Aerogel greenhouse heat preservation quilt multi-mode sensing fusion control system

Abstract

The invention relates to the technical field of agricultural intelligent equipment, and discloses a multi-mode sensing fusion control system for an aerogel greenhouse heat preservation quilt. An environment multi-mode monitoring module of the system collects internal temperature and humidity distribution, illumination intensity gradient, soil moisture content profile and external meteorological change data of a greenhouse in real time to generate an environment comprehensive monitoring data set; based on the data set, the thermal insulation mode decision module analyzes the difference between the thermal energy loss rate and the thermal insulation demand of different areas in combination with the physical characteristic parameters and historical regulation and control records of the aerogel thermal insulation quilt, matches a dynamic thermal insulation strategy and generates a multi-stage partition regulation and control instruction; the cooperative parameter optimization module calculates a combination scheme of the unfolding angle, the coverage area and the thickness of the heat preservation quilt, and adjusts a partition cooperative control parameter set; the multi-mode execution controller generates a heat preservation quilt dynamic regulation and control map; the anomaly fusion diagnosis module monitors anomalies and generates a set of regulation error correction factors. The system realizes fine, dynamic and intelligent regulation and control of greenhouse heat preservation.

Description

Technical Field

[0001] This invention relates to the field of agricultural intelligent equipment technology, specifically to a multimodal sensing fusion control system for aerogel greenhouse insulation blankets. Background Technology

[0002] In the field of agricultural facility cultivation, greenhouses serve as an important carrier for ensuring a stable growing environment for crops, and their heat preservation performance directly affects crop yield and quality. Traditional greenhouse insulation mostly relies on single-layer or multi-layer composite insulation materials, achieving heat control through manual laying or mechanical rolling, but this method has many limitations.

[0003] From an environmental adaptability perspective, traditional insulation methods struggle to cope with complex and ever-changing weather conditions. For instance, when there are significant day-night temperature differences, the timing of covering and removing insulation materials relies entirely on experience, which can easily lead to excessive temperature fluctuations inside the greenhouse, affecting crop growth cycles. Furthermore, existing systems cannot accurately respond to dynamic changes in light intensity; both excessively strong and weak light can reduce crop photosynthetic efficiency.

[0004] In terms of spatial control precision, traditional greenhouse insulation often adopts a holistic control approach, neglecting the environmental differences between different areas within the greenhouse. For example, there are significant differences in temperature, humidity, and soil moisture between the greenhouse edge and the central area. A uniform insulation strategy can lead to heat waste in some areas and insufficient insulation in others. Furthermore, soil moisture monitoring is mostly limited to surface data, making it difficult to reflect the moisture status of deeper soil layers, resulting in a lack of scientific basis for the coordinated control of irrigation and insulation.

[0005] From a materials application perspective, aerogel, as a novel high-efficiency thermal insulation material, possesses extremely low thermal conductivity and excellent thermal insulation performance. However, existing technologies have failed to fully utilize its physical properties. Traditional control systems cannot dynamically adjust based on parameters such as the deformation characteristics and thickness variations of aerogel, resulting in low material performance utilization and susceptibility to material loss and shortened service life due to improper control.

[0006] Existing systems have weak fault diagnosis capabilities. When deviations occur in the insulation actuators or environmental parameters change abruptly, there is a lack of effective compensation mechanisms and fault-tolerance strategies, which can easily lead to equipment failure and affect the continuous and stable operation of the greenhouse. These problems collectively restrict the level of intelligence in greenhouse insulation control and make it difficult to meet the development needs of modern agriculture for precision and efficiency. Summary of the Invention

[0007] The purpose of this invention is to provide a multimodal sensing fusion control system for aerogel greenhouse insulation blankets to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides a multimodal sensing fusion control system for aerogel greenhouse insulation blankets, the system comprising: The environmental multimodal monitoring module is used to collect real-time environmental parameters inside the greenhouse, including temperature and humidity distribution data, light intensity gradient data, soil moisture profile data, and external meteorological change data, and generate a comprehensive environmental monitoring dataset. The thermal insulation mode decision module, based on the comprehensive environmental monitoring dataset, combined with the physical property parameters of the aerogel insulation blanket and historical control records, analyzes the differences in heat loss rate and thermal insulation requirements in different areas, matches dynamic thermal insulation strategies, and generates multi-level zone control instructions. The collaborative parameter optimization module calculates the combination scheme of the aerogel insulation blanket unfolding angle, coverage area and insulation layer thickness according to the multi-level partition control instructions, calibrates the action timing conflict and energy consumption threshold constraint between each execution unit, and adjusts the partition collaborative control parameter set. The multimodal execution controller, based on the partitioned collaborative parameter set, statistically analyzes the deformation feedback data of the aerogel insulation blanket and the target position deviation, pairs the material deformation characteristics with the drive mechanism response curve, calculates the execution compensation amount and path replanning range, reassembles the drive signal sequence and the execution unit collaborative strategy, and generates a dynamic control map of the insulation blanket. The abnormal fusion diagnosis module, based on the dynamic control map of the insulation blanket, monitors material stress anomalies and environmental mutation interference during the execution process, adjusts the driving parameter compensation mechanism and collaborative timing fault tolerance strategy, and generates a set of control error correction factors.

[0009] Preferably, the steps for generating the comprehensive environmental monitoring dataset are as follows: To meet the needs of monitoring the internal environment of greenhouses, a raw data acquisition model for multimodal sensors was developed. A spatiotemporal alignment algorithm was used to perform timestamp calibration and spatial grid mapping on temperature and humidity distribution data, light intensity gradient data, and soil moisture profile data; Identify the associated feature vectors of each sensor data in the spatial grid, including the features of heat conduction hysteresis effect, the nonlinear features of light attenuation, and the associated features of soil moisture migration, and form a multimodal feature fusion matrix; The environmental parameters in the multimodal feature fusion matrix are coupled and analyzed separately. The thermodynamic coupling region, the photothermal conversion efficiency region, and the soil moisture dynamic influence region are extracted in sequence as the basis for partition mapping of the comprehensive environmental monitoring dataset.

[0010] Preferably, the step of forming the multimodal feature fusion matrix further includes: The characteristics of heat conduction hysteresis effect and the nonlinear characteristics of light attenuation are superimposed on the time series to obtain the multiphysics coupling results; Temperature gradient extrema are extracted from the multiphysics coupling results. These extrema are then compared with a phase transition threshold library for aerogel materials to generate a thermodynamic stability identifier set. The cross-influence factors of diffusion rate and heat conduction rate in soil moisture migration correlation characteristics are extracted from the multiphysics coupling results. The multiphysics coupling results are reconstructed according to the diffusion rate threshold and heat conduction fluctuation rate to generate a soil moisture-thermodynamic dynamic equilibrium parameter table.

[0011] Preferably, the generation step of the multi-level partition control instruction specifically includes: The thermodynamic stability identifier set and soil moisture-thermodynamic dynamic equilibrium parameter table in the comprehensive environmental monitoring dataset are called up, and the heat preservation priority of different areas is analyzed through energy consumption optimization algorithm. Identify the deployment mechanism state parameters of each section of the aerogel insulation blanket, including the motor torque response curve and the material tensile strength attenuation coefficient; Based on the state parameters of the deployment mechanism and the heat preservation priority, a target control model is constructed: Successful control samples from historical control records are collected, and the optimal value of driving energy consumption and the threshold of deformation control accuracy are extracted as training labels. The thermodynamic stability identifier set is input into a temporal prediction network with variable kernel width to output an initial control scheme for the action sequence of the driving mechanism. The timing conflict nodes of the initial control scheme are corrected by the moisture-thermodynamic dynamic equilibrium parameter table, and multi-level zone control instructions are generated.

[0012] Preferably, the adjustment steps for the partitioned collaborative control parameter set are as follows: Extract the target position sequence and action trigger time window of each execution unit from the multi-level partition control instructions; Analyze the mechanical interference area and peak power load period between adjacent execution units, and mark the action timing conflict domain and instantaneous energy consumption over-limit risk points; The spatiotemporal obstacle avoidance algorithm is used to recalculate the action delay compensation of each execution unit, and combined with the deformation relaxation characteristics of aerogel material, a partitioned collaborative fault-tolerant time window parameter table is generated. By integrating the mechanical interference region marking results with the fault-tolerant time window parameter table, the phase difference of the driving signal and the power compensation coefficient are calibrated to generate a partitioned collaborative control parameter set.

[0013] Preferably, the step of generating the dynamic control map of the thermal insulation blanket specifically includes: The partitioned collaborative control parameter set is input into the deformation-drive coupling model to predict the real-time deformation trajectory of the aerogel insulation blanket. Collect the actual deformation trajectory point set fed back by high-precision displacement sensor, and calculate the point-by-point position deviation between the predicted trajectory and the actual trajectory; Identify abnormal deformation regions where the positional deviation exceeds the material's elastic threshold, and extract the current fluctuation characteristics of the drive mechanism and the records of sudden changes in environmental wind speed in the corresponding regions. Based on the current fluctuation characteristics and wind speed change records, the amplitude-frequency adjustment curve of the driving signal is reconstructed to generate a coverage path replanning vector. By integrating the position deviation analysis results with the path replanning vector, the driving pulse width modulation sequence is recombined, and the dynamic control spectrum of the insulation blanket is output.

[0014] Preferably, the step of generating the set of adjustment error correction factors specifically includes: Historical regulation records and current environmental mutation data of abnormal deformation regions were extracted from the dynamic regulation map of thermal insulation blankets. A coupled analysis model of material stress and environmental disturbance is constructed to calculate the nonlinear relationship between deformation compensation and actual energy consumption increment; Identify the periods of continuous disturbance and instantaneous impact events in environmental wind speed change records, and calibrate the passive fault tolerance time threshold by combining the creep recovery time constant of aerogel materials. Based on the aforementioned nonlinear relationship and passive fault tolerance time threshold, the overload protection coefficient and torque attenuation compensation gradient of the drive mechanism are dynamically adjusted. A set of regulation error correction factors is generated by using a multi-execution unit synergy evaluation matrix.

[0015] Preferably, the calibration step of the passive fault tolerance time threshold further includes: Extract real-time operating data of adjacent execution units, including motor temperature rise curves and gearbox vibration spectrum; Analyze the heat dissipation efficiency decay node in the motor temperature rise curve and the abnormal resonance frequency band of the gearbox vibration spectrum; By mapping the heat dissipation efficiency decay node and the abnormal resonant frequency band to the time constraint domain, the failure early warning time window of the mechanical system is calculated. The minimum value between the mechanical system failure warning time window and the duration of the environmental wind speed change record is taken as the passive fault tolerance time threshold.

[0016] Preferably, the system further includes: The global energy efficiency assessment module is used to integrate the real-time data stream of the environmental multimodal monitoring module with the error correction factor set of the anomaly fusion diagnosis module to construct a dynamic assessment model of thermal insulation performance. The predictive maintenance module, based on the output of the thermal insulation performance dynamic evaluation model, correlates with the historical fault feature library of the drive mechanism to predict the life decay curve of the execution unit and the material fatigue critical point. The fault-tolerant strategy iteration module dynamically updates the safety margin coefficient in the partitioned collaborative control parameter set and the compensation weight value in the control error correction factor set based on the life decay curve and the material fatigue critical point.

[0017] Preferably, the system further includes a dynamic interference quantization module, which is implemented in the following ways: The system acquires environmental wind speed abrupt change records and material stress anomaly data from the anomaly fusion diagnosis module, and extracts the spatial gradient distribution characteristics of wind speed abrupt changes and the duration sequence of stress anomalies. A wind speed-stress coupled response model was constructed, and the nonlinear correlation coefficient between stress fluctuation and wind speed change per unit time was calculated. The system identifies periods of continuous interference where the nonlinear correlation coefficient exceeds the material creep threshold, and generates dynamic interference intensity grading parameters by combining the deformation recovery hysteresis characteristics of aerogel insulation blankets. Based on the dynamic interference intensity classification parameters, the dynamic adjustment step size of the overload protection coefficient of the drive mechanism is calibrated, and the fault-tolerant response priority in the control error correction factor set is updated.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This multimodal sensing fusion control system for aerogel greenhouse insulation blanket uses an environmental multimodal monitoring module to collect real-time data on various dimensions, including temperature and humidity distribution inside the greenhouse, light intensity gradient, soil moisture profile, and external meteorological changes. The resulting comprehensive environmental monitoring dataset fully reflects the environmental conditions inside and outside the greenhouse, providing rich and accurate basic information for subsequent control measures. This multi-parameter collaborative monitoring approach overcomes the limitations of traditional single-parameter monitoring, allowing the system to perceive the environment more comprehensively and meticulously.

[0019] The insulation mode decision module combines the physical properties of the aerogel insulation blanket with historical control records to analyze the differences in heat loss rates and insulation requirements in different areas. This allows for the matching of dynamic insulation strategies and the generation of multi-level zone control commands. This transforms insulation control from a simple overall operation into a differentiated approach tailored to the actual needs of different areas within the greenhouse. It fully considers the impact of spatial heterogeneity on insulation performance, ensuring that insulation measures for each area are more aligned with actual requirements.

[0020] The collaborative parameter optimization module calculates the combination of the aerogel insulation blanket's unfolding angle, coverage area, and insulation layer thickness based on multi-level zone control instructions. It also calibrates the timing conflicts and energy consumption threshold constraints between each execution unit and adjusts the zone-based collaborative control parameter set. This process optimizes the combination of various physical parameters of the insulation blanket while also considering the coordination between execution units. It avoids inefficiencies or equipment damage caused by action conflicts, and effectively controls energy consumption while meeting insulation requirements.

[0021] The multimodal actuator controller, based on a partitioned collaborative parameter set, statistically analyzes the deformation feedback data of the aerogel insulation blanket and the deviation from the target position, pairs the material deformation characteristics with the response curves of the drive mechanism, calculates the execution compensation amount and path replanning range, reassembles the drive signal sequence and the collaborative strategy of the execution unit, and generates a dynamic control map of the insulation blanket. This step fully utilizes the characteristics of aerogel materials, ensuring that the insulation blanket can act according to the expected trajectory and state through real-time processing of deformation feedback and precise control of the drive mechanism, thus improving the accuracy and stability of the control.

[0022] The anomaly fusion diagnostic module, based on the dynamic control map of the insulation blanket, monitors material stress anomalies and environmental abrupt changes during execution. It adjusts the driving parameter compensation mechanism and collaborative timing fault-tolerant strategy to generate a set of control error correction factors. This enables the system to react promptly to unexpected situations, compensating for errors by adjusting relevant parameters and strategies, reducing the impact of anomalies on the insulation effect, and ensuring the long-term reliability of the system. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the working principle of the multimodal sensor fusion control system for aerogel greenhouse insulation blankets described in this invention. Figure 2 A flowchart generated for an integrated environmental monitoring dataset; Figure 3 A flowchart for generating multi-level zone control commands; Figure 4 This is a flowchart for generating dynamic control patterns for thermal insulation blankets. Detailed Implementation

[0024] The technical solutions of 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.

[0025] Please see Figure 1 This invention provides a multimodal sensing fusion control system for aerogel greenhouse insulation blankets, the system comprising: The closed-loop control system comprises an environmental multimodal monitoring module, a thermal insulation mode decision-making module, a collaborative parameter optimization module, a multimodal execution controller, and an anomaly fusion diagnosis module. The environmental multimodal monitoring module collects multi-dimensional environmental parameters such as temperature, humidity, light intensity, and soil moisture through a distributed sensor array, and establishes a unified comprehensive environmental monitoring dataset using a spatiotemporal alignment algorithm. The thermal insulation mode decision-making module generates multi-level zoned control commands based on a thermodynamic stability identifier set and a soil moisture-thermodynamic dynamic equilibrium parameter table, using a convolutional neural network. The collaborative parameter optimization module analyzes the mechanical interference and electrical load constraints between execution units and outputs a zoned collaborative control parameter set. The multimodal execution controller combines material deformation characteristics and drive response curves to generate a real-time dynamic control spectrum of the thermal insulation layer.

[0026] Example 1: See Figure 2 The environmental comprehensive monitoring dataset generation process employs a multi-source heterogeneous sensor collaborative acquisition architecture. A three-dimensional monitoring network is established within the greenhouse space. Temperature and humidity distribution data acquisition utilizes a digital sensor array, with each measurement node integrating temperature and relative humidity sensors. These nodes are evenly distributed within the greenhouse space at 0.5-meter intervals, forming a spatial gridded monitoring system. Light intensity monitoring uses a planar array composed of silicon-based photoelectric sensors. Optical diffusers are installed on the sensor surfaces to eliminate angle dependence. The array spacing is dynamically adjusted according to the greenhouse span and height to ensure accurate capture of light gradient changes. Soil moisture monitoring employs probe groups based on the time-domain reflectometry principle, buried in layers at different depths to form a vertical soil moisture profile data acquisition structure.

[0027] Data synchronization of the multimodal sensors employs a precision clock protocol to achieve a unified time base. Each sensor node incorporates a high-precision clock chip, and time synchronization calibration is performed via a wired network to eliminate data deviations caused by asynchronous sampling times. The spatial registration process establishes a three-dimensional coordinate system with the geometric center of the greenhouse as the origin, mapping measurement data from different sensors to a unified spatial reference system. Temperature and humidity point clouds are correlated with illumination point clouds through a spatial indexing structure, and a nearest neighbor search algorithm is used to match spatially close measurement points. Soil moisture data is processed continuously in the vertical direction using an interpolation algorithm to construct a complete three-dimensional monitoring data field.

[0028] The heat conduction feature extraction process analyzes the dynamic relationship between adjacent temperature and humidity measuring points, establishing a heat flow transfer model to describe the evolution of heat distribution inside the greenhouse. Feature parameters include thermal response delay time and temperature gradient change rate, reflecting differences in thermal inertia across different areas. Light intensity feature processing considers the propagation characteristics of light within the greenhouse, analyzing the ratio of direct to diffuse light to establish a feature model of light intensity variations over time and space. Soil moisture migration features are analyzed by examining the correlation of soil moisture data at different depths to identify the movement trends and rates of moisture within the soil profile.

[0029] Multiphysics coupling analysis spatiotemporally superimposes thermal conduction and illumination characteristics to identify the interaction between the temperature and radiation fields. The coupling analysis process establishes a correlation matrix between characteristic parameters, quantifying the degree of mutual influence between different physical fields. A temperature gradient extremum detection algorithm scans the entire monitoring area, marking key locations with drastic temperature changes; these locations typically correspond to regions of thermal instability. The extremum features are then matched against a pre-established material property database to determine whether the current temperature conditions exceed the stable operating range of the aerogel material.

[0030] The dynamic equilibrium analysis of soil moisture and thermodynamic parameters employs multivariate statistical methods to calculate the correlation indices between soil moisture movement and temperature changes. The analysis process establishes a model relating the rate of soil moisture change to the amplitude of temperature fluctuations, identifying the hysteresis response characteristics between the two. The generation of the dynamic equilibrium parameter table includes the normalization of characteristic parameters and time series alignment to ensure the comparability of data for different physical quantities. The parameter table is organized according to spatial grid locations, with each grid cell storing the corresponding equilibrium state index.

[0031] Thermodynamic stability assessment, based on coupled analysis results, classifies and labels the environmental stability of each zone within the greenhouse. The assessment process considers the combined effects of multiple factors, including temperature fluctuation amplitude, heat transfer rate, and light intensity variations. A multi-level classification system is used for stability labeling to distinguish areas with different risk levels, providing a basis for subsequent control decisions. The label set is updated frequently in sync with the rate of environmental change to ensure the timeliness of the assessment results.

[0032] The storage structure of the environmental integrated monitoring dataset adopts a spatiotemporal cube model, unifying the time, spatial, and physical quantity dimensions. The dataset comprises three layers: raw measurement data, feature extraction data, and fusion analysis data, supporting data access needs at different granularities. The data update mechanism employs incremental processing, recalculating only data areas that have changed, thus improving system efficiency. The data access interface provides multiple query methods, supporting searches based on combinations of time range, spatial region, and physical quantity type.

[0033] Sensor data quality control includes multiple verification steps, monitoring the operating status and parameters of each sensor in real time. Anomaly detection algorithms analyze the reasonable range and temporal continuity of measured values, filtering out outliers that significantly deviate from the normal range. Sensor fault diagnosis identifies potential hardware problems by analyzing signal characteristics and power supply parameters. A data completion mechanism ensures the integrity of the dataset by making reasonable estimations based on measurements from neighboring nodes and historical data models when some sensors fail.

[0034] The spatiotemporal feature extraction process employs a sliding time window analysis technique, with the window size adaptively adjusted based on environmental changes. Shorter time windows are used to capture transient features during rapid changes, while longer windows are used during stable phases to improve the stability of feature extraction. Feature parameter calculations consider spatial neighborhood relationships, introducing spatial weighting factors to reflect the degree of influence between different locations. Dimensionality compression techniques for the feature matrix preserve key feature components, reducing the computational complexity of subsequent processing.

[0035] Multimodal data fusion employs a hierarchical and progressive processing flow. First, it performs spatiotemporal alignment and feature extraction within the same physical quantity, then establishes a correlation model between different physical quantities. The fusion process preserves the original characteristics of each physical quantity while uncovering implicit cross-modal relationships. The fusion results can be represented in both visual and machine-analyzed formats, facilitating both human supervision and automated processing. The data update trigger mechanism sets multi-level thresholds, determining the processing frequency at different levels based on the degree of environmental change.

[0036] Example 2: See Figure 3 The multi-level zoning control command generation process is based on a comprehensive environmental monitoring dataset, achieving precise control through a hierarchical decision-making mechanism. The system first calls the preprocessed thermodynamic stability identifier set and soil moisture-thermodynamic dynamic equilibrium parameter table. These data have undergone spatiotemporal alignment and feature fusion processing to form a structured environmental state description. The energy consumption optimization algorithm employs a multi-objective programming framework, simultaneously considering three dimensions: temperature maintenance accuracy, energy consumption efficiency, and equipment operating life. The equilibrium point is selected through a Pareto optimal solution set. The heat preservation priority calculation incorporates fuzzy logic reasoning, quantifying qualitative factors such as temperature deviation, crop accumulated temperature requirements, and regional importance into comparable numerical indicators.

[0037] The unfolding mechanism condition monitoring system consists of a distributed sensor network that collects motor operating parameters and material mechanical property data in real time. Torque sensors are installed at key locations on the drive shaft, and the measured signals are processed with anti-interference filtering to extract characteristic waveforms reflecting load characteristics. Material performance degradation analysis, combined with real-time environmental data and an accelerated aging model, predicts the trend of mechanical property changes in the aerogel insulation blanket under different temperature and humidity conditions. The condition parameter update mechanism adopts an event-triggered approach, immediately initiating a reassessment process when a significant change is detected.

[0038] The training process of the target regulation model employs a deep reinforcement learning framework, and historical regulation samples undergo rigorous data cleaning and feature engineering. Training label generation combines expert knowledge and automatic optimization algorithms. The optimal energy consumption value is determined by minimizing the power integral curve, and the deformation control accuracy threshold is set based on a comprehensive consideration of material properties and crop requirements. The network architecture design takes into account the need for temporal feature extraction, employing a long short-term memory network structure with an attention mechanism, capable of capturing long-term dependencies in environmental parameter changes.

[0039] The input data of the time series prediction network undergoes standardization preprocessing to eliminate dimensional differences between different physical quantities. Network parameter initialization employs an adaptive method, adjusting the initial weight distribution based on the statistical characteristics of the input data. The training process uses a phased strategy: first, feature extraction training is performed with some layer parameters fixed, followed by end-to-end fine-tuning. The initial control scheme output by the network includes the action sequences and time parameters of each execution unit, forming a preliminary control instruction set.

[0040] The initial control scheme revision process incorporates a multi-constraint optimization algorithm, with soil moisture-thermodynamic balance parameters integrated as soft constraints into the objective function. Timing conflict detection identifies overlapping areas and resource contention points between execution units by establishing a resource occupancy timeline model. The revision algorithm adjusts the timing and amplitude of local actions while maintaining the overall control objective, eliminating potential mechanical interference and power load conflicts. A dynamic buffer time allocation strategy automatically adjusts the action interval based on regional environmental sensitivity, increasing control margin in critical areas.

[0041] The encoding format of multi-level partition control instructions adopts a hierarchical structure, including two levels: global control parameters and local adjustment instructions. Global parameters define the overall control strategy and coordination rules, while local instructions specify the detailed action parameters for each execution unit. The instruction set includes validity verification information, including data integrity checks and logical consistency checks, to ensure the reliability of the control instructions. The instruction transmission protocol is designed with real-time requirements in mind, employing a priority queue mechanism to ensure the timely delivery of critical instructions.

[0042] The execution unit status feedback mechanism forms a closed-loop control circuit, monitoring command execution and actual control effects in real time. Feedback data includes position sensor readings, motor current waveforms, and material deformation measurements, which are compared and analyzed against the expected control targets. A deviation detection algorithm identifies anomalies exceeding allowable limits, triggering command adjustment procedures. An adaptive learning module records the differences between actual operation and expected results, progressively optimizing the control parameter mapping relationship.

[0043] The dynamic adjustment process of control commands considers the response characteristics of the actuators, establishing a composite model that includes mechanical delay and electrical inertia. The adjustment algorithm predicts the actual effect of the control actions, compensating in advance for deviations caused by inherent system delays. The emergency stop mechanism operates independently of the main control loop, immediately cutting off power output when an abnormal situation that may endanger equipment safety is detected. Safety protection strategies are implemented in a hierarchical manner, with different protective measures taken according to the risk level.

[0044] The multi-level zone control system supports switching between multiple control modes, including fully automatic operation, semi-automatic assisted operation, and manual intervention. Mode switching is seamless, maintaining the continuity and stability of the control process. The operation log records a complete control history, including the time-series correspondence between environmental states, control commands, and execution results, providing data support for system optimization and maintenance. Log analysis tools extract typical control modes and empirical parameters, continuously enriching the system's knowledge base.

[0045] The system performance monitoring module evaluates the control effect in real time, calculating key indicators such as temperature uniformity, energy efficiency, and equipment load. The evaluation results are visualized, helping operators intuitively understand the system's operating status. An anomaly warning mechanism sets multi-level thresholds, issuing different levels of warning signals based on the degree of deviation. The maintenance reminder function predicts the remaining service life of critical components based on equipment operating time and workload, allowing for the rational scheduling of maintenance plans.

[0046] The adaptive adjustment process of control parameters is continuous, dynamically optimizing the control strategy based on environmental changes and equipment status. The adjustment algorithm balances short-term control performance with long-term equipment wear and tear, avoiding excessive consumption of equipment lifespan in pursuit of immediate performance. The parameter update mechanism employs a gradual adjustment approach to prevent system instability caused by drastic changes in control strategy. Version control records the parameter change history, supporting quick revert to previous stable versions.

[0047] Example 3: See Figure 4 The dynamic control pattern generation process of the thermal insulation blanket is based on a partitioned collaborative control parameter set, and precise control is achieved through a deformation-drive coupling model. The deformation prediction model adopts an improved Cosserat theoretical framework, considering the nonlinear deformation characteristics of aerogel materials, and its constitutive relation can be expressed as:

[0048] in Represents total strain energy. For material volume domain, It is the Green's strain tensor. It is a fourth-order elastic tensor. This represents the strain rate-dependent dissipation potential function. This model can accurately describe the mechanical behavior of aerogel materials under large deformation conditions.

[0049] Real-time deformation trajectory prediction employs a variable-step numerical integration algorithm, automatically adjusting the calculation step size based on the material stress state. The prediction process considers the influence of ambient temperature on material stiffness, obtaining elastic parameters at the current temperature through interpolation. The trajectory point generation frequency is synchronized with the control cycle, outputting a set of predicted position coordinates for each control cycle. A trajectory smoothing algorithm eliminates high-frequency oscillations introduced by numerical calculations, ensuring the stability of control commands.

[0050] The actual deformation measurement system consists of a laser displacement sensor array, with the sensor density determined based on control accuracy requirements. Measurement data undergoes median filtering to remove random noise, and then a Kalman filter algorithm is used to fuse multi-sensor data, improving the accuracy of position estimation. A data synchronization mechanism ensures strict alignment between measurement time and control cycle, eliminating errors caused by time delays. An outlier detection algorithm identifies measurement points outside the physically possible range, and nearest-neighbor interpolation is used for data repair.

[0051] Position deviation analysis employs Mahalanobis distance metric, considering the varying control precision requirements across different spatial directions. The deviation threshold is dynamically adjusted based on material mechanical properties, automatically tightening the control tolerance as it approaches the elastic limit. An abnormal deformation region detection algorithm scans the entire insulation blanket surface, marking control units where deviations exceed allowable limits. A region growing method clusters discrete anomaly points into continuous anomaly regions, providing a complete target range for subsequent adjustments.

[0052] Current fluctuation characteristic analysis was achieved through online spectrum monitoring, acquiring three-phase current signals from the drive motor and performing Fast Fourier Transform. The feature extraction process focused on frequency band energy changes related to mechanical resonance, establishing a correlation model between current harmonics and external interference. Wind speed abrupt changes were recorded from an ultrasonic anemometer; data preprocessing included outlier removal and moving average smoothing. Spatiotemporal correlation analysis determined the causal relationship between wind speed changes and mechanical vibration, distinguishing between direct interference and indirect effects.

[0053] The drive signal reconstruction algorithm, based on impedance control principles, treats environmental interference as an external force. The amplitude-frequency adjustment curve generation considers the frequency response characteristics of the actuator, maximizing anti-interference capability while maintaining system stability. The phase compensation module eliminates inherent delays in the control loop, improving the system's response speed. The gain scheduling strategy adjusts control parameters according to interference intensity levels, achieving stable control over a wide operating range.

[0054] The path replanning employs a local correction method based on spline interpolation to optimize the motion path in abnormal regions while preserving the global trajectory shape. Control point density adaptively increases according to the severity of deformation, providing finer control in critical areas. Smoothing of the transition zone ensures the corrected trajectory has continuous curvature changes, avoiding impact on the actuators. The replanning efficiency optimization algorithm only recalculates in the affected areas, keeping the overall control timing unaffected.

[0055] The pulse width modulation (PWM) sequence optimization takes into account the switching characteristics of the power devices and eliminates control dead time through dead time compensation. The duty cycle analysis algorithm converts the target torque into corresponding PWM parameters based on the torque-current characteristic curve. Timing constraint checks are introduced during the sequence reassembly process to ensure that the switching frequency of the power module does not exceed the rated value. A safety protection mechanism monitors the temperature of the power devices in real time and automatically reduces the frequency when there is an overheating risk.

[0056] The encoding format of the dynamic control map adopts a hierarchical data structure, including global control parameters, regional adjustment instructions, and security verification information. A data compression algorithm removes redundant control points, improving transmission efficiency. The refresh mechanism employs a differential update strategy, transmitting only the changed control parameters. Version control records the map's modification history, supporting the retrospective analysis and comparison of control strategies.

[0057] The system validation process comprises two stages: static verification and dynamic simulation. Static verification checks the rationality and completeness of the control parameters, including value range checks and logical relationship verification. Dynamic simulation predicts control effects by simplifying the model and identifying potential conflicts in advance. Only validated control commands are issued to the actuators to ensure the safe operation of the system.

[0058] The execution status monitoring system provides real-time feedback on the control effect, forming a closed-loop control circuit. Data acquisition frequency exceeds the control cycle, and feature values ​​are extracted statistically using a sliding window. Performance evaluation indicators calculate the matching degree between the actual deformation and the target trajectory, quantifying control accuracy. An adaptive learning module records the control effect under different operating conditions, progressively optimizing the parameters of the deformation prediction model.

[0059] The anomaly handling mechanism employs a tiered response strategy, taking different measures based on the severity of the anomaly. Minor deviations are compensated for through parameter fine-tuning, moderate deviations trigger local reprogramming, and severe deviations activate safety protection procedures. A fault diagnosis expert system analyzes the causes of anomalies and provides handling suggestions. The knowledge base continuously accumulates typical cases to improve the system's self-learning capabilities.

[0060] The maintenance reminder function predicts the remaining lifespan of critical components based on equipment operating status. The wear model considers both mechanical load and environmental influences to calculate the equivalent service life. Warning thresholds are set according to component importance, allowing for proactive preventative maintenance. The maintenance record database stores complete equipment health records, supporting continuous optimization of the lifespan prediction model.

[0061] The human-computer interface intuitively displays the execution effect of the dynamic control spectrum, including 3D deformation simulation and key parameter trend charts. Operation permissions are managed hierarchically, with different levels of users having corresponding control privileges. Operation logs fully record every step of the control process, meeting audit trail requirements. The system provides context-sensitive guidance information to reduce the difficulty of use.

[0062] Example 4: The passive fault tolerance time threshold calibration process achieves accurate determination through multi-source data fusion. Taking the actual operation scenario of a winter tomato planting greenhouse as an example, the system detected abnormal vibration in the B3 area of ​​the drive unit and immediately initiated the fault tolerance threshold calculation process. The temperature sensor installed on the drive motor recorded that the winding temperature started to rise continuously from 45°C, while the gearbox accelerometer detected a significant increase in vibration energy in the 3.2kHz frequency band. The environmental monitoring module simultaneously reported that the area was experiencing a level 6 gust of wind, with wind speed fluctuations ranging from 8 to 12 m / s.

[0063] The real-time operating data acquisition system records the operating parameters of the motor and transmission mechanism at a sampling frequency of 500Hz. Temperature monitoring points are arranged at three key locations: the motor windings, bearing housings, and heat sinks, forming a temperature field distribution monitoring network. Vibration sensors employ triaxial accelerometers, installed at three locations: the motor housing, the gearbox input shaft, and the output shaft. The data preprocessing stage eliminates power grid frequency interference and mechanical background noise, extracting effective characteristic signals. Temperature trend analysis uses a moving average method to identify sustained temperature rise segments, and vibration spectrum analysis uses a peak detection algorithm to locate abnormal resonant frequency bands.

[0064] The calculation of the mechanical system failure early warning time window is based on the theory of cumulative damage from material fatigue. Analysis of the motor temperature rise curve identifies the inflection point where heat dissipation efficiency decreases, after which the temperature rise rate accelerates significantly. Gearbox vibration characteristic analysis shows that the energy proportion of abnormal frequency bands exceeds the baseline level by more than 30%. The remaining safe operating time is estimated by correlating the temperature rise rate, vibration energy growth trend, and material performance parameters. Environmental wind speed data is incorporated into the calculation process to correct the influence coefficient of wind load on the lifespan of mechanical components.

[0065] The global energy efficiency assessment module integrates current operating data and historical records to construct a dynamic assessment model. The assessment indicators include three dimensions: temperature control accuracy, energy efficiency ratio, and equipment health, with several secondary indicators under each dimension. The data acquisition cycle is dynamically adjusted according to the rate of environmental change, automatically shortening to 5 seconds during sudden wind speed changes. Assessment results are expressed as quantitative scores and also generate intuitive trend graphs. System maintenance personnel can quickly grasp the overall operating status through these assessment results.

[0066] The predictive maintenance module is linked to a fault feature database, which contains vibration spectrum templates and temperature variation characteristics for 32 typical fault modes. Current monitoring data is matched against the feature database to identify potential fault types. The life prediction model considers three main influencing factors: load cycle count, environmental corrosion level, and operating time, calculating the probability distribution of remaining service life. Prediction results are presented in time intervals, including both optimistic and pessimistic scenarios.

[0067] The fault-tolerance strategy iteratively adjusts control parameters dynamically based on prediction results. Safety margin coefficients are set according to equipment health status levels: standard coefficients are used for new components, appropriately relaxed in the mid-term, and tightened near the end of their lifespan. Compensation weights are allocated based on fault risk levels, with high-risk components receiving higher compensation priority. Parameter adjustments employ a gradual strategy to avoid drastic changes that could lead to system instability. A historical parameter version saving function supports rapid rollback to a stable state. The following is an example of the correspondence between equipment status and fault-tolerance thresholds recorded during a specific operation.

[0068] Table 1: An example of the correspondence between the recorded equipment status and the fault tolerance threshold during a certain actual operation.

[0069]

[0070] Based on this status data, the system automatically calculates a comprehensive fault tolerance time threshold of 63 minutes, which is the upper limit for ensuring safe operation under the current conditions. This threshold is continuously and dynamically updated. When the motor temperature rise rate is detected to accelerate to 1.2℃ / min, the system immediately adjusts the threshold to 47 minutes. At the same time, preventive measures are initiated, including reducing the load sharing of adjacent units and starting the backup cooling fan in advance.

[0071] The fault diagnosis expert system analyzes the spatiotemporal correlation of abnormal data to distinguish between primary faults and transmissive anomalies. The diagnostic process employs a combination of rule-based and case-based reasoning, first matching predefined rule patterns and then retrieving similar historical cases for corroboration. The diagnostic results include a fault probability assessment and suggested remedial measures to assist maintenance personnel in decision-making. A knowledge base update mechanism automatically incorporates validated new cases, continuously improving diagnostic capabilities.

[0072] The maintenance reminder function generates differentiated early warning messages based on the equipment's health status. Regular maintenance reminders are issued 24 hours in advance, including suggested maintenance content and estimated time. Emergency fault warnings immediately trigger audible and visual alarms and automatically reduce system load. Maintenance task priorities are comprehensively assessed based on equipment criticality and fault risk, optimizing maintenance resource allocation. Maintenance records are electronically archived, forming a complete equipment lifecycle file.

[0073] The human-computer interface intuitively displays fault tolerance status information, using color coding to distinguish different warning levels. Real-time monitoring curves overlay display current values ​​and historical trends to assist in judging the development of anomalies. Operation logs record in detail the reasons for each automatic adjustment action and parameter changes, supporting post-event analysis. The access control system ensures that modifications to critical parameters require double confirmation to prevent accidental operations.

[0074] Example 5: The dynamic disturbance quantification module achieves accurate environmental disturbance assessment through multi-dimensional data analysis. This module continuously receives wind speed mutation records and material stress anomaly data from the anomaly fusion diagnostic system, and establishes a spatiotemporal correlation model to analyze the impact mechanism of external disturbances on the system. Wind speed data acquisition uses a three-dimensional ultrasonic anemometer array, with measurement points covering the entire perimeter of the greenhouse according to a three-dimensional grid layout. The sampling frequency is set to 32Hz to ensure the capture of transient characteristics of turbulence. The stress monitoring system consists of an embedded strain gauge network, with the measurement point density determined based on structural sensitivity analysis, focusing on monitoring the edge connection areas and easily deformable central areas of the aerogel insulation blanket.

[0075] The wind speed abrupt change feature extraction process includes spatial gradient analysis and temporal duration determination. Spatial gradient calculation employs the central difference method to establish a wind speed change rate field in a three-dimensional coordinate system, identifying the location and extent of high-gradient regions. Time series analysis uses a change point detection algorithm to segment different wind speed stages, extracting the start time, duration, and intensity change curves of the abrupt change event. Feature parameters include maximum wind speed increment, gradient change rate, and energy spectrum characteristics; these parameters collectively describe the dynamic characteristics of the disturbance.

[0076] Stress anomaly data analysis focuses on the spatiotemporal distribution of material response. Anomaly detection algorithms compare the deviations between real-time strain values ​​and those predicted by the material constitutive model, marking areas exceeding the allowable range of elastic deformation as anomalies. Duration statistics record the entire process of each anomaly event from occurrence to recovery, analyzing the cumulative strain effect. A spatiotemporal correlation model maps stress anomaly points to a three-dimensional coordinate system, establishing a correspondence with the wind speed field, revealing the causal relationship between external excitation and structural response.

[0077] The wind speed-stress coupled response model is established using a nonlinear system identification method. The model input includes the three components of the wind speed vector and its time derivative, and the output is the predicted strain response at key locations. The parameter identification process uses correlation analysis to determine the contribution weight of each input term, highlighting the dominant influencing factors. Model validation employs cross-validation, dividing the measured data into training and test sets to ensure generalization ability. The model update mechanism automatically triggers retraining during seasonal environmental changes to maintain prediction accuracy.

[0078] The dynamic disturbance intensity grading system comprehensively considers both wind speed characteristics and material response. The grading criteria are based on the creep properties and fatigue damage threshold of aerogel materials, classifying disturbance events into five severity levels. Grading parameters include four dimensions: peak wind speed, duration, spatial impact range, and stress response amplitude. The comprehensive intensity value is calculated using a fuzzy comprehensive evaluation method. The grading results are updated in real time, reflecting the dynamic changes in environmental disturbances.

[0079] The adjustment of the drive mechanism protection parameters is based on a differentiated strategy implemented according to the graded results. The dynamic adjustment of the overload protection coefficient employs a smooth transition algorithm to avoid control oscillations caused by sudden parameter changes. The adjustment step size is adaptively determined based on the rate of change of the disturbance intensity; a larger step size is used during the fast response phase, while the step size is reduced during the stable phase to improve accuracy. The protection parameters include multiple dimensions such as current limits, speed thresholds, and acceleration limits, forming a complete set of protection constraints.

[0080] Fault-tolerant response priority management incorporates multi-objective optimization principles. Priority assessment considers a balance between equipment safety, control accuracy, and energy efficiency, dynamically adjusting weighting coefficients based on different operating conditions. High-priority tasks receive faster response times and more system resources, while low-priority tasks are allowed appropriate performance compromises. The priority mapping table is dynamically adjusted based on real-time system load to optimize overall operational efficiency.

[0081] The abnormal event database records complete environmental disturbance cases, forming a knowledge accumulation system. Each event includes raw monitoring data, feature extraction results, and records of handling measures, supporting multi-dimensional retrieval and similar case matching. Data analysis tools provide trend statistics and correlation mining functions to identify regular patterns in disturbance occurrences. The case database maintenance mechanism regularly cleans up low-quality data and adds typical new cases to keep the knowledge fresh.

[0082] The system status visualization interface intuitively displays the spatiotemporal evolution of dynamic disturbances. The wind speed field uses particle animation to simulate flow patterns, stress distribution is represented by color gradients to indicate strain levels, and disturbance intensity levels are marked with warning icons. Historical trend graphs overlay wind speed and stress variation curves to aid in analyzing their response relationship. The operation console allows manual adjustment of some parameters, but critical protection settings require authorization verification.

[0083] An adaptive learning mechanism continuously optimizes the interference handling strategy. The performance evaluation module records the processing effect of each interference event, quantifying indicators such as control deviation, increased energy consumption, and equipment wear. The strategy optimization algorithm analyzes the correspondence between processing measures and results, gradually improving parameter adjustment rules. The learning process employs incremental updates to ensure system stability; major adjustments require offline verification before deployment.

[0084] The early warning information distribution system initiates different notification processes based on the interference level. Low-level alerts are displayed via a status bar icon, medium-level alerts trigger sound prompts and SMS notifications, and high-level alarms simultaneously activate on-site audible and visual alarms and link with the remote monitoring center. The information includes the type of interference, the scope of impact, and recommended measures to help operators quickly understand the situation. A confirmation mechanism requires responsible personnel to sign off on important early warnings, forming a closed-loop management system.

[0085] The maintenance decision support function links equipment health status data. Interference event records are cross-analyzed with equipment maintenance logs to assess the impact of the external environment on equipment lifespan. Maintenance recommendations are generated considering cumulative damage effects, scheduling preventative maintenance before critical components reach fatigue thresholds. Decision reports include risk analysis and cost assessments to assist in developing scientific maintenance plans.

[0086] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0087] 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. A multimodal sensing fusion control system for aerogel greenhouse insulation blankets, characterized in that, include: The environmental multimodal monitoring module is used to collect real-time environmental parameters inside the greenhouse, including temperature and humidity distribution data, light intensity gradient data, soil moisture profile data, and external meteorological change data, and generate a comprehensive environmental monitoring dataset. The thermal insulation mode decision module, based on the comprehensive environmental monitoring dataset, combined with the physical property parameters of the aerogel insulation blanket and historical control records, analyzes the differences in heat loss rate and thermal insulation requirements in different areas, matches dynamic thermal insulation strategies, and generates multi-level zone control instructions. The collaborative parameter optimization module calculates the combination scheme of the aerogel insulation blanket unfolding angle, coverage area and insulation layer thickness according to the multi-level partition control instructions, calibrates the action timing conflict and energy consumption threshold constraint between each execution unit, and adjusts the partition collaborative control parameter set. The multimodal execution controller, based on the partitioned collaborative parameter set, statistically analyzes the deformation feedback data of the aerogel insulation blanket and the target position deviation, pairs the material deformation characteristics with the drive mechanism response curve, calculates the execution compensation amount and path replanning range, reassembles the drive signal sequence and the execution unit collaborative strategy, and generates a dynamic control map of the insulation blanket. The abnormal fusion diagnosis module, based on the dynamic control map of the insulation blanket, monitors material stress anomalies and environmental mutation interference during the execution process, adjusts the driving parameter compensation mechanism and collaborative timing fault tolerance strategy, and generates a set of control error correction factors.

2. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 1, characterized in that, The specific steps for generating the comprehensive environmental monitoring dataset are as follows: To meet the needs of monitoring the internal environment of greenhouses, a raw data acquisition model for multimodal sensors was developed. A spatiotemporal alignment algorithm was used to perform timestamp calibration and spatial grid mapping on temperature and humidity distribution data, light intensity gradient data, and soil moisture profile data; Identify the associated feature vectors of each sensor data in the spatial grid, including the features of heat conduction hysteresis effect, the nonlinear features of light attenuation, and the associated features of soil moisture migration, and form a multimodal feature fusion matrix; The environmental parameters in the multimodal feature fusion matrix are coupled and analyzed separately. The thermodynamic coupling region, the photothermal conversion efficiency region, and the soil moisture dynamic influence region are extracted in sequence as the basis for partition mapping of the comprehensive environmental monitoring dataset.

3. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 2, characterized in that, The steps for forming the multimodal feature fusion matrix further include: The characteristics of heat conduction hysteresis effect and the nonlinear characteristics of light attenuation are superimposed on the time series to obtain the multiphysics coupling results; Temperature gradient extrema are extracted from the multiphysics coupling results. These extrema are then compared with a phase transition threshold library for aerogel materials to generate a thermodynamic stability identifier set. The cross-influence factors of diffusion rate and heat conduction rate in soil moisture migration correlation characteristics are extracted from the multiphysics coupling results. The multiphysics coupling results are reconstructed according to the diffusion rate threshold and heat conduction fluctuation rate to generate a soil moisture-thermodynamic dynamic equilibrium parameter table.

4. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 1, characterized in that, The specific steps for generating the multi-level partition control instructions are as follows: The thermodynamic stability identifier set and soil moisture-thermodynamic dynamic equilibrium parameter table in the comprehensive environmental monitoring dataset are called up, and the heat preservation priority of different areas is analyzed through energy consumption optimization algorithm. Identify the deployment mechanism state parameters of each section of the aerogel insulation blanket, including the motor torque response curve and the material tensile strength attenuation coefficient; Based on the state parameters of the deployment mechanism and the heat preservation priority, a target control model is constructed: Successful control samples from historical control records are collected, and the optimal value of driving energy consumption and the threshold of deformation control accuracy are extracted as training labels. The thermodynamic stability identifier set is input into a temporal prediction network with variable kernel width to output an initial control scheme for the action sequence of the driving mechanism. The timing conflict nodes of the initial control scheme are corrected by the moisture-thermodynamic dynamic equilibrium parameter table, and multi-level zone control instructions are generated.

5. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 4, characterized in that, The specific steps for adjusting the partitioned collaborative control parameter set are as follows: Extract the target position sequence and action trigger time window of each execution unit from the multi-level partition control instructions; Analyze the mechanical interference area and peak power load period between adjacent execution units, and mark the action timing conflict domain and instantaneous energy consumption over-limit risk points; The spatiotemporal obstacle avoidance algorithm is used to recalculate the action delay compensation of each execution unit, and combined with the deformation relaxation characteristics of aerogel material, a partitioned collaborative fault-tolerant time window parameter table is generated. By integrating the mechanical interference region marking results with the fault-tolerant time window parameter table, the phase difference of the driving signal and the power compensation coefficient are calibrated to generate a partitioned collaborative control parameter set.

6. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 5, characterized in that, The specific steps for generating the dynamic control map of the thermal insulation blanket are as follows: The partitioned collaborative control parameter set is input into the deformation-drive coupling model to predict the real-time deformation trajectory of the aerogel insulation blanket. Collect the actual deformation trajectory point set fed back by high-precision displacement sensor, and calculate the point-by-point position deviation between the predicted trajectory and the actual trajectory; Identify abnormal deformation regions where the positional deviation exceeds the material's elastic threshold, and extract the current fluctuation characteristics of the drive mechanism and the records of sudden changes in environmental wind speed in the corresponding regions. Based on the current fluctuation characteristics and wind speed change records, the amplitude-frequency adjustment curve of the driving signal is reconstructed to generate a coverage path replanning vector. By integrating the position deviation analysis results with the path replanning vector, the driving pulse width modulation sequence is recombined, and the dynamic control spectrum of the insulation blanket is output.

7. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 6, characterized in that, The specific steps for generating the set of adjustment error correction factors are as follows: Historical regulation records and current environmental mutation data of abnormal deformation regions were extracted from the dynamic regulation map of thermal insulation blankets. A coupled analysis model of material stress and environmental disturbance is constructed to calculate the nonlinear relationship between deformation compensation and actual energy consumption increment; Identify the periods of continuous disturbance and instantaneous impact events in environmental wind speed change records, and calibrate the passive fault tolerance time threshold by combining the creep recovery time constant of aerogel materials. Based on the aforementioned nonlinear relationship and passive fault tolerance time threshold, the overload protection coefficient and torque attenuation compensation gradient of the drive mechanism are dynamically adjusted. A set of regulation error correction factors is generated by using a multi-execution unit synergy evaluation matrix.

8. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 7, characterized in that, The calibration step for the passive fault tolerance time threshold also includes: Extract real-time operating data of adjacent execution units, including motor temperature rise curves and gearbox vibration spectrum; Analyze the heat dissipation efficiency decay node in the motor temperature rise curve and the abnormal resonance frequency band of the gearbox vibration spectrum; By mapping the heat dissipation efficiency decay node and the abnormal resonant frequency band to the time constraint domain, the failure early warning time window of the mechanical system is calculated. The minimum value between the mechanical system failure warning time window and the duration of the environmental wind speed change record is taken as the passive fault tolerance time threshold.

9. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 1, characterized in that, Also includes: The global energy efficiency assessment module is used to integrate the real-time data stream of the environmental multimodal monitoring module with the error correction factor set of the anomaly fusion diagnosis module to construct a dynamic assessment model of thermal insulation performance. The predictive maintenance module, based on the output of the thermal insulation performance dynamic evaluation model, correlates with the historical fault feature library of the drive mechanism to predict the life decay curve of the execution unit and the material fatigue critical point. The fault-tolerant strategy iteration module dynamically updates the safety margin coefficient in the partitioned collaborative control parameter set and the compensation weight value in the control error correction factor set based on the life decay curve and the material fatigue critical point.

10. The multimodal sensing fusion control system for aerogel greenhouse insulation blankets according to claim 1, characterized in that, It also includes a dynamic interference quantization module, the implementation of which includes: The system acquires environmental wind speed abrupt change records and material stress anomaly data from the anomaly fusion diagnosis module, and extracts the spatial gradient distribution characteristics of wind speed abrupt changes and the duration sequence of stress anomalies. A wind speed-stress coupled response model was constructed, and the nonlinear correlation coefficient between stress fluctuation and wind speed change per unit time was calculated. The system identifies periods of continuous interference where the nonlinear correlation coefficient exceeds the material creep threshold, and generates dynamic interference intensity grading parameters by combining the deformation recovery hysteresis characteristics of aerogel insulation blankets. Based on the dynamic interference intensity classification parameters, the dynamic adjustment step size of the overload protection coefficient of the drive mechanism is calibrated, and the fault-tolerant response priority in the control error correction factor set is updated.

Images