signal to drive the local motor to perform the action. After the motor starts, the node controller continuously judges the progress of the action based on the actual opening degree fed back by the position unit; when the actual opening degree enters the allowable error range of the segmented target opening degree, the drive stops. When the running time reaches the maximum allowable running time but still does not enter the allowable error range, the node controller stops the drive, sets the node action timeout fault flag, and reports the abnormality to the main controller.

[0069] The scheduling version number, command number, and controlled area together constitute the scheduling task identifier. The execution node only compares scheduling versions that match its own node identifier, command number, and controlled area; version numbers of different commands or different controlled areas do not substitute for each other. If the execution node fails to receive a valid lease renewal frame within the set lease duration, causing the scheduling lease to expire, or receives a lease renewal frame carrying a higher scheduling version number matching its current scheduling task, or receives an instruction that the current scheduling version is marked as invalid and does not simultaneously receive a valid scheduling hold instruction, the node controller immediately cuts off the output of the drive pulse signal, forcibly locks the local motor control authority to enter a safe suspension mode, and prevents doors and windows unsupervised by the main controller from continuing to operate, which could lead to mechanical jamming or air pressure imbalance.

[0070] The scheduling hold instruction is used to atomically revoke the permissions of nodes that have not yet started to start based on the current version, while carrying a hold disposal flag, hold authorization duration, and safe action endpoint for nodes that are already running. Within the hold authorization duration, nodes that are already running will execute control according to the hold disposal flag to continue to the current segment target, stop, or move towards the safe action endpoint; after the hold authorization expires, they will stop unconditionally.

[0071] Reference Figure 6 In the process of performing step S40, load residual calculation and unsteady-state classification, the system extracts features reflecting changes in physical boundaries and classifies risk levels through node-side load sensing and multi-source data fusion, including the following sub-steps: S41. Microscopic aerodynamic load electrification calculation. During motor operation, the execution node acquires node operating parameters, specifically by collecting its actual opening degree and operating speed through the position unit, collecting the motor's operating temperature through the temperature unit, and obtaining the actual supply voltage as a power supply parameter through the analog-to-digital converter interface in the feedback loop of the drive circuit. Based on the acquired actual supply voltage and motor operating temperature, the node controller queries and obtains the corresponding voltage compensation coefficient and temperature compensation coefficient, and multiplies the real-time operating current collected by the current unit by the voltage compensation coefficient and temperature compensation coefficient to obtain the normalized operating current. Both compensation coefficients are obtained through node calibration tests consistent with the installation position of the temperature unit, and are set to 1 at the rated supply voltage and standard reference temperature.

[0072] Then, the node controller reads the internally stored two-dimensional calibration table to obtain the basic operating current under the current actual opening degree and operating speed conditions without external wind resistance. Finally, it subtracts the basic operating current from the normalized operating current to obtain the motor load residual and feeds it back to the fieldbus. When the actual power supply voltage or operating temperature exceeds the effective calibration range of the compensation coefficient, the motor load residual is not calculated further, the corresponding data is marked as invalid, and node sensing anomaly handling is triggered.

[0073] S42. High-Frequency Feature Extraction and Status Feedback. Since the real-time operating current of the motor includes high-frequency commutation ripple and mechanical vibration noise, the node controller uses an internally configured digital low-pass filter to smooth the motor load residuals, eliminating meaningless data under starting surges, commutation dead zones, and extreme end conditions. The cutoff frequency of the digital low-pass filter is set between 2Hz and 5Hz to filter out high-frequency electrical noise and retain low-frequency load characteristics caused by ambient wind pressure. The cutoff frequency range is calibrated based on the dominant frequency bandwidth of ambient wind pressure extracted from building wind tunnel tests. The digital low-pass filter can be implemented using the Butterworth filter design method with parameter configuration and difference equations. After filtering, the execution node uses the smoothed motor load residuals, along with the peak value and rate of change calculated within a sliding time window, as abnormal condition tags, encapsulating them in a response message and feeding them back according to the polling sequence of the fieldbus. The master controller receives the node feedback through the status determination module.

[0074] S43. Non-steady-state classification determination of multi-source data fusion. After receiving feedback from nodes through the status determination module, the main controller will analyze the situation when the ambient air pressure is abnormal, as multiple doors and windows will be subjected to unexpected wind resistance impacts simultaneously. Therefore, the status determination module will use spatial clustering characteristics for analysis and perform integrated calculations on the motor load residuals uploaded by each execution node according to the controlled area. The main controller will calculate the normalized regional residual index, which is defined as the ratio of the sum of the absolute values ​​of the motor load residuals of each execution node that is currently running and whose residual data is valid within the controlled area to the sum of the rated operating currents of the corresponding motors.

[0075] When only one execution node's motor load residual exceeds the node anomaly threshold, or when that node simultaneously reports stall, overcurrent, anti-pinch, sensor malfunction, or position instability faults, the status determination module identifies it as a local node fault, stops and isolates that node, and does not conclude that the controlled area has entered a local unsteady state. Only when at least two execution nodes, or a preset regional node ratio, simultaneously exceed the node anomaly threshold, and the normalized regional residual index exceeds the set load safety threshold, is the regional load anomaly condition considered met. The node anomaly threshold is determined based on the maximum load fluctuation boundary of the motor under normal, uninterrupted wind pressure disturbances, for example, set to 20% of the motor's rated current; the preset regional node ratio is set based on the total number of doors and windows and their spatial distribution within the controlled area to filter out misjudgments caused by individual mechanical faults, for example, set to 30% of the total number of doors and windows in the area.

[0076] After determining the anomaly, the status determination module combines the air conditioning static pressure data and the building area pressure difference for graded judgment. When the area load anomaly conditions are met, or the absolute value of the air conditioning static pressure deviation exceeds the set allowable deviation, and the inverter status display device transmitted by the air conditioning controller is in a saturated state, indicating that the HVAC equipment has lost its environmental regulation capability, the status determination module determines that the system has entered a local unsteady state. The load safety threshold is calibrated based on the extreme value of motor stress disturbance under normal airflow regulation of the building ventilation network, and is set to 10% to 15% of the total rated operating current of motors in the area; the allowable deviation is set based on the control dead zone parameters of the air conditioning equipment, for example, set to ±20Pa.

[0077] Simultaneously, the state determination module continuously extracts the building area differential pressure feedback from the differential pressure unit and calculates the rate of change of differential pressure per unit time between two adjacent global synchronization time windows. When the building area differential pressure or the rate of change of differential pressure exceeds the preset safety boundary, the system is determined to enter a global critical state. The preset safety boundary includes the maximum static pressure extreme value used to evaluate the building area differential pressure, which is set based on the physical pressure bearing standards of building doors, windows, and interior partitions, for example, 50 Pa; the preset safety boundary also includes a safety boundary rate of change threshold used to evaluate the rate of change of differential pressure per unit time, which is experimentally determined based on the human body's tolerance gradient to eardrum pressure, for example, set at a rate of change of differential pressure reaching 15 Pa / s.

[0078] See Figure 7 During the execution of step S50, which involves scheduling maintenance and physical truncation in response to abnormal conditions, the system issues instructions based on different risk levels to implement precise physical intervention on the underlying door and window actions, including the following sub-steps: S51. High-priority scheduling hold instruction issuance and timing freeze. In response to the triggering of local non-steady state or global critical state, the master controller, through the exception control module, forcibly revokes the start permissions of nodes that have not yet started in the current scheduling version based on the scheduling version number, freezes subsequent action batches, and issues a scheduling hold instruction with high communication priority. This instruction internally encapsulates the currently determined risk level status word.

[0079] To improve the real-time transmission performance of hold-and-hold instructions and reduce data queuing delays, the anomaly control module sets a higher transmission priority for hold-and-hold instructions than for normal state polling messages and log messages at the communication scheduling layer. When the fieldbus is a CAN bus, the anomaly control module allocates a reserved high-priority message identifier for the hold-and-hold instructions and uses the CAN bus arbitration mechanism to send them preferentially. When the fieldbus is a Modbus RTU bus, the master controller stops generating new normal polling transactions after detecting an anomaly; for currently transmitting messages, it waits for the message transmission to complete and the frame interval specified by the protocol to end before sending a broadcast write instruction at address zero as the next transaction, causing all execution nodes to write to the hold-and-hold state. The broadcast write instruction is not used to interrupt currently transmitting messages, nor does it directly clear the receive buffer of the execution nodes. After the broadcast is completed, the master controller reads the hold-and-hold status register of each execution node through sparse unicast polling to confirm the execution result of the hold-and-hold instruction.

[0080] Upon receiving the schedule hold instruction, the execution node that has not yet started has its node controller forcibly stopped counting down, stores the existing remaining delay parameters as historical records, and prohibits the starting of the motor according to the current schedule version.

[0081] S52. Asymmetric physical blocking control based on state hierarchy. The node controller of each running execution node parses the risk level status word encapsulated in the received scheduling and holding instructions, and executes the corresponding physical processing logic according to the risk level indicated by the risk level status word.

[0082] When the status word indicates that the system is in a locally unstable state, the control logic for continuing the action of the running execution node is executed. Specifically, the running execution node only continues to complete the current segment action and does not continue to the final target opening degree in the remote scheduling instruction; after reaching the target opening degree of the current segment, it stops and waits for reconfiguration of the scheduling. If the node triggers overcurrent, stall, anti-pinch, overtemperature, or maximum running time protection during the completion of the action path, it stops immediately and does not continue the action. This control logic prevents frequent opening and stopping of doors and windows caused by ordinary fluctuations due to air conditioning response lag, thereby protecting the mechanical structure of the motor.

[0083] When the status word indicates that the system is in a global critical state, the running execution node performs blocking and recovery control under the global critical state. The anomaly control module calculates the distance change between the predicted pressure difference and the preset pressure target interval of each affected area when each running node continues to run in the current direction to the next segment target, based on the current building area pressure difference, pressure difference change rate, facade where doors and windows are located, effective opening increment, and area pressure difference prediction model.

[0084] If continuing the current action can bring the predicted differential pressure of all affected areas close to the corresponding pressure target range, and will not cause any area's absolute differential pressure boundary or differential pressure gradient boundary to exceed the limit further, the action is determined to be in the direction of pressure recovery; otherwise, it is determined to be in the direction of pressure deterioration. The anomaly control module sends a continue flag or stop flag and a safety action endpoint to the corresponding execution node through a scheduling hold command.

[0085] When the movement of doors and windows is determined to indicate a deterioration in pressure, the node controller immediately stops outputting drive pulse signals to the drive circuit. When the drive circuit and motor support short-circuit braking, the node controller uses short-circuit braking; when the drive circuit or motor does not support short-circuit braking, the node controller cuts off the drive output and maintains the current position through a motor self-locking structure, mechanical brake, or electromagnetic brake to limit further airflow into or out of the controlled area.

[0086] When the movement of a door or window is determined to be in the direction of pressure recovery, the node controller maintains the output of the drive pulse signal, allowing it to continue operating to the safe set opening degree within the authorized holding time. The safe set opening degree is a fixed opening value preset based on the mechanical limit requirements of the door and window and the pressure safety requirements of the controlled area. The actuator controls the opening degree appropriately to suppress system load dispersion and accelerate the recovery of environmental steady state. When the differential pressure data of the building area is invalid and the direction of the movement cannot be determined, the running actuator stops driving and maintains its current position, no longer presuming its safety based solely on the opening or closing direction.

[0087] During the execution of step S60, system stabilization assessment and closed-loop scheduling reconfiguration, after blocking abnormal actions, the system relies on the environment itself and the adjustment capabilities of HVAC equipment to restore air pressure balance, and then continues to complete the remaining control actions within the safety boundary, including the following sub-steps: S61. Inverter State Mapping and System Settling Time Calculation. In the state of scheduling hold, the system does not employ static trial recovery but instead transitions to a dynamic adaptive calculation phase. The main controller continuously monitors the air conditioning static pressure deviation and inverter status fed back by the air conditioning controller through the closed-loop reconfiguration module. Based on the absolute value of the air conditioning static pressure deviation, the absolute value of the rate of change of the air conditioning static pressure deviation, the current output frequency included in the inverter status, and the inverter's limiting saturation state, the main controller interpolates the estimated settling time from the built-in parameter matrix.

[0088] The index of the air conditioner's expected settling time parameter matrix includes at least the air conditioner static pressure deviation range, the air conditioner static pressure deviation change rate range, the inverter output frequency range, and the limiting state. Each entry records the upper quantile of the time required for the air conditioner static pressure deviation to reach the steady-state error bandwidth under the corresponding operating condition. The pneumatic additional waiting time table uses the interval containing the normalized region residual index as its index, and the entries do not decrease as the region residual index increases. Linear interpolation is used when the input is between adjacent intervals, and the most conservative boundary entries are used when the input exceeds the calibration range.

[0089] Simultaneously, the closed-loop reconfiguration module maps the combined residual value of the controlled area to the minimum aerodynamic additional waiting time. The closed-loop reconfiguration module then calculates the system settling time. When the air conditioning controller communication is interrupted or the data is invalid, the expected settling time is set to zero. The expression for calculating the system settling time is as follows: ; in, The calculated system stabilization time; The estimated settling time is obtained by interpolation based on the air conditioning system operating parameter matrix; The minimum aerodynamic additional waiting time is derived from the partitioned load index mapping. This refers to the communication transmission overhead time of the fieldbus; Allow sufficient buffer time for the project to accommodate fault tolerance and redundancy; This is the maximum of the expected stabilization time and the minimum aerodynamic additional waiting time.

[0090] S62. Multi-dimensional System Recovery Condition Determination. The closed-loop reconfiguration module starts its internal recovery timer, accumulating the waiting time the system spends in the scheduling hold state. System recovery conditions include normal recovery conditions and degraded recovery conditions. The closed-loop reconfiguration module compares the accumulated waiting time with the system settling time. Normal recovery conditions are applied when the air conditioning static pressure data and inverter status are valid. Normal recovery conditions include: the waiting time reaches the system settling time; each controlled area is configured with a directional pressure target range. The restored building area pressure difference should fall within this range, rather than simply judging whether the absolute value of the building area pressure difference is between 0 Pa and 15 Pa. For areas requiring negative pressure, the entire pressure target range can be negative; for areas requiring positive pressure, the entire pressure target range can be positive. Furthermore, the rate of change of building area pressure difference per unit time should fall back to within the preset safety boundary rate of change threshold; the air conditioning static pressure deviation should be continuously maintained within the preset steady-state error bandwidth during the confirmation period; and there should be no unresolved hardware faults at the relevant nodes.

[0091] When the air conditioning controller communication is interrupted but the building area differential pressure is valid, a degraded recovery condition is adopted. The degraded recovery condition no longer includes air conditioning static pressure deviation or inverter status conditions, but requires the waiting time to reach a preset minimum degraded hold time, and the building area differential pressure and its rate of change to remain within the degraded recovery boundary for multiple consecutive global synchronization time windows. The minimum degraded hold time must not be less than the maximum system stabilization time achievable in normal mode. Only when the normal recovery condition or the degraded recovery condition is met does the closed-loop reconfiguration module determine that the system meets the system recovery condition, allowing the release of the underlying scheduling hold state.

[0092] When the accumulated waiting time exceeds the preset system circuit breaker threshold and neither the normal recovery condition nor the degraded recovery condition is met, the closed-loop reconfiguration module directly determines that the current scheduled task has failed to prevent the system from getting stuck due to environmental or equipment abnormalities. It then terminates the corresponding scheduled task and reports the reason for the recovery failure and the environmental out-of-control alarm to the cloud platform. The preset system circuit breaker threshold is set to, for example, 3 times the system calming time.

[0093] S63. Remaining Displacement Extraction and Closed-Loop Scheduling Reconstruction. After determining that the system meets the system recovery conditions, the closed-loop reconstruction module initiates position polling to each execution node. The execution nodes feed back their current actual opening degree after interruption or short-circuit braking to the closed-loop reconstruction module. The closed-loop reconstruction module extracts the remaining displacement based on the target opening degree in the received remote scheduling instruction and the current actual opening degree fed back by each execution node, and discards the original remaining delay parameters saved by each execution node during the scheduling hold phase.

[0094] The closed-loop reconfiguration module, based on the altered initial physical conditions including the updated current actual opening degree and the restored steady-state building area pressure difference, treats incomplete actions as initial disturbance states and re-executes the initial scheduling generation process under pressure difference constraints to calculate and generate a reconfigured scheduling matrix. This prevents secondary pressure shocks caused by concurrent release of remaining travel and ensures that the scheduling version number within the generated reconfigured scheduling matrix is ​​automatically updated. During generation, the effective opening increment and predicted pressure difference of the controlled area within the predicted time step are also estimated, and the action batches and start-up delay parameters are re-allocated based on the absolute pressure difference upper limit, pressure difference gradient upper limit, and fieldbus message load rate upper limit. The data structure of the reconfigured scheduling matrix remains consistent with the initial scheduling matrix, containing only information on execution nodes that have not reached the target opening degree.

[0095] For the specific calculation logic of re-dividing batches and allocating parameters, those skilled in the art can refer to the initial scheduling generation process under the pressure difference constraint in step S20 for understanding. The calculation logic belongs to the same mathematical model, and will not be repeated here.

[0096] S64. Reconstruction Data Distribution and Global Closed-Loop Feedback. The master controller sends the generated reconstruction scheduling matrix to the corresponding execution nodes for loading. After loading, the master controller sends a unified activation command to the fieldbus to continue the remaining control actions. The execution nodes restart the delay countdown according to the time parameters in the reconstruction scheduling matrix, and drive the motor to run when the countdown reaches zero and a valid scheduling lease is available.

[0097] The system continuously operates through a closed-loop iterative process of impact, global truncation, fusion reconstruction, and continued execution until the instruction sequence is cleared and the final actual opening of each execution node reaches the corresponding target opening, or the number of scheduled reconstruction triggers reaches the preset maximum reconstruction iteration threshold. When the preset maximum reconstruction iteration threshold, used to avoid the system falling into an infinite retry deadlock, is reached and the target opening is still not achieved, the control loop of the corresponding node is forcibly terminated. The maximum reconstruction iteration threshold is set to 3 times.

[0098] After all remote scheduling commands have been executed or the forced exit conditions have been met, the closed-loop reconfiguration module aggregates the final actual opening degree of each execution node and generates an execution result status for each command number and execution node. The execution result status includes at least the following: execution successful, node offline, node hardware failure, invalid sensor data, pressure differential constraint unmet, command expired, system circuit breaker triggered, and maximum reconfiguration count reached. For execution nodes that have not reached the target opening degree, the system operation log also records the final actual opening degree, target opening degree, the difference in uncompleted opening degree, the last valid scheduling version, and the reason for exit.

[0099] The closed-loop reconfiguration module packages the number of unsteady-state triggers, system stabilization time logs, and node power consumption parameters recorded during system operation into a system operation log. This log is then transmitted to the cloud platform via an encrypted link, along with the final actual opening degree of each execution node, the system operation log, and the execution results status at both the command and node levels. A command is marked as successfully executed only when all necessary execution nodes included in the corresponding command have entered the target opening degree's allowable error range; commands that meet the forced exit conditions are marked as failed or partially completed. This achieves global closed-loop feedback for remote control. The node power consumption parameters are derived by the underlying execution nodes through time integration calculations based on real-time data collected during operation of the supply voltage and motor current.

[0100] The present invention uses embodiments to illustrate the principles and implementation methods of the invention. The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of the invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the invention. Therefore, the content of this specification should not be construed as a limitation of the invention.

Claims

1. An IoT-based intelligent door and window remote control system, comprising a fieldbus, multiple execution nodes, and a master controller communicating with the execution nodes via the fieldbus, characterized in that, The main controller includes: The status acquisition module is used to acquire the target opening degree, the actual opening degree of the execution node, and external environment parameters; The scheduling generation module is used to generate an initial scheduling matrix based on the actual opening degree, the target opening degree, and the external environment parameters. The instruction issuing module is used to issue the initial scheduling matrix to the execution node to drive the corresponding execution node action; The state determination module is used to determine whether a local unsteady state or a global critical state has been entered based on the external environment parameters and the running state data fed back by the execution node. An anomaly control module is used to intervene by issuing a scheduling and holding command in response to the local unsteady state or the global critical state. The closed-loop reconfiguration module is used to generate a reconfiguration scheduling matrix and continue to drive the execution node actions when the system recovery conditions are met under the execution state of the scheduling hold instruction.

2. The IoT-based intelligent door and window remote control system according to claim 1, characterized in that, The external environmental parameters include building area pressure difference, outdoor wind speed and direction, air conditioning static pressure data and inverter status; The system also includes a cloud platform that is connected to the main controller and used to issue remote scheduling commands, and an air conditioner controller that provides the static pressure data of the air conditioner and the status of the frequency converter.

3. The IoT-based intelligent door and window remote control system according to claim 2, characterized in that, The status acquisition module extracts the target opening degree by parsing the remote scheduling command and collects the actual opening degree via the fieldbus; The status acquisition module is also used for: Verify whether the current control time is within the effective time range of the remote scheduling instruction; The external environment parameters are timestamped and invalid data is removed by using a global synchronization time window. Monitor the current message load rate of the fieldbus; When a sensor malfunction is detected that is used to collect the external environmental parameters, a preset key execution node is offline, or the current message load rate exceeds a preset communication congestion threshold, the initial scheduling matrix is ​​stopped from being sent. When the communication of the air conditioning controller is interrupted but the pressure difference of the building area continues to be updated, the pressure difference constraint degradation control logic is executed to forcibly reduce the upper limit of the number of doors and windows that are allowed to act concurrently and to generate the initial scheduling matrix only based on the pressure difference of the building area.

4. The IoT-based intelligent door and window remote control system according to claim 2, characterized in that, The scheduling generation module is further used to divide the multiple execution nodes into discrete action batches based on preset constraints. Based on the preset initial prediction time step and the calibration motion speed and motion direction coefficients corresponding to the execution node, the predicted displacement increment is determined, thereby generating the expected opening and the corresponding predicted running time that are limited by the target opening, and the effective opening increment is obtained based on the expected opening. The predicted pressure difference of the controlled area is generated by combining the predicted pressure difference of the previous action batch, the effective opening increment, the expected change of the inverter output frequency extracted based on the inverter status, and the air conditioning static pressure deviation obtained based on the air conditioning static pressure data using a multivariate autoregressive model.

5. The IoT-based intelligent door and window remote control system according to claim 4, characterized in that, The initial scheduling matrix includes at least the start delay parameter and the segmented target opening degree; The scheduling generation module sorts and pushes multiple execution nodes into the current action batch for verification based on the pressure difference influence weight; If the addition of a new node causes the preset constraints to be broken, the scheduling generation module will postpone the execution node that violates the preset constraints to the next action batch, or re-verify and estimate by halving the prediction time step of the corresponding execution node and generate a smaller segmented target opening. If the predicted time step has reached the preset minimum predicted time step and still does not meet the preset constraint conditions, the scheduling generation module will mark the corresponding execution node as unschedulable under the current boundary conditions.

6. The IoT-based intelligent door and window remote control system according to claim 5, characterized in that, The instruction issuing module is also used for: Broadcast lease renewal frames according to a set period; Furthermore, the start-up delay parameters of subsequent action batches are determined based on the sum of the start-up delay parameters of the previous action batch, the longest predicted running time within the previous batch, the regional pressure response observation time, and the fieldbus communication margin time. The execution node controls the action timing according to the start delay parameter and the set lease duration, and stops outputting the drive pulse signal when the actual opening degree enters the allowable error range of the segment target opening degree or when the lease renewal frame is not received.

7. The IoT-based intelligent door and window remote control system according to claim 1, characterized in that, The operating status data includes motor load residuals; The execution node includes a motor and a node controller, and the node controller is used to acquire the actual opening degree, actual power supply voltage, operating temperature and real-time operating current. The node controller dynamically compensates the real-time operating current based on the actual power supply voltage and the operating temperature to generate a normalized operating current, and generates the motor load residual based on the difference between the normalized operating current and the basic operating current without external wind resistance, which is then fed back to the fieldbus.

8. The IoT-based intelligent door and window remote control system according to claim 4, characterized in that, The state determination module is also used for: The normalized region residual index is calculated by combining the motor load residual of the execution node running in the controlled region with the rated operating current of the corresponding motor. When the regional load abnormality condition is met or the absolute value of the air conditioning static pressure deviation exceeds the set allowable deviation and the inverter is in the limited saturation state, the system is determined to enter the local unsteady state. When the pressure difference or its rate of change in the building area exceeds the preset safety boundary, the system is determined to enter the global critical state.

9. The IoT-based intelligent door and window remote control system according to claim 5, characterized in that, The specific intervention performed by the anomaly control module is as follows: When forcibly deprecating the current scheduling version, control the execution nodes that have not yet started to stop forcibly, and control the running execution nodes to perform blocking and recovery control under the global critical state; When performing the blocking and recovery control, the abnormal control module evaluates the impact of continued action on the predicted pressure difference in the controlled area, and executes the control logic for continued action under the local unsteady state, so as to continue to complete only the action corresponding to the current segmented target opening.

10. The IoT-based intelligent door and window remote control system according to claim 2, characterized in that, The reconstructed scheduling matrix is ​​generated based on the current actual opening degree and the pressure difference of the restored building area. The closed-loop reconfiguration module, under the scheduling hold command execution state, continuously monitors the air conditioner static pressure deviation and the inverter status to determine whether the system recovery conditions are met. After the system recovery conditions are met, a position polling is initiated to the execution node to obtain the actual opening degree. The unfinished actions are treated as the initial disturbance state, and the reconstructed scheduling matrix is ​​recalculated, generated, and issued to continue the remaining control actions until the instructions are executed.

Images