An environmental regulation control system and method for an emergency rescue double-cabin extended shelter

Abstract

The present application relates to the technical field of shelter environment control, and discloses an environmental regulation control system and method for an emergency rescue double-cabin expansion shelter, which comprises a data acquisition module and a control unit; the control unit is configured with a steady-state closed-loop regulation and a transient feedforward control dual mode. When the shelter is stationary, the system performs steady-state closed-loop regulation; when the expansion action of the shelter is monitored and the kinematic parameters reach the set threshold, the transient feedforward control mode is activated, the volume variable is analyzed based on the kinematic parameters, and a feedforward compensation instruction is generated to drive the actuator, so as to offset the flow field pressure loss and thermal load mutation caused by the volume increase in advance; when the expansion action is completed, the dynamic weight stripping logic is executed, the integral anti-saturation back-calculation mechanism is combined to smoothly return to the steady-state closed-loop regulation mode, and the redundant memory resources are released. The present application overcomes the response delay of the environmental control in the process of the volume change of the shelter, eliminates the instruction jump during mode switching, and improves the regulation accuracy and operation stability of the system.

Description

Technical Field

[0001] This invention relates to the field of mobile cabin environmental control technology, specifically to an environmental regulation and control system and method for an emergency rescue dual-cabin extended mobile cabin. Background Technology

[0002] Emergency rescue expandable modular units are widely used in medical rescue, communication command, and other missions. The dual-unit expandable modular unit is transported in a collapsed state; upon arrival at the site, it unfolds via a linkage mechanism to increase the usable internal space. The unit is typically equipped with an environmental control system consisting of air conditioning and ventilation equipment to maintain the temperature and air pressure difference within the enclosed working space. Most existing environmental control systems employ conventional steady-state closed-loop control logic, using sensors to collect real-time temperature and pressure difference data within the unit, comparing this data with system setpoints, and then generating feedback control commands to drive the relevant temperature control and ventilation actuators.

[0003] This conventional control system has significant limitations when the modular shelter (MSB) is deploying. As the mechanical cabin expands outwards, the internal volume increases rapidly within a short period, directly causing transient depressurization. At this time, untreated external air rapidly flows back into the cabin under the pressure difference, causing drastic fluctuations in internal temperature and airflow. The existing closed-loop control system relies entirely on delayed feedback from internal sensors, only starting to operate after a substantial shift in the overall environmental parameters and after a complete sensor sampling cycle. This inherent physical lag in the control system's response and mechanical actions prevents it from providing sufficient thermal load handling power and air momentum compensation in advance during the sudden change in internal volume.

[0004] Furthermore, existing technologies lack a smooth transition and coordination mechanism when handling the system's operational state after the expansion of the modular shelter has ceased. When the mechanical expansion of the shelter reaches its structural limit and suddenly stops, if the control unit directly and forcibly switches control authority back to the previous steady-state closed-loop regulation loop, the accumulated computational deviations in the underlying algorithm during the previous passive following phase will lead to severe saturation of the integral term in temperature control. This forced takeover of command data will cause sudden jumps in the signals sent to the actuators, resulting in violent oscillations in the blower speed and compressor frequency. Simultaneously, the system consumes microcontroller memory stack resources when performing complex action state determination and compensation calculations. Existing systems lack a low-level cleanup mechanism for redundant data generated by such transient actions. When the shelter faces repeated expansion and contraction tests or long-term stationary tasks, memory fragmentation can easily clog the microprocessor's operating cycle. Especially under extremely inefficient conditions such as outdoor unit defrosting, the underlying microcontroller unit is prone to division-by-zero overflow or crashes due to computational parameters approaching boundary values, affecting the overall operational reliability of the equipment in emergency rescue environments. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an environmental regulation and control system and method for an emergency rescue dual-cabin extended container. This solves the problem that when existing extended containers perform structural deployment, the instantaneous increase in the internal volume of the container leads to overall flow field depressurization and backflow of unregulated external airflow. Conventional steady-state closed-loop temperature control systems suffer from physical response hysteresis and cannot preemptively offset transient thermal loads and momentum impacts.

[0006] The first aspect of the present invention provides an environmental regulation and control system for an emergency rescue dual-cabin extended modular container, comprising a data acquisition module, an actuator module, and a control unit; the data acquisition module is used to acquire internal and external environmental parameters of the container and kinematic parameters of the movable cabin body of the container; the actuator module includes a ventilation component and a temperature control component; the control unit is configured with a steady-state closed-loop regulation mode and a transient feedforward control mode. When the modular shelter is stationary, the control unit executes a steady-state closed-loop regulation mode. When the control unit detects that the modular shelter is performing an expansion maneuver and the collected kinematic parameters reach a set threshold, it activates a transient feedforward control mode. Based on the kinematic parameters, it analyzes the volume-related variables inside the modular shelter and generates feedforward compensation commands to drive the actuator module. When the expansion maneuver ends, the control unit executes dynamic weight stripping logic, gradually exits the transient feedforward control mode, and returns to the steady-state closed-loop regulation mode.

[0007] The transient feedforward control mode includes transient mass compensation logic: the control unit represents the volume-related variable as the instantaneous volume change rate inside the cabin, and encapsulates the real-time motion speed, acceleration, measured temperature inside the cabin, pressure difference between inside and outside the cabin, and ambient temperature into a state vector for dimensionality reduction and normalization; the processed state vector is input to the long short-term memory neural network model for forward inference to obtain the target frequency feedforward of the blower and the target opening feedforward of the exhaust valve in the ventilation component; the output result is algebraically summed and limited with the feedback value of the steady-state closed-loop operation and then sent to the ventilation component.

[0008] The training process of the long short-term memory neural network model is as follows: sensor feedback data induced by flow field fluctuations under simulated traction deformation conditions are selected as input feature samples; a mean square error loss function is applied, and a regularization term with a quadratic penalty constraint parameter is embedded in the loss function to constrain the absolute range of the interconnection weights of each nonlinear neural node.

[0009] For multiple node blowers in array mode, the control unit executes control logic based on momentum allocation: based on ambient air density and system pipeline flow cross-sectional area, the target frequency feedforward is converted into a global air momentum flow rate constraint equation benchmark; the node utility mapping array constructed based on the physical flow resistance parameters of the pipeline branches is retrieved, and the current power frequency of each blower is read back in real time; a diagonal weight penalty matrix is ​​constructed by comparing the rated maximum frequency of the blowers; and the frequency distribution sequence of each node blower is calculated by combining the node utility mapping array and the diagonal weight penalty matrix through a pseudo-inverse matrix solving algorithm with damping parameter constraint terms.

[0010] The transient feedforward control mode also includes transient energy control logic: the control unit extracts the gradient difference benchmark between the ambient outside temperature and the cabin reference temperature; it performs a product operation using the equivalent fluid displacement rate derived from the volume change rate, ambient air density, constant pressure heat capacity coefficient, and the gradient difference benchmark to obtain the absolute power demand for transient heat load; it obtains the cycle efficiency ratio parameter through the real-time cycle efficiency two-dimensional interpolation table of the addressing device, and converts the absolute power demand for transient heat load into an equivalent drive execution quantity for the action of the temperature control component.

[0011] When calculating the equivalent drive execution quantity, a hysteresis dual-judgment dead zone logic is preset: when the amplitude of the absolute demand power of the transient heat load exceeds the set dead zone threshold and the direction of the heat deviation remains in the same polarity, the authority to issue the equivalent drive execution quantity is activated; for the calculation process with the cycle energy efficiency ratio parameter as the denominator, an overflow-preventing minimal constant constraint term is added for replacement calculation.

[0012] The process of the control unit executing the dynamic weight stripping logic is as follows: when the measured motion rate decreases below the shock detection threshold and the state maintains the set continuous judgment period, the current natural timestamp is locked as the retreat origin; the monotonically decreasing real-time control weight factor is calculated according to the cosine half-period function decay algorithm, and the feedforward control weight is smoothly decayed from the peak limit to zero within the set buffer time.

[0013] During the execution of the dynamic weight stripping logic, the control unit executes truncation and inverse calculation rules: freezing the original error integral term of the steady-state closed-loop control mode; reading the current driving quantity actually sent to the actuator module, and inversely calculating the internal equivalent cumulative value corresponding to the current output state through the transfer function; and overwriting the frozen error integral term using the internal equivalent cumulative value. When the control weight factor decays to zero, the memory resources occupied by the variance comparison array, feature buffer pool, and transition stage pointer stack used in the transient feedforward control mode are released.

[0014] The second aspect of this invention provides an environmental regulation and control method for an emergency rescue dual-cabin extended modular container, applied to the aforementioned system. The method includes: acquiring internal and external environmental parameters of the container and kinematic parameters of the movable cabin; when the container is stationary, operating a steady-state closed-loop regulation mode for temperature control; when the container is detected to be performing an extension action and the kinematic parameters reach a set threshold, switching to a transient feedforward control mode; analyzing the predicted negative pressure and corresponding transient heat load caused by changes in the container's internal volume based on characteristic parameters, and generating feedforward compensation commands to drive the ventilation and temperature control components; when the action reaches a stopping point, performing a smooth handover of control authority based on the internal integral quantity back-calculation overwrite and cosine decay dynamic weight stripping logic, returning to the steady-state closed-loop regulation mode, and clearing the memory resources occupied by the transient mode.

[0015] This invention provides an environmental control system and method for an emergency rescue dual-cabin extended modular shelter. It has the following beneficial effects: 1. This invention employs a dual-mode control framework that combines steady-state closed-loop and transient feedforward. When the container undergoes expansion, the control system pre-calculates the internal volume change rate by collecting kinematic parameters of the moving cabin and directly generates feedforward compensation commands based on multi-dimensional environmental data to drive the ventilation and temperature control mechanisms. This control mode proactively responds to potential negative pressure deficits and heat load demands before sudden volume changes, effectively overcoming the physical response lag caused by measurement and mechanical inertia in traditional steady-state feedback control, and reducing the drastic temperature and pressure fluctuations caused by external air backflow during rapid volume increases.

[0016] 2. This invention employs an anti-oscillation dynamic weight stripping logic and an integral anti-saturation mechanism. When the system detects that the extended action has stopped, the control unit smoothly reduces the feedforward control ratio within a set time period using a cosine decay function. It then uses the actual hardware drive quantity sent to the execution port at this time to reverse-engineer the closed-loop deviation, using this value to overwrite and erase the frozen historical error integral term in the steady-state closed-loop regulation loop. Because the internal calculation state is forced to be homomorphically aligned with the actual external mechanical operating state at the switching node, the parameter integral saturation phenomenon and equipment execution command jumps that are prone to occur during control handover are completely eliminated.

[0017] 3. This invention leverages the stability of the underlying hardware and software system during long-term operation of the control module. By embedding a minimal constant constraint term in the denominator of the heat load conversion calculation, the system automatically switches to constant control for division operations when the outdoor unit enters defrosting mode, causing the energy efficiency ratio to drop to zero. This directly prevents microprocessor crashes due to arithmetic division by zero overflow. Furthermore, once the system returns to a stable, closed-loop control state, the control module proactively clears and resets the feature comparison buffer pool and pointer stack memory resources used for transient feedforward calculations according to the first-in-first-out rule. This prevents memory overflow lock-up faults caused by the long-term accumulation of temporary redundant data fragments. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the environmental regulation and coordinated control system of the emergency rescue dual-cabin extended container according to an embodiment of the present invention; Figure 2 This is a flowchart of the environmental regulation and coordinated control method according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating the signal feature extraction and operating condition determination in an embodiment of the present invention. Figure 4 This is a steady-state closed-loop regulation diagram according to an embodiment of the present invention; Figure 5 This is a flowchart of the transient quality compensation logic in an embodiment of the present invention; Figure 6 This is a flowchart of the momentum distribution control logic according to an embodiment of the present invention; Figure 7 This is a flowchart of the transient energy feedforward control logic according to an embodiment of the present invention; Figure 8 This is a flowchart of the control weight stripping and steady-state regression logic in an embodiment of the present invention; Figure 9 This is a comparison diagram of extended voltage drop under two control schemes in an embodiment of the present invention. Detailed Implementation

[0019] 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.

[0020] See attached document Figure 1 , Figure 1 This is a schematic diagram of the environmental regulation and coordination control system for an emergency rescue dual-compartment extended container according to an embodiment of the present invention. The present invention provides an environmental regulation and coordination control system for an emergency rescue dual-compartment extended container, which may include: a main control unit, a data acquisition module, and an actuator module.

[0021] The data acquisition module includes an absolute displacement encoder, a differential pressure transmitter, and a temperature transmitter. The absolute displacement encoder is installed at the mechanical deployment and telescopic structure of the extension compartment to acquire the physical quantity of the mechanical displacement of the extension compartment relative to the main compartment. The differential pressure transmitter is located at the inner and outer boundaries of the compartment to acquire the pressure difference data between the inside and outside of the compartment. The temperature transmitter is located inside the main compartment and the extension compartment to measure spatial thermal parameters.

[0022] The data acquisition module establishes a data connection with the main control unit through a standard industrial communication protocol. The main control unit receives the underlying basic physical parameters sent by the absolute displacement encoder, the micro differential pressure transmitter, and the temperature transmitter.

[0023] The actuator module includes a supply fan, an exhaust fan, a frequency converter, a main compartment variable air volume (VAV) valve, an expansion compartment VAV valve, a fresh air electrically operated airtight valve, an exhaust electrically operated airtight valve, a compressor, and hot and cold water valves. The output of the main control unit establishes a control connection with the actuator module.

[0024] The frequency converter reads the frequency commands issued by the main control unit and adjusts the speed of the supply fan and exhaust fan. The main control unit outputs analog commands to adjust the opening degree of the fresh air electric shut-off valve, the exhaust air electric shut-off valve, the main compartment variable air volume valve, and the expansion compartment variable air volume valve, respectively. The main control unit outputs control flow to adjust the compressor load rate and the opening degree of the hot and cold water valves.

[0025] Based on the aforementioned hardware architecture, the control unit of this invention incorporates a steady-state closed-loop regulation mode and a transient feedforward control mode. During operation, the data acquisition module continuously acquires environmental parameters both inside and outside the shelter, as well as the kinematic parameters of the shelter's moving parts, and transmits them to the control unit. When the shelter is stationary, the control unit executes the steady-state closed-loop regulation mode. When the shelter is detected to be performing an expansion maneuver and the kinematic parameters reach a set threshold, the control unit activates the transient feedforward control mode. Based on the kinematic parameters, it analyzes the volume-related variables inside the shelter and generates corresponding feedforward compensation commands to drive the ventilation and temperature control components located within the shelter in the actuator module.

[0026] Furthermore, when executing the transient feedforward control mode, the control unit is configured to simultaneously execute transient mass compensation logic, momentum-based load balancing protection logic, and transient energy control logic with safety judgment constraints. When the extended action is detected to be over, the control unit executes dynamic weight stripping logic to gradually exit the transient feedforward control mode and return to the steady-state closed-loop regulation mode. During this process, the control unit also simultaneously executes an anti-integral saturation mechanism to achieve a disturbance-free switching, and initiates a memory cleanup mechanism when the buffer duration ends to release the transitional cache resources occupied by the transient feedforward control mode.

[0027] See attached document Figure 2 This invention provides a method for coordinated environmental regulation and control of an emergency rescue dual-cabin extended container, comprising the following steps: S10, Signal Feature Extraction and Operating Condition Determination: Read the absolute displacement encoder data and generate the effective displacement through moving average filtering; obtain the effective speed and effective acceleration through discrete difference calculation. When the absolute value of the effective speed is less than or equal to the structural anti-shake speed threshold, it is determined to be a static parking condition, and step S20 is executed; when it is greater than the threshold, it is determined to be an extended operating condition, the closed-loop feedback of S20 is suspended, and steps S30, S40 and S50 are triggered synchronously.

[0028] S20, steady-state closed-loop regulation: Under parking conditions, based on temperature and micro-pressure difference feedback measurements, a PID algorithm with anti-integral saturation is applied to output regulation commands, driving the actuators to maintain the cabin reference temperature and micro-positive pressure.

[0029] S30, Transient Mass Compensation Control: Under extended operating conditions, kinematic parameters and thermodynamic parameters inside and outside the cabin are encapsulated and normalized, input into a Long Short-Term Memory Neural Network (LSTM) model, and the opening degree of the exhaust air sealing valve and the frequency conversion feedforward command are deduced.

[0030] S40, Momentum Distribution Load Balancing Predictive Control: Calculates global air momentum demand based on feedforward commands, constructs an asymmetric impedance penalty diagonal matrix based on the current fan speed margin, and applies pseudo-inverse operation with damping constraints to solve for the individual command sequence of the underlying distributed fans.

[0031] S50, Energy Feedforward Interception Control: Based on the instantaneous expansion rate of the cabin volume and the internal and external temperature difference, the transient sensible heat load power is calculated. Combined with the dynamic energy efficiency ratio of the air conditioning system, a dimension reduction mapping is performed to synthesize the basic frequency enhancement feedforward component of the inverter.

[0032] S60, Smooth Transition and Steady-State Weighted Regression: After the cabin movement is detected to have ended, the feedforward control weights are gradually reduced by applying a cosine smooth decay function based on the continuity of the first derivative, and combined with the integral back calculation mechanism to achieve a disturbance-free switch to S20 steady-state closed-loop regulation, releasing redundant memory.

[0033] See attached document Figure 2 , Figure 2 This is a flowchart illustrating signal feature extraction and operating condition determination according to an embodiment of the present invention. Based on this flowchart, the environmental regulation coordinated control method of the present invention performs high-order kinematic feature extraction of the underlying sensor signals during the startup phase, and then performs the determination process of the operating condition boundary of the system. The specific process includes the following sub-steps: S110: Acquire the values ​​from the absolute displacement encoder to establish a data sequence. To ensure the timing consistency of multi-source signals, the system establishes a unified time reference. A fixed sampling period is set. ( >0), reads the raw displacement scalar data output by the displacement encoder in real time and stores it in the controller's first-in-first-out (FIFO) data queue. The length of this queue is set to a fixed parameter (preferably 10 to 50) to store continuous data within the historical sampling period, so as to balance the computing power consumption of embedded hardware with the preservation of signal characteristics.

[0034] S120, the effective displacement is obtained by applying a moving average filter. To avoid high-frequency differential noise introduced by mechanical clearance vibration during the deformation of the extended cabin, the control node applies a moving average filter algorithm to smooth the data queue.

[0035] The formula for calculating the effective displacement is as follows: ; In the formula, Indicates the current time The calculated effective displacement; This represents the storage length of the first-in-first-out (FIFO) data queue. Its value range is usually determined based on the mechanical inertia time constant characteristics of the system. In this embodiment, it is preferably a constant integer between 10 and 50. Indicates the current calculation time; This indicates the fixed sampling period of the system, and satisfies Physical limitations; Indicates at time Read the raw displacement scalar data; The dimensionless index parameter is used to backtrack the time of the summation operation. By configuring the above smoothing layer, the system can effectively isolate control misleading signals caused by external wind load fluctuations or internal personnel disturbances at the physical level.

[0036] S130: Based on the effective displacement, discrete difference operations are performed to obtain kinematic parameters. After acquiring smooth and continuous spatial position data, the system needs to further extract the transient change characteristics of the physical structure to support forward compensation calculations. Based on the general mathematical principles of discrete state-space observation, the controller extracts the effective displacement at the current moment and the effective displacement recorded in the previous sampling period, and calculates the effective velocity of the mechanical structure's extended motion using a first-order backward difference algorithm. The effective velocity data from the previous sampling period is further acquired, and the effective acceleration is obtained through difference operations with the effective velocity at the current moment. The kinematic parameters obtained directly reflect the instantaneous rate and degree of abruptness of the change in air volume inside the cabin.

[0037] The formula for calculating effective speed is as follows: ; In the formula, Indicates the effective speed at the current moment; This represents the effective displacement obtained in the previous sampling period. It should be noted that, due to... As a rigid non-zero constant driven by the underlying crystal oscillator clock, this division operation logic is naturally immune to the risk of computing power overflow caused by the denominator approaching zero.

[0038] The formula for calculating effective acceleration is as follows: ; In the formula, Indicates the effective acceleration at the current moment; This represents the effective velocity obtained in the previous sampling period. The derived effective velocity and effective acceleration constitute the core boundary conditions and input reference scalar of the subsequent transient feedforward control loop.

[0039] S140 compares kinematic parameters with structural thresholds to determine the current operating mode. For determining the switching boundary of operating conditions, relying solely on single-point extreme values ​​is prone to false triggering. A pre-set structural anti-shake speed threshold and a corresponding continuous confidence time window are used, their values ​​determined jointly by the measured drag parameters of the wind tunnel and the inherent temperature drift error of the displacement encoder.

[0040] when When the value is less than or equal to the threshold, the extended cabin is determined to be in a static parking condition without extension deformation, and the system maintains the internal steady-state closed-loop regulation loop.

[0041] when If the value exceeds the threshold and the crossing state is maintained stably within a continuous confidence time window, the container is determined to be in a dynamic deformation condition. The system suspends the execution right of the steady-state closed-loop regulation loop and simultaneously triggers the transient intervention process.

[0042] See attached document Figure 4 , Figure 4 This is a steady-state closed-loop control diagram according to an embodiment of the present invention. As a preferred implementation, this execution process aims to compensate for environmental parameter drift caused by natural heat conduction and conventional leakage within the extended container, maintaining the reference set temperature and slight positive pressure parameters inside the container. The control logic specifically includes the following sub-steps: S210 retrieves the measured values ​​of differential pressure and temperature feedback. After establishing a global soft clock for timestamp alignment and resampling, the system reads the measured temperature inside the cabin and the measured differential pressure inside and outside the cabin according to a fixed inspection cycle.

[0043] S220 incorporates feedback measurements into the closed-loop control algorithm for deviation calculation. To establish a digital drive source for the actuator's actions, it's necessary to quantify the mathematical difference between the current physical state and the ideal state. Based on the engineering standards for constant-temperature cleanrooms, the controller's internal safety register presets the system's baseline temperature and target micro-positive pressure parameters. The algorithm's core extracts the aforementioned feedback measurements and calculates the difference between them and the preset target values, thereby deriving the instantaneous control deviation scalar for each controlled physical quantity.

[0044] The formulas for calculating temperature deviation and pressure deviation are as follows: ; ; In the formula, Indicates the first Temperature deviation value for each sampling period; The reference set temperature represents the system setting temperature, the value of which is constrained by the physiological thermal comfort standards of the personnel stationed in the cabin. In this embodiment, it is preferably set between 22 degrees Celsius and 26 degrees Celsius. Indicates the first Each sampling cycle receives and analyzes the feedback measurement value obtained via a temperature transmitter. Indicates the first Pressure deviation value for each sampling period; This indicates the target micro-positive pressure parameter agreed upon by the system. The purpose of setting this value is to prevent unfiltered external air from seeping back into the cabin through gaps. It is usually set between ten and thirty Pascals. Indicates the first Measured values ​​of cross-boundary micro-pressure differential pressure in each sampling period; These are the discrete-time sequence index labels. The generated sequence of deviation values ​​constitutes the direct driving force for subsequent damping operations.

[0045] S230 uses a discrete PID calculation model with anti-integral saturation condition clamping to generate the output. To prevent the system response from being delayed due to oversaturation of the integral depth caused by the inability to pull back the deviation when the system is in an open structure, the arithmetic kernel embeds conditional freeze accumulation logic; if it is detected that the output of the previous action cycle is already at the edge of the limit full scale, the current residual is temporarily blocked from sliding into the integral accumulation area.

[0046] The mathematical description formula incorporating the concept of anti-integral overinflation correction is as follows: ; In the formula, Indicates the controller in the The output control quantity of the corresponding controlled loop is calculated from each sampling period; This variable represents the input deviation value at the current moment, and corresponds to the value generated in the previous step. or ; This represents the proportional gain coefficient, which is used to provide the basic driving force for the correction response; This represents the integral time coefficient, used to eliminate steady-state error; The index sequence number for the traversal index sequence accumulated by summing the integral; This parameter represents the discrete sampling control step size of the system. Since this parameter is rigidly configured by the underlying clock crystal oscillator and forced to be greater than zero, it avoids the risk of computing power collapse caused by the denominator of the differential loop approaching zero. This represents the differential time coefficient, used to suppress overshoot fluctuations of the controlled parameter; This represents the input deviation value from the previous sampling period. In this embodiment, considering the characteristics of the large hysteresis thermodynamic system in the modular shelter, 、 、 The initial optimization of the three gain coefficients was determined by testing the embedded Ziegler-Nichols empirical curves and burning them into the dead-zone memory at the factory.

[0047] S240, regulating the execution component matrix. The communication stack translates the regulation scalar into the corresponding node instructions for hardware action. For the temperature control closed-loop sequence, it drives the variable frequency compressor and water flow valve to perform hot and cold medium replacement, or sets the electric heater / four-way reversing valve for heating. Corresponding to the micro-positive pressure feedback channel of the flow field, the pulse width and amplitude of the regulation signal perform cross-action on the fresh air valve and exhaust valve, creating a surface pressure effect where the intake air velocity exceeds the exhaust air velocity to retain air mass and achieve pressure differential support.

[0048] See attached document Figure 5 , Figure 5 This is a flowchart of the transient mass compensation logic according to an embodiment of the present invention. As a preferred embodiment, the core technical objective of this part is to overcome the problem of sudden pressure increase or collapse instability caused by sudden changes in cabin volume in traditional hysteresis feedback control. It achieves dynamic decoupling between physical space changes and thermodynamic mass replenishment by extracting kinematic parameters to establish a forward-looking compensation feedforward. The transient compensation control logic specifically includes the following sub-steps: S310 establishes a transient physical mapping of the internal volume expansion of the modular container based on kinematic parameters. During the mechanical extension of the modular container, the volume of the sealed space inside the container expands linearly or nonlinearly with the extension displacement. This physical expansion directly triggers a transient decrease in the internal gas density. To quantify the negative pressure coupling disturbance generated by this disturbance source on the flow field, the calculation unit extracts the system's preset standard cross-sectional area parameters of the modular container and, combined with the effective velocity calculated in the aforementioned process, deduces the current instantaneous volume change rate.

[0049] The formula for calculating the instantaneous rate of change of volume is as follows: ; In the formula, Indicates the current time The instantaneous volume change rate inside the shelter calculated by the control system; This parameter represents the effective expanded longitudinal cross-sectional area of ​​the mobile cabin, and is determined by the geometric dimensions of the cabin's mechanical drawings from the factory. This indicates the effective speed of the cabin extension at the current moment; This represents the dynamic air leakage compensation coefficient at the door gaps and transmission mechanism. This value is set based on the nonlinear leakage induced by the deformation of the sealing strip during the extension process, and in this embodiment, it is preferably set within the dimensionless range of 1.05 to 1.15. The instantaneous volume change rate generated by the above formula physically represents the additional air quality reference source that must be injected to maintain the current thermodynamic and hydrodynamic steady state.

[0050] The S320 performs multi-source sensor data feature matrix encapsulation and normalization preprocessing. It encapsulates five physical scalars—effective velocity, effective acceleration, measured cabin temperature, measured cross-boundary micro-pressure difference, and ambient temperature—into a single-frame state vector, and applies a variance normalization algorithm to eliminate dimensional differences. ; In the formula, Indicates the first Normalized values ​​for each feature dimension. The value range is 1 to 5, corresponding to the five physical scalars mentioned above; This represents the original physical observations extracted from the feature matrix at the current time. and These represent the statistical mean and standard deviation of the feature variable in this dimension under historical baseline conditions, respectively. These statistical parameters are stored in non-volatile memory. The compensation amount is a very small constant, which is set to 10 in this embodiment. -6 The mathematical purpose of introducing this constant at this time is to avoid the risk of computing power overflow and singularity downtime caused by division operations when the characteristic variance approaches zero due to the long period of static operation of the mobile cabin.

[0051] The S330 introduces a multi-dimensional long short-term memory neural layer to calculate the pre-compensation vector. The computing board embeds and runs a long short-term memory neural network (LSTM), whose network architecture is as follows: it receives the input tensor of the aforementioned five feature signals, followed by two hidden layers, each with 64 neurons, and finally connects to a fully connected layer with two output dimensions. The output is directly mapped to the feedforward frequency conversion command of the blower and the feedforward opening setting value of the exhaust sealing valve.

[0052] S340 extracts the experimental calibration physical information set to feed back and improve the network's weight matrix correction. To verify and ensure the industrial re-enactment conditions for lower-level technicians, the core computing brain that performs the aforementioned forward-looking compensation inference underwent rigorous supervised training during the offline production line deployment period. The specific training set was extracted from: the temperature and pressure boundary sensor data streams during extreme weather environment testing of a full-size modular cabin in conjunction with a push-pull rig; the corresponding benchmark truth labels are the optimal driving parameter sequence that can suppress transient pressure drop fluctuations to the minimum amplitude, obtained through laboratory gridded speed regulation iterative testing.

[0053] During the offline training phase, the mean squared error loss function with added regularization constraints is used to update the inter-layer physical connection weights, as shown in the following formula: ; In the formula, This represents the scalar evaluation value of the global loss function within a single iteration batch; Indicates the sample number index within the training batch; This represents the total number of valid samples in each training batch. The channel represents the feedforward dimension of the prediction, where channel 1 is mapped to the feedforward command of the exhaust valve and channel 2 is mapped to the feedforward command of the fresh air fan frequency converter. This represents baseline intervention label data obtained from offline wind tunnel testing and physical calibration. This represents the forward inference output of the neural network model; express The regularization penalty coefficient is preferably 0.01. Represents the set of all weights to be optimized in the model's network structure. The specific elements within the formula. The parameter sum of squares constraint term added to the formula aims to limit the divergence of individual weights, forcing the network structure to learn general features rather than memorizing overfitted isolated samples, thereby ensuring the system's generalization robustness in the face of stressful and sudden changes.

[0054] The S350 synthesizes global control commands and executes hardware resource anti-limiting scheduling. The controller adds the generated feedforward control quantity to the smoothed PID closed-loop quantity of the last frame suspended under the previous parking condition. To prevent the synthesized command from overflowing the physical boundaries of the execution element, a forced clamping constraint is executed. If the calculated quantity exceeds the limit, the limit is forcibly applied according to the nearest extreme value, and the motion buffer is refreshed simultaneously to drive the servo response.

[0055] See attached document Figure 6 , Figure 6 This is a flowchart of momentum distribution control logic according to an embodiment of the present invention. When implementing a system array with multi-branch layout and locally grid-connected wind turbines working in coordination, the forward-looking commands transmitted must not be crudely distributed along an average path. A load configuration mechanism aimed at overcoming differences caused by pipe resistance voltage division includes: S410 extracts the global fluid momentum requirement. The feedforward total opening command, combined with air density and flow area, is converted back to the global mass flow rate. This scalar constitutes the equality constraints for subsequent non-independent node allocation calculations.

[0056] S420, retrieves the hardware ledger to establish an evaluation projection matrix for the physical contribution rate. Given that each distributed blower has a separate scaling loss constant for the bifurcation resistance of the main duct, the central data system pulls the execution node utility mapping matrix group based on the underlying actual mapping storage.

[0057] Linear matrix form expressing mapping relationships: ; In the descriptive paragraph, The momentum scalar quantity required for global air transport compensation to meet the above requirements; Describes the similarity The distributed impedance segregation attenuation matrix; The equivalent aerodynamic damping multiple ratio for a single node; The final result is a planned and assigned array of plungers that are disassembled and dissolved into multiple dimensions of independent directional rotation steps.

[0058] S430 employs an asymmetric diagonal constraint setting based on the implementation margin penalty allocation. To prevent rotor demagnetization and thermal damage induced by prolonged near-full-frequency load operation of the motor located in the short air duct upstream, the core coordinator constructs a transcoding weight penalty diagonal network that adjusts based on the current idling inventory.

[0059] The calculation rules for the diagonal elements of the impedance penalty matrix are as follows: ; In the formula, The element located at the center of the diagonal of the square configuration represents the pressure weight distribution of the specified object. Used to represent the penalty-enhanced sensitivity tone scaling reference number; This indicates that the relevant motor drive module hardware data specifies the nominal maximum critical frequency peak. The current job feedback frequency is based on the underlying encoding and readback; the configured infinitesimal number. Locked to 10 -5 This operation effectively eliminates the problem of matrix operation crash caused by division by zero when a specific blower encounters a sudden overload and is locked at full speed. The logic of this formula reflects that once a module is close to physical full load, the cost of assigning any additional burden to it and requesting parameters is immediately increased exponentially.

[0060] S440 uses a pseudo-inverse solution with boundary protection buffers to perform sequence reprogramming distribution on the matrix. When the system equations are under-localized with multiple wind turbine nodes and the matrix is ​​not strictly rank (leading to overdetermined system equations or redundant solutions), the control engine rejects the traditional coarse Gaussian inverse and calls a flat, damped, penalized pseudo-inverse equation to extract the sequence output solution with the minimum comprehensive operational losses.

[0061] The core analytical equation expansion with mandatory crash-resistant generality: ; In this planning description that includes complex linear algebra: It will output the drive fine-tuning frequency band array index representing multi-axis synchronization; the simple inversion and finishing action of pushing the reciprocal of the diagonal matrix from the previous inversion operation has the guarantee of convergence continuity. The forward assignment matrix is ​​represented by its transpose; the internal Serrat-introduced independent damping constant parameters are... The mathematical injection not only ensures that the solution of the equation does not produce jump points or spikes during drastic switching, but more importantly, it effectively avoids the rank reduction of the system correlation matrix caused by the physical communication interruption of some wind turbine nodes, which in turn causes the singularity calculation failure in the inverse matrix solution process, thus ensuring uninterrupted calculation.

[0062] S450, approved for interception and instruction issuance. Before sending communication word data to the physical address, a mandatory clamping security audit is performed to ensure that the output sequence meets the thermal dissipation protection limits.

[0063] See attached document Figure 7 , Figure 7 This is a flowchart of the transient energy feedforward control logic according to an embodiment of the present invention. In response to the forced intake of external natural climate temperature accompanying airflow compensation, which inevitably disrupts the original temperature range within the bulkhead, a proactive energy injection anti-drop design must be implemented to prevent the internal air conditioning system from reactively catching up with the load. The flowchart includes the following: S510 is used to calibrate the field meteorological boundary parameters. Real-time temperatures at environmental meteorological calibration points and cabin temperatures are simultaneously extracted to establish an initial differential benchmark.

[0064] S520 applies a steady-state physicochemical model to quantify the sensible heat surge preload power caused by the rapid changes in air mass iteration. Ignoring the secondary effects of slow, long-term heat transfer and static leakage from the plate material (which the system treats as static feedback), this step specifically addresses the instantaneous dynamic load increment caused by the sudden increase in the deformation volume of the container by establishing a formula based on the first law of mass transfer to describe the compensating heat flow.

[0065] The core preset total flow equation for calculating heat load demand is: ; In the formula, surface Showing the current time node The system is expected to release or absorb additional transient heat load power required to neutralize disturbances introduced into the fresh air. This represents the air density calibration constant under local atmospheric pressure conditions. In this embodiment, it is preferably initialized to 1.205 kg / m³, taking into account the local altitude and static pressure parameters. 3 ; The specific isobaric heat capacity coefficient represents the thermal capacity of a dry air medium. This coefficient characterizes the inherent physical limitation of the fluid's isobaric heat storage capacity, and is typically taken as 1.005 kJ / (kg· ... K); This indicates the instantaneous volume change rate command issued by the preceding control link based on the speed measuring device and geometric dimensions. and These represent the external natural environment meteorological temperature points and the measured reference temperature inside the container, respectively, acquired within the same processing epoch. This formula, by converting dynamic volumetric flow rate into a mass flow rate reference and integrating the enthalpy gradient inside and outside the system, quantifies the absolute amplitude of heat load fluctuations caused by transient fluid exchange from a physical perspective.

[0066] The S530 maps the theoretical heat load power to an operational-level target quantity for the terminal electromechanical equipment. After obtaining the absolute energy demand calculated from macroscopic physical formulas, the controller must further characterize it as the duty cycle or Hertz value that can directly drive the silicon rectifier bridge of the air conditioner inverter or PTC heater. To avoid frequent start-ups and shutdowns of the hardware unit caused by high-frequency fluctuations in the heat load power near zero, the system does not rely solely on the positive or negative polarity of the theoretical power for switching between heating and cooling modes. The controller internally incorporates dual-decision logic with hysteresis dead zone: only when... Only when the absolute value of the temperature deviation crosses the set anti-vibration threshold, and its polarity remains positively correlated with the polarity of the measured temperature deviation inside the cabin, can the corresponding refrigeration unit's four-way reversing valve or heating switching node array be activated. Furthermore, the conversion formula for the frequency conversion parameter of the integrated unit's energy efficiency ratio is set as follows: ; In the formula, This refers to the transient frequency compensation feedforward action or equivalent inverter drive parameters derived from the derivation of a high-power variable frequency compressor. This represents the absolute perturbation power envelope amplitude remaining after removing the positive and negative signs from the previous matrix operations; This represents the real-time cycle energy efficiency ratio characterization constant for the unit's energy transfer under the current environmental temperature and pressure load curve. Considering the nonlinear refrigeration characteristics of the reverse Carnot cycle, this dimensionless variable is obtained by bilinear interpolation through real-time consultation of the two-dimensional calibration table built into the air conditioning controller. The input addressing dimension of the calibration table is clearly divided into the current outdoor ambient temperature measurement value and the set reference temperature value of the cabin. This indicates the rated electrical power of the selected temperature control unit under full load. Indicates the maximum allowable operating threshold frequency for the frequency converter drive; To pre-determine the minimum constant protection value for preventing computational collapse, a threshold value of 10 is used in this embodiment. -5 The algorithm that integrates a minimal constant within the above formula aims to consider physical fault tolerance: when the outdoor unit of the air conditioner temporarily enters deep defrosting, causing the real-time calculated energy efficiency ratio to approach zero, this additional constant can sustainably resist the non-numerical singularity overflow caused by the denominator drop, avoiding abnormal interruptions due to computing power overflow in the microprocessor core. This dimensionality reduction conversion stage rigorously achieves the constrained convergence from a purely thermodynamic scalar to an execution-side driving capability that includes hardware performance barrier characteristics.

[0067] The S540 features instruction algebra coordination and safety takeover. It performs a full summation of the feedforward frequency command and the underlying PID closed-loop output, and performs an over-limit check before sending it to the pulse width modulation bus. If the synthesized frequency exceeds the physical limit, it suspends incremental adjustment and performs a saturation limiting overwrite to avoid overload risks.

[0068] See attached document Figure 8 , Figure 8 This is a flowchart illustrating the control weight stripping and steady-state regression logic according to an embodiment of the present invention. After the system establishes that the mechanical extension action of the modular shelter has ended and it has stopped at the physical parking dead point, the transient high-power feedforward intervention must be terminated to allow the control system to smoothly transition and return to the normal steady-state closed-loop control mode. Specific smooth exit mechanisms include: S610 extracts dynamic indicators of parking and docking to construct a dimensionality reduction exit window beacon for the system. To prevent erroneous switching of operating modes caused by the flexible deformation and vibration of mechanical components, a dual-track time amplitude merging judgment criterion is constructed to address the low-point drop in feedback reception speed. When the bottom layer's return rate drops to a minimum and breaks through the established threshold for vibration prevention, and can maintain a certain continuous non-jumping cycle time on the established time axis, the scheduler issues and locks this current physical natural stamp as the reference time root for subsequent smooth transition calculations.

[0069] The S620 utilizes a mathematical cosine periodic retreat formula algorithm with extremely smooth derivative connections to reduce the weight occupied by the registration quota. Considering that direct truncation of the command would cause a step change in the inverter control voltage signal, leading to severe consequences such as violent fluctuations in the pipeline flow field and transient oscillations in the fan motor, a semi-periodic function retreat trajectory without oblique angle sharp change connection is introduced.

[0070] The formula for calculating the control weight decay factor is as follows: ; In the formula, Represents the current natural time series The system assigns normalized control weights to the transient feedforward command channel, and their value range is strictly clamped between 0 and 1. This represents the physical origin timestamp generated by the beacon triggered by the steady-state regression established in the previous step. This indicates the duration of the smooth transition of control rights set by the system. This parameter is determined based on the thermal inertia and aerodynamic large inertia damping characteristics of the actuator, and is usually preferred to be a constant time constant of 10 to 15 seconds. To pre-determine a minimum constant to prevent computational anomalies, a threshold value of 10 is used in this embodiment. -6 .

[0071] The S630 employs an anti-integral saturation mechanism to achieve a smooth handover of control. To prevent the steady-state PID controller from experiencing deep integral saturation and runaway control during suspension, the system forcibly clamps abnormally accumulated integral deviation terms in the controller's memory to within the effective operating space, ensuring that the underlying control unit always matches the current control authority, thus achieving a smooth, uninterrupted handover of control.

[0072] S640, resource release and steady-state return. When the aforementioned transition time expires, control is completely transferred to the steady-state feedback loop. The execution layer cleans up the pre-push feature temporary storage, redundant data pointers, and fragmented cache according to the first-in-first-out principle to avoid stack deadlock and ensure that the main kernel is clean to handle the next control loop.

[0073] To aid in understanding the present invention, the following description is based on a typical deployment scenario of an emergency rescue medical dual-module extended modular unit.

[0074] Initial scene setup: External environmental conditions: The mobile shelter is deployed in a high-altitude and cold region, with an external temperature of... .

[0075] Internal baseline of the mobile medical unit: The steady-state internal temperature of the medical unit is required to be set to... Target micro-positive pressure (Prevents the intrusion of cold air containing germs or that has not been pre-filtered).

[0076] Hardware specifications: 1 built-in variable frequency compressor and matching electric heating components; 1 set of exhaust air sealing valves; 3 parallel distributed blowers (marked as F1, F2, F3).

[0077] Specific action deduction: In the static parking state, the effective displacement change rate is 0, and the PID controller maintains stable operation of the blower and exhaust valve. When the extension action is initiated, the system detects the bulkhead pushing outward at 0.05 m / s, and calculates a surge of 0.4 m / s. 3 Volume.

[0078] In this transition state, feedforward energy injection is triggered: for this -10°C 0.4m 3 The system calculates that approximately [amount missing] transient sensible heat power needs to be neutralized in the S50 for the cold fresh air intake. The 16.5kW, bottom-level electric heater and compressor are forced to increase their frequency setpoint by feedforward to actively dissipate heat. Simultaneously, in the momentum distribution phase, the fan speed must be increased to maintain a positive pressure of 15Pa. If fan F1 is already at 80% load due to its shorter duct, the impedance penalty matrix will dynamically give F1 a very small incremental command, smoothly distributing the responsibility of filling the negative pressure to F2 and F3. After the extension action stops at 30 seconds, the strong feedforward disturbance command smoothly yields to the PID steady-state controller within 15 seconds using a cosine decay curve.

[0079] To verify the effectiveness and advancement of this invention, a full-scale modular cabin environmental thermodynamic testing platform was constructed, and a real-vehicle dynamic extension test was conducted to compare it with a traditional pure PID closed-loop control scheme (comparative). The test conditions were as follows: the modular cabin deployed at a speed of 0.05 m / s, and the internal and external temperature difference was... .

[0080] The comparison and verification data are as follows: Capability to maintain micro-positive pressure in the compartment (fluid defense test): Conventional PID scheme: Starting from the 2nd second after the sudden change in the volume of the modular cabin, due to the integral response lag, the internal air pressure experiences an avalanche-like collapse, dropping to a minimum of The pressure was 12Pa negative. This state lasted for 18 seconds, during which the mobile cabin lost its biosafety-level micro-positive pressure defense, and a large amount of unfiltered external air flowed back in.

[0081] The present invention utilizes the transient mass compensation and Long Short-Term Memory (LSTM) network prediction of the S30 to activate the exhaust valve and blower in advance. The cabin pressure experienced only a slight drop throughout the entire deployment period (reaching a minimum of +11 Pa), remaining within the positive pressure range throughout, and the aerodynamic defense line was not breached, improving the resistance to volumetric disturbances by nearly 100%.

[0082] Temperature control robustness (thermal shock test): Conventional PID scheme: The cabin temperature drops sharply from 24°C to 18.5°C during deployment, and after the operation stops, it still experiences about 40 seconds of low-frequency thermodynamic oscillation before returning to steady state, resulting in extremely poor human comfort.

[0083] The solution of this invention, thanks to the S50 energy feedforward blocking control, allows the external energy estimation to enable the electric heating components and compressor to intervene in advance to offset sensible heat. The maximum temperature drop in the main temperature zone measured inside the cabin is only 0.6℃ (minimum 23.4℃), which basically achieves constant temperature expansion without being noticeable under severe mechanical distortion, demonstrating significant engineering progress.

[0084] Unit load stability test (dynamic distribution test): Conventional PID scheme: When the parallel fan units receive a unified input signal, due to the asymmetry of pipe resistance, the No. 1 front-stage fan instantly hits the bottom at 100% full frequency, triggering the inverter overcurrent alarm, and the system falls to the edge of dangerous overload.

[0085] The solution of this invention is as follows: By using the S40 pseudo-inverse impedance penalty matrix, the weight of fan 1 is dynamically limited when it reaches 85% bandwidth, and fans 2 and 3 actively take over the remaining make-up air volume. The entire unit operates smoothly, and the risk of thermal demagnetization and mechanical damage is reduced to zero.

[0086] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An environmental regulation and control system for an emergency rescue dual-cabin extended modular container, characterized in that, include: The data acquisition module is used to acquire parameters of the internal and external environment of the modular shelter and the kinematic parameters of the movable cabin. The actuator module includes ventilation and temperature control components located within the cabin; The control unit is communicatively connected to the data acquisition module and the actuator module, and the control unit is equipped with a steady-state closed-loop regulation mode and a transient feedforward control mode. Specifically, when the modular shelter is stationary, the control unit executes the steady-state closed-loop regulation mode; when the modular shelter is detected to be performing an expansion action and the kinematic parameters reach a set threshold, the control unit activates the transient feedforward control mode, analyzes the volume-related variables inside the modular shelter based on the kinematic parameters, and generates corresponding feedforward compensation commands to drive the actuator module; when the expansion action is detected to be over, the control unit executes dynamic weight stripping logic to gradually exit the transient feedforward control mode and return to the steady-state closed-loop regulation mode.

2. The environmental regulation and control system for an emergency rescue dual-cabin extended modular container according to claim 1, characterized in that, When the control unit executes the transient feedforward control mode, it is configured to execute transient quality compensation logic, specifically including: The volume-related variable is represented as the instantaneous volume change rate inside the cabin, which is the result of multiplying the real-time motion speed parameter by the preset effective cross-sectional area of ​​the active cabin. The system acquires the real-time motion velocity, real-time motion acceleration, measured temperature inside the cabin, pressure difference between inside and outside the cabin, and ambient temperature during the current sampling period. These parameters are then encapsulated into a single-frame state vector and subjected to dimensionality reduction and normalization. The normalized state vector is input into the long short-term memory neural network model for forward reasoning to obtain the target frequency feedforward of the blower and the target opening feedforward of the exhaust valve in the ventilation component. The calculated feedforward quantity is algebraically summed with the feedback output value of the steady-state closed-loop operation, and then sent to the ventilation component after the amplitude is limited.

3. The environmental regulation and control system for an emergency rescue dual-cabin extended modular container according to claim 2, characterized in that, The ventilation assembly includes multiple node fans configured in an array pattern, and the control unit is configured to execute momentum-based load balancing protection logic when issuing a target frequency feedforward. Based on the ambient air density and the cross-sectional area of ​​the system pipeline, the target frequency feedforward is converted into a global baseline for fluid demand allocation. Retrieve the node utility mapping array pre-constructed based on the physical flow resistance parameters of the pipeline branch for each corresponding node blower; It also reads back the current actual operating frequency of each node's blower in real time, and constructs a diagonal weighted penalty matrix by comparing it with its rated maximum boundary frequency; By combining the node utility mapping array and the diagonal weight penalty matrix, and applying the penalty pseudo-inverse matrix solution formula configured with damping constant, the frequency distribution sequence of each node's blower is output to maintain the total momentum requirement and minimize the total power consumption of the system.

4. The environmental regulation and control system for an emergency rescue dual-cabin extended modular container according to claim 1, characterized in that, When executing the transient feedforward control mode, the control unit is also configured to execute transient energy control logic: The external detection parameters of the ambient temperature are sampled and compared with the reference temperature parameters inside the cabin at the same time to establish a cross-domain meteorological temperature gradient difference benchmark. Based on the volume-related variables, the unit time fluid replacement rate, the preset ambient air absolute density parameter, and the dry air ratio isobaric heat capacity coefficient are derived and multiplied by the cross-domain meteorological temperature gradient difference benchmark to calculate the transient heat load absolute demand power to offset the interference factors of internal and external air mass replacement. The corresponding system cycle energy efficiency ratio parameter is obtained by addressing the two-dimensional interpolation table of the real-time operating condition of the air conditioning unit. The energy efficiency ratio parameter is then used to reduce the absolute power demand of the transient heat load to the equivalent drive execution quantity for the operation of the compressor and four-way valve in the temperature control component.

5. The environmental regulation and control system for an emergency rescue dual-cabin extended container according to claim 4, characterized in that, When calculating the equivalent drive execution quantity, the control unit is configured to execute the following safety decision constraint: The system has a preset hysteresis double-judgment dead zone. The system will only activate the authority to issue the equivalent drive execution quantity to the temperature control component when the amplitude of the absolute power demand of the transient heat load calculated exceeds the preset dead zone threshold and the polarity of the theoretical heat load direction is consistent with the current actual temperature deviation direction in the cabin. In the division calculation formula of the dimensionality reduction, a minimum constant value is added after the system cycle energy efficiency ratio parameter, which is used as the denominator. This minimum constant value replaces the efficiency state that approaches zero, thus avoiding the processor from failing to perform a division-by-zero operation.

6. The environmental regulation and control system for an emergency rescue dual-cabin extended modular container according to claim 2, characterized in that, The long short-term memory neural network model was pre-trained using the following method: The boundary sensor feedback data stream induced by the flow field oscillation inside the container under simulated mechanical traction deformation conditions is extracted as the input sample set; A mean squared error loss function with a quadratic penalty measure regularization term is applied to network weight optimization to constrain and smooth the absolute range of interconnection weights of each nonlinear neural node.

7. The environmental regulation and control system for an emergency rescue dual-cabin extended modular container according to claim 1, characterized in that, When executing the dynamic weight stripping logic, the control unit is configured to: The system analyzes dynamic docking indicators in real time. When it detects that the running speed in the kinematic parameters is lower than the static anti-vibration threshold of the system and the state is maintained for a set continuous period, the current timestamp is latched as the physical starting point timestamp reference. A cosine half-cycle decay function is established based on the physical starting point timestamp, and a monotonically decreasing real-time control weight factor is obtained by solving it. Within the set buffer time, the feedforward control weight factor is smoothly decayed to zero.

8. The environmental regulation and control system for an emergency rescue dual-cabin extended container according to claim 7, characterized in that, During the execution of the dynamic weight stripping logic, the control unit is also configured to execute an anti-integral saturation mechanism: Freeze the integral term of the error accumulated in the original steady-state closed-loop control mode within the control unit; Read the final driving quantity actually sent to the execution module, and call the transfer function of the corresponding execution module to reverse calculate the corresponding internal equivalent cumulative value; The frozen error integral term is overwritten using the internal equivalent cumulative value to achieve a seamless switch to the steady-state closed-loop control mode.

9. The environmental regulation and control system for an emergency rescue dual-cabin extended container according to claim 7, characterized in that, When the control factor drops to zero and the buffer duration ends, the control unit is configured to initiate a memory cleanup mechanism: Release memory resources occupied by the variance comparison array, feature buffer pool, and transition phase resident pointer stack, which are dedicated to transient feedforward control mode; Implement a memory overwrite initialization action based on the first-in-first-out rule, clean up redundant data, and restore the memory state in steady-state closed-loop regulation mode.

10. An environmental regulation and control method for an emergency rescue dual-module extended container, executed by an environmental regulation and control system for an emergency rescue dual-module extended container as described in any one of claims 1-9, characterized in that, The environmental regulation and control method includes the following steps: The parameters of the internal and external environment of the modular shelter and the kinematic parameters of the movable cabin are obtained cyclically. When it is determined that the modular cabin has not performed an expansion action and the kinematic parameters are lower than the set threshold, a steady-state closed-loop regulation mode is activated, and steady-state closed-loop feedback regulation of the cabin temperature and micro-pressure difference is performed based on the environmental parameters. When the expansion action of the modular cabin is detected and the kinematic parameters reach the set threshold, the steady-state closed-loop control mode is suspended and the transient feedforward control mode is switched on. In the transient feedforward control mode, the volume-related variables caused by the expansion of the internal volume of the modular cabin are analyzed based on the kinematic parameters, and feedforward compensation commands are generated. Transient feedforward adjustment is performed according to the feedforward compensation commands. When the extended action is detected to have reached the limit end and ended, the dynamic weight stripping logic is executed. The control authority is transferred by controlling the smooth decay of the weight and the internal integral quantity back calculation and overwrite mechanism, returning to the steady-state closed-loop adjustment mode, and clearing the transition period cache resources occupied by the transient feedforward control mode.

Images