Variable air volume valve control method and system, electronic device, and readable storage medium

Abstract

The application relates to the technical field of air volume control, and discloses a variable air volume valve control method and system, an electronic device and a readable storage medium, wherein the variable air volume valve control method comprises the following steps: collecting multi-source sensing data in the operation process of a variable air volume valve, checking the multi-source sensing data to identify abnormal data; replacing the abnormal data based on historical reliable data to generate repaired sensing data; inputting the repaired sensing data into a trained air volume fusion calculation model to output a fused air volume value; obtaining a target air volume set value, generating a valve control instruction based on the air volume deviation between the fused air volume value and the target air volume set value to control the variable air volume valve. The application effectively prevents frequent valve shaking, air volume oscillation and system surge caused by sudden changes in original data, and improves the basic robustness and operation continuity of variable air volume valve control.

Description

Technical Field

[0001] This application relates to the field of air volume control technology, and in particular to a variable air volume valve control method, system, electronic device and readable storage medium. Background Technology

[0002] In building automation systems, variable air volume (VAV) air conditioning systems are widely used in commercial offices, hospitals, laboratories, and other places where high precision in controlling indoor environmental parameters is required. This system maintains a set temperature by adjusting the airflow delivered to each area. Its core actuator is the VAV valve, and the dynamic adjustment of the valve position relies on real-time feedback data from sensors such as temperature, pressure, and wind speed, which is then controlled by a controller using a preset control algorithm in a closed-loop manner. In actual engineering deployments, various sensors operate under complex electromagnetic environments and fluctuating temperature and humidity conditions for extended periods, making them susceptible to interference, aging, or momentary contamination. This can lead to millisecond-level abrupt changes, spikes, or step jumps in the output signal. Although these abnormal data are short-lived, if they are directly involved in control calculations without effective identification and suppression, they will cause severe oscillations in the controller output, resulting in frequent start-stop, over-adjustment, or even mechanical shock of the valve actuator. This leads to significant deviations of the airflow parameters from the set value, deterioration of regional thermal comfort, and significantly increased actuator wear and system energy consumption, ultimately reducing the operational stability and long-term reliability of the entire VAV control system. Summary of the Invention

[0003] In view of this, embodiments of this application provide a variable air volume valve control method, system, electronic device, and readable storage medium, which can effectively solve the problems of control instability and system reliability degradation caused by sudden changes in sensor data in the existing variable air volume valve control system.

[0004] In a first aspect, embodiments of this application provide a variable air volume valve control method, the method comprising: Collect multi-source sensor data during the operation of the variable air volume valve; Verify multi-source sensor data to identify abnormal data; Abnormal data is replaced based on historical reliable data to generate repaired sensor data; The repaired sensor data is input into the trained air volume fusion calculation model, and the fused air volume value is output. The target air volume setpoint is obtained, and based on the air volume deviation between the fused air volume value and the target air volume setpoint, a valve control command is generated to control the variable air volume valve.

[0005] In some embodiments, the collected multi-source sensor data is verified, and abnormal data is identified, including: Construct a sliding time window containing a preset number of continuous sampling points to calculate the standardized values ​​of each detection signal of the multi-source sensor data within the sliding time window; Assign a first weighting coefficient to each detection signal, and calculate the weighted anomaly index based on the standardized value of each detection signal and the corresponding first weighting coefficient. Map the weighted abnormality index to a health score; If the health score is outside the preset health threshold range, the multi-source sensor data within the sliding time window is determined to be an abnormal data group. Detection signals whose standardized values ​​in the abnormal data group are outside the preset standardized value range are identified as abnormal data.

[0006] In some embodiments, replacing anomalous data with historical reliable data to generate repaired sensing data includes: Filter multi-source sensor data samples whose health scores fall within a preset health threshold range within a historical time window; A second weighting coefficient is assigned to the detection signal samples of the same type as the abnormal data in each multi-source sensor data sample. The second weighting coefficient decreases exponentially with the increase of time distance. Based on each detected signal sample and its corresponding second weighting coefficient, a weighted average operation is performed on the abnormal data to obtain the replacement value; Abnormal data is replaced with substitute values ​​to generate repaired sensor data.

[0007] In some embodiments, training the air volume fusion calculation model includes: The feedforward neural network model was initially trained using an offline dataset to obtain the air volume fusion calculation model; When the health score of the current multi-source sensor data is within the preset health threshold range and the corresponding air volume deviation is less than the preset deviation threshold, the current multi-source sensor data and the corresponding fused air volume value are saved as a fine-tuning sample. When the number of accumulated fine-tuning samples reaches the preset number of samples to form a sample set, the air volume fusion calculation model is incrementally updated and trained based on the sample set.

[0008] In some embodiments, generating valve control commands based on the airflow deviation between the fused airflow value and the target airflow setpoint includes: Calculate the rate of change between the target air volume setpoint and the target air volume setpoint of the previous period; When the rate of change exceeds a preset change threshold within a preset time, an intermediate target air volume sequence is generated based on a preset step size. Based on the current fused air volume value and the intermediate deviation air volume of each intermediate target value in the intermediate target air volume sequence, corresponding intermediate valve control commands are generated to control the variable air volume valves in sequence.

[0009] In some embodiments, after generating corresponding intermediate valve control commands to sequentially control the variable air volume valve, the method further includes: Based on each intermediate target value and its corresponding fused air volume value, calculate the tracking error corresponding to each intermediate target value; When the tracking error exceeds the first preset error threshold for a preset number of consecutive times, the generation time of the remaining intermediate target values ​​in the intermediate target air volume sequence is extended. When the tracking error is negative and the absolute value of the tracking error is greater than the second preset error threshold, the next intermediate target value of the intermediate target air volume sequence will be reduced.

[0010] In some embodiments, after obtaining the target airflow setpoint and generating a valve control command based on the airflow deviation between the fused airflow value and the target airflow setpoint, the control method further includes: Calculate the absolute difference between the current valve control command and the valve control command of the previous cycle; When the absolute difference is within the preset range, the current valve control command is output to control the variable air volume valve; When the absolute difference is outside the preset range, the instruction from the previous cycle remains unchanged.

[0011] Secondly, embodiments of this application provide a variable air volume valve control system, comprising: The data acquisition module is used to collect multi-source sensor data during the operation of the variable air volume valve; The verification module is used to verify multi-source sensor data in order to identify abnormal data; The replacement module is used to replace abnormal data based on historical reliable data and generate repaired sensor data; The input module is used to input the repaired sensor data into the trained air volume fusion calculation model and output the fused air volume value. The generation module is used to obtain the target air volume setpoint and generate valve control commands to control the variable air volume valve based on the air volume deviation between the fused air volume value and the target air volume setpoint.

[0012] Thirdly, embodiments of this application provide an electronic device, which includes a processor and a memory. The memory stores a computer program, and the processor executes the computer program to implement the variable air volume valve control method of the first aspect described above.

[0013] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the variable air volume valve control method of the first aspect described above.

[0014] The embodiments of this application have the following beneficial effects: This solution collects multi-source sensor data, including differential pressure, temperature, atmospheric pressure, and actual valve opening signals. It dynamically verifies the data to identify anomalies, then replaces the anomalies with historical reliable data to generate repaired sensor data. This repaired data is then input into a trained airflow fusion calculation model to output a fused airflow value. Finally, valve control commands are generated based on the deviation between this fused airflow value and the target airflow setpoint. This approach achieves proactive sensing and online suppression of transient sensor distortions, avoiding misjudgments caused by single-point sensor failures or interference. It ensures that airflow calculations are always based on reliable inputs and guarantees the rationality and stability of valve control commands, thereby improving the control stability and system reliability of the variable airflow valve. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This paper shows a first flowchart of a variable air volume valve control method according to an embodiment of the present application. Figure 2 This paper shows a second flowchart of the variable air volume valve control method according to an embodiment of the present application; Figure 3 A schematic diagram of the third process of the variable air volume valve control method according to an embodiment of this application is shown; Figure 4 A schematic diagram of the fourth process of the variable air volume valve control method according to an embodiment of this application is shown; Figure 5 A schematic diagram of a variable air volume valve control system according to an embodiment of this application is shown. Detailed Implementation

[0017] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0018] The components of the embodiments of this application described and illustrated in the accompanying drawings can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0019] In the following text, the terms "comprising," "having," and their cognates, which may be used in various embodiments of this application, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more combinations thereof. Furthermore, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0020] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this application pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be construed as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this application.

[0021] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0022] Considering the problems of control instability and decreased system reliability caused by sudden changes in sensor data in existing variable air volume (VAV) valve control systems, this application provides a VAV valve control method, system, electronic device, and readable storage medium. This method achieves active sensing and online suppression of transient sensor distortion, avoids misjudgments caused by single-point sensor failures or interference, ensures that airflow calculations are always based on reliable input, guarantees the rationality and stability of valve control commands, and improves the control stability and system reliability of the VAV valve.

[0023] The following describes the control method for the variable air volume valve using specific examples.

[0024] Figure 1 A schematic flowchart of a variable air volume (VAV) valve control method according to an embodiment of this application is shown. Exemplarily, the VAV valve control method of this application embodiment can be applied to any type of VAV valve system. It can be integrated into the valve's built-in controller or configured independently as an edge computing unit. It interacts with the building automation system via a communication protocol to achieve real-time identification, reliable assessment, and autonomous correction of sudden sensor data changes, thereby ensuring the continuity, accuracy, and system-level robustness of airflow control.

[0025] Exemplary, this variable air volume valve control method includes S101-S105: S101 collects multi-source sensor data during the operation of the variable air volume valve. Exemplary examples show that multi-source sensor data includes differential pressure signals, temperature signals, atmospheric pressure signals, and actual valve opening signals. Different types of sensors can be configured to collect various data points during the operation of the variable air volume (VAV) valve; for example, a differential pressure sensor can be used to measure the differential pressure signal of the VAV valve. Temperature signals are measured using a temperature sensor. Atmospheric pressure signals are measured using a barometric pressure sensor. Or calculate atmospheric pressure signals using preset altitude. The actual valve opening signal is calculated through voltage feedback from the variable air volume valve. And form a time-series data vector. Furthermore, time-series data vectors can be processed. Preprocessing is performed to obtain standardized vectors. As multi-source sensor data, for example, unifying the units of various signals, such as temperature signals... Converted to Kelvin (K), actual valve opening signal Using binary units, low-pass filtering can be applied to each signal to suppress high-frequency noise, thereby ensuring that multi-source sensor data participate in subsequent joint analysis within a unified physical dimension and dynamic range, laying a data foundation for high-precision anomaly identification.

[0026] S102 verifies multi-source sensor data to identify abnormal data.

[0027] As an example, a time-series anomaly detection algorithm can be used to analyze the multi-source sensor data corresponding to each sampling point in a preset time window to obtain a health score. Based on the health score and relevant thresholds, it can be determined whether the multi-source sensor data in the time window is an abnormal data group. Based on the standardized value corresponding to each detection signal in the abnormal data group, it can be determined whether each detection signal is abnormal data.

[0028] For example, the time series anomaly detection algorithm uses a sliding time window, standardized values ​​formed by multiple variables, and multi-weight constraints, which takes into account both lightweight design and real-time performance and deployment feasibility in resource-constrained environments, avoiding the high computing power requirements and uninterpretable defects brought about by using complex deep learning models such as LSTM-AE.

[0029] For example, in one implementation, such as Figure 2 As shown, S102 includes the following sub-steps: S201, construct a sliding time window containing a preset number of continuous sampling points, so as to calculate the standardized value of each detection signal of the multi-source sensor data within the sliding time window.

[0030] For example, the sliding time window length N=5, the sampling period is 100ms, and the total coverage time is 500ms; the multi-source sensor data of each group of sliding time window are collected, and the detection signals of each group of multi-source sensor data are calculated respectively. The corresponding mean and standard deviation, and based on each detection signal. Calculate the corresponding mean and standard deviation for each detection signal. The corresponding standardized value, where if the current detection signal If the standard deviation is less than the minimum standard deviation threshold, then the current detection signal will be... The standardized value is set to zero to prevent the standard deviation from approaching zero due to minimal signal fluctuations when the system is in a static or steady-state condition, which could lead to divergence of the standardized value and false alarms. This ensures that the anomaly criteria have stable triggering capability and a low false alarm rate under all actual operating conditions.

[0031] S202, assign a first weighting coefficient to each detection signal, and calculate the weighted anomaly index based on the standardized value of each detection signal and the corresponding first weighting coefficient.

[0032] The first weighting coefficient can be set according to the physical sensitivity of each detection signal in the variable air volume valve control system: for example, differential pressure signal. The weight is 0.4, due to the differential pressure signal. It is a core parameter for airflow calculation and is highly sensitive to sudden changes; temperature signal The weight is 0.3, reflecting the temperature signal. Physical characteristics of high thermal inertia and slow change under normal operating conditions; atmospheric pressure signal. Atmospheric pressure signal with a weight of 0.1. The daily rate of change is extremely small; any significant sudden change is highly likely to indicate a sensor malfunction; the actual valve opening signal The weight is 0.2, reflecting the objective constraint that the actuator has a mechanical response speed limit.

[0033] The weighted anomaly index A is defined as follows: ,in, Indicates the detection signal The weights; Indicates the detection signal The standardized value; =5, used to truncate extreme standardized values ​​to suppress strong single-point interference; different detection signals are assigned through a weighting mechanism. The discriminant weights, which are matched to its physical importance, give the weighted anomaly index A clear engineering interpretability.

[0034] S203 maps the weighted abnormality index to a health score.

[0035] For example, the mapping function uses an exponential decay form, and the health score... ,in Ensure that when the weighted abnormal index Health score Weighted abnormality index Health score Final health score Strictly limited to the interval This nonlinear mapping correlates slight deviations with severe faults in health scores. The above exhibits significant differentiation, facilitating the subsequent implementation of tiered response strategies.

[0036] S204. If the health score is outside the preset health threshold range, the multi-source sensor data within the sliding time window is determined to be an abnormal data group.

[0037] For example, the preset health threshold range is [85, 100]. This range has been verified through extensive field calibration and typical fault reproduction, ensuring that the differential pressure signal ΔP returns to zero, the temperature signal T rises by more than 5℃ / s, and the actual valve opening signal... The trend of the differential pressure signal ΔP is significantly deviated from that of the atmospheric pressure signal. While achieving a high detection rate for key abnormal patterns such as mutations, the false alarm rate of small fluctuations during normal adjustment processes is controlled to an acceptable level, providing a reliable and reproducible decision-making basis for accurate positioning of subsequent abnormal data.

[0038] S205, the detection signal whose corresponding standardized value in the abnormal data group is outside the preset standardized value range is determined as abnormal data.

[0039] Among them, each detection signal The corresponding preset standardized value ranges can be the same or different, and can be set according to the actual application. By setting the preset standardized value range, precise location of anomalies from data group level to signal level is achieved, providing a clear target for subsequent differentiated weighted repair strategies based on historical reliable data.

[0040] S103, replaces abnormal data with historical reliable data to generate repaired sensor data.

[0041] Exemplary, when the health scores corresponding to multi-source sensor data When the data score falls below the preset health threshold range of 85, a data reliability weighting and outlier replacement mechanism is activated. This mechanism uses historical valid data with a health score ≥ 85 to calculate a moving weighted average, replacing outlier data identified in multi-source sensor data, or temporarily excluding it from the current cycle calculation in extreme cases. This is to prevent single-point outlier data from polluting the overall control link, and ultimately repair the sensor data. This mechanism abandons crude fault tolerance methods such as simple discarding or static mean filling. Through dynamic reliable data screening and time-aware weighting, it ensures that the repaired data not only reflects the true physical state of the system, but also has sufficient time continuity and trend rationality, significantly improving the system's continuous operation capability and decision reliability in scenarios of partial sensor failure or strong interference.

[0042] For example, in one implementation, such as Figure 3 As shown, S103 includes the following sub-steps: S301, filter multi-source sensor data samples whose health scores are within the preset health threshold range within the historical time window.

[0043] Exemplary, the historical time window is the current moment. The time interval was traced back 10 seconds, with a sampling frequency of 5Hz, meaning a set of data was collected every 200ms, covering a total of 50 sampling points; only the health score was included. Samples that meet the criteria are included in the trusted sample set, while other outliers are automatically removed; the trusted sample set is the set of samples taken at the current time. The corresponding history window is All those who meet the health score The sample set; through the dual-constraint window design of fixed time length and quality screening, it is ensured that the data time span is sufficient to cover the typical dynamic response process of the system, and the health score hard threshold ensures that all historical samples involved in the calculation have high confidence, providing a high-quality data foundation for subsequent weighted repair. Furthermore, if there are too few multi-source sensor data samples, for example, only 3 multi-source sensor data samples are selected within the historical time window, the valve control command is generated using the fused air volume value of the previous period, avoiding inaccurate repair due to data sparsity.

[0044] S302 assigns a second weighting coefficient to the detection signal samples of the same type as the abnormal data in each multi-source sensor data sample. The second weighting coefficient decreases exponentially with the increase of time distance.

[0045] Specifically, an exponential decay weighting strategy is used to assign weights to the k-th historical valid sample within the window. = Where K = 0, 1, 2, 3, 4, 5..., k = 0 represents the latest sample. =0.95 is an empirical value; the weights satisfy the normalization condition. This fully reflects the slow changes in the system's thermal and airflow processes, making the weight of data from 1 second ago approximately 0.77 and data from 5 seconds ago approximately 0.28. This truly achieves a consistent physical weighting with a focus on near-term data and a focus on distant data, avoiding the distortion of current repair values ​​due to outdated data from the past. This ensures that the repaired sensor data can quickly track changes in the system's transient operating conditions while maintaining steady-state accuracy.

[0046] S303: Based on each detected signal sample and its corresponding second weighting coefficient, perform a weighted average operation on the abnormal data to obtain a replacement value.

[0047] Furthermore, considering the differences in physical recoverability of different detection signals, a graded processing strategy is adopted for differential pressure signals. Temperature signal and atmospheric pressure signal Forced substitution using an exponentially weighted average is used. These signals have large physical inertia, and short-term abrupt changes almost always originate from sensor noise or malfunction; the actual state has a high degree of temporal continuity. For the actual valve opening signal... This only applies if the host computer has not issued a new instruction, or if it has issued one but the current opening degree is not yet set. With instruction opening Weighted average substitution is only used when the absolute deviation is less than 5°; if Then retain the current opening degree. This prevents erroneous corrections caused by actuator response lag or jamming. The hierarchical strategy takes into account the slow-changing characteristics of airflow and thermodynamic systems as well as the electromechanical response characteristics of actuators, avoiding control logic confusion caused by one-size-fits-all replacements, and maintaining the integrity and traceability of system action intent while ensuring data reliability.

[0048] S304 uses substitute values ​​to replace abnormal data and generates repaired sensor data.

[0049] By replacing abnormal data with substitute values, the air volume fusion calculation model can have continuous, smooth and reliable multi-source input under any operating condition. This fundamentally eliminates the risk of valve malfunction, air volume oscillation and system surge caused by abnormal data, and realizes the key data governance link of preventing data mutation and adaptive correction.

[0050] S104 inputs the repaired sensor data into the trained airflow fusion calculation model and outputs the fused airflow value.

[0051] The air volume fusion calculation model is a lightweight feedforward neural network, replacing the traditional model that relies solely on differential pressure signals and valve opening signals. The lookup table method, through deep fusion of differential pressure signals Temperature signal and atmospheric pressure signal and valve actual opening signal Four-dimensional input enables dynamic modeling and online compensation of non-ideal flow characteristics of variable air volume (VAV) valves in actual engineering environments, caused by factors such as installation deviations, valve plate dust accumulation, component aging, temperature drift, and sensor zero-point drift. Furthermore, the airflow fusion calculation model strictly meets edge deployment constraints: it is compatible with Cortex-M7 core controllers, Flash usage <50KB, peak RAM usage ≤256KB, and single inference time <1ms, ensuring real-time and stable operation in resource-constrained valve built-in controllers, providing millisecond-level airflow sensing capabilities for high-precision closed-loop control.

[0052] As an example, the air volume fusion calculation model adopts a three-layer fully connected feedforward neural network architecture, with the input layer receiving 4D repaired sensor data. The first hidden layer contains 16 neurons and uses the ReLU activation function, the second hidden layer contains 8 neurons and uses the ReLU activation function, and the output layer is a one-dimensional linear output, directly regressing to predict the fused air volume value. This structure ensures nonlinear fitting capability while compressing the number of parameters and computational complexity to the greatest extent, balancing model expressiveness and MCU real-time requirements, avoiding the introduction of high-overhead structures such as LSTM and Transformer, and conforming to the engineering implementation principles of being sufficient, reliable, and embeddable under slow-changing system conditions.

[0053] As an example, the feedforward neural network model can be initially trained using an offline dataset to obtain the airflow fusion calculation model; the offline training set consists of differential pressure signals synchronously acquired in a wind tunnel laboratory or on a factory calibration bench. Temperature signal and atmospheric pressure signal and valve actual opening signal The dataset uses four-dimensional sensor data and high-precision thermal anemometer measurements as the fused air volume data set; the operating conditions cover valve opening of 10° to 90°, wind speed of 3 to 15 m / s, and ambient temperature of 10 to 40°, and specifically simulates the dust accumulation state of valve plates, accumulating ≥10,000 sets of high-quality samples; through offline training, the model has the ability to generalize and recognize multi-source disturbances from the beginning, laying the foundation for calibration-free start-up on site.

[0054] The model training uses weighted average absolute error as the main loss function and introduces a physical constraint regularization term with a weight coefficient λ = 0.1. This forces the model to output air volume Q close to zero when the pressure difference signal ΔP approaches zero. This embeds the laws of mass conservation and fluid mechanics into the mathematical modeling, preventing non-physical interpretations such as output air volume Q ≠ 0 when the pressure difference signal ΔP = 0. This ensures that the model output always has engineering interpretability and safety boundaries. After training, the model weights are exported in standard C language array format and directly compiled and embedded into the controller firmware. No additional runtime inference framework is required, significantly reducing the deployment threshold and maintenance costs.

[0055] Furthermore, when the health score of the current multi-source sensor data is within a preset health threshold and the corresponding airflow deviation is less than a preset deviation threshold, the current multi-source sensor data and the corresponding fused airflow value are saved as a fine-tuning sample. When the cumulative number of saved fine-tuning samples reaches a preset number to form a sample set, the airflow fusion calculation model is incrementally updated and trained based on the sample set. For example, the preset deviation threshold is... The preset sample size is 100. Training uses a single-step stochastic gradient descent method with a learning rate η = 1e^(-1 / n). -5 Furthermore, only the output layer bias parameters are updated, while the remaining weights are frozen. This approach not only allows for continuous optimization of model accuracy using real-world operational data, but also minimizes the risk of catastrophic forgetting and model oscillations, ensuring long-term operational reliability.

[0056] Traditional methods only establish the fusion of air volume value Q and pressure difference signal. Valve actual opening signal The two-dimensional mapping lacks the ability to identify and compensate for interference such as sudden changes in sensor movement, dust accumulation, and temperature drift. Once the differential pressure signal... A sudden change would cause the output of the fused air volume value Q to fluctuate violently; however, the air volume fusion calculation model of this application uses four-dimensional reliable sensor data as input, significantly improves anti-interference through a multivariate cross-validation mechanism, and has adaptive capability, realizing software-level autonomous calibration of the performance drift of the equipment throughout its entire life cycle, greatly improving the long-term control accuracy, energy efficiency stability and engineering applicability of the system.

[0057] S105: Obtain the target air volume setpoint, and based on the air volume deviation between the fused air volume value and the target air volume setpoint, generate a valve control command to control the variable air volume valve.

[0058] The target airflow setpoint can be sent in real time by the airflow control system to the controller integrated within the variable airflow valve, ensuring that the control commands are dynamically synchronized with the regional load demand. The controller compares the current integrated airflow value with the target airflow setpoint in real time. If there is a deviation between the two, the actual valve opening signal is dynamically adjusted through a PID algorithm. This drives the current integrated air volume value to approach the target air volume setting value infinitely; if the current integrated air volume value is equal to the target air volume setting value, the current opening degree remains unchanged, significantly improving the stability of the air supply system and the thermal comfort of the terminal.

[0059] Furthermore, a predictive smoothing strategy can be introduced for the target airflow setpoint. If the change in the target airflow setpoint does not exceed a corresponding threshold within a unit of time, the target airflow setpoint is output. If the change in the target airflow setpoint exceeds the corresponding threshold, a progressive intermediate target airflow sequence is generated.

[0060] For example, in one implementation, such as Figure 4 As shown, S105 includes the following sub-steps: S401, calculate the rate of change between the target air volume setpoint and the target air volume setpoint of the previous cycle.

[0061] Specifically, rate of change ,in, This indicates the latest received target airflow setting. This indicates the target airflow value used at the current moment, i.e., the target airflow setpoint from the previous period, and the rate of change. Calculation is the core criterion for triggering subsequent smoothing strategies. It directly reflects the intensity of dynamic adjustment of the upper system to the terminal air volume demand, and provides a quantitative basis for the controller to predict the mechanical stress of the actuator and the risk of airflow disturbance.

[0062] S402, when the rate of change exceeds the preset change threshold within a preset time, an intermediate target air volume sequence is generated based on a preset step size.

[0063] As an example, the preset change threshold is: ,when ,in, To control the sampling period of the system, a typical value of 1 second is used. If this is deemed a drastic command change, a predictive smoothing strategy is initiated. The smoothing process uses linear interpolation to generate a progressive intermediate target airflow sequence. ,in Increment at each step , Smoothing steps The time span in the smoothing process can be a fixed time window or based on the rate of change. The time window for dynamic adjustment is provided. The linear strategy of this application is simple to implement and has a smooth transition, taking into account both the mechanical response limit of the actuator and the requirements of airflow stability. While ensuring the traceability of commands, it avoids sudden valve opening and stopping, air duct pressure oscillation and air supply surge caused by step jump changes. It is especially suitable for clean environments where the stability of airflow organization is extremely sensitive.

[0064] S403 generates corresponding intermediate valve control commands based on the intermediate deviation air volume of each intermediate target value in the current fused air volume value and the intermediate target air volume sequence, and controls the variable air volume valves in sequence.

[0065] During smooth execution, based on the intermediate deviations between the current fused air volume value and each intermediate target value, corresponding intermediate valve control commands are generated sequentially to control the variable air volume valve.

[0066] Furthermore, during the process of controlling the variable air volume valve sequentially according to the intermediate valve control command, the intermediate target air volume sequence can be adjusted according to the tracking error. For example, in one embodiment, the tracking error corresponding to each intermediate target value is calculated based on each intermediate target value and the corresponding fused air volume value. When the tracking error exceeds the first preset error threshold for a preset number of times, the generation time of the remaining intermediate target values ​​in the intermediate target air volume sequence is extended. When the tracking error is negative and the absolute value of the tracking error is greater than the second preset error threshold, the next intermediate target value of the intermediate target air volume sequence is reduced.

[0067] Specifically, the difference between the intermediate target value at each step and the measured fused air volume value at the corresponding time is calculated in real time as the tracking error, and the subsequent trajectory is dynamically optimized based on the tracking error: when the tracking error exceeds the first preset error threshold twice in a row, the remaining smoothing step size is automatically extended and the subsequent slope is reduced to match the actual response capability; when the tracking error is negative and the absolute value is greater than the second preset error threshold, the next intermediate target value is slightly adjusted back to suppress overshoot. The first preset error threshold and the second preset error threshold can be set according to the actual application. The closed-loop feedback trajectory correction mechanism of this application gives the smoothing process adaptive capability, so that the controller can maintain high-precision tracking performance under unknown disturbances such as actuator aging, changes in air duct resistance or increased dust accumulation, fundamentally eliminating the problem of steady-state error accumulation and repeated oscillation that is easily caused by open-loop smoothing strategies.

[0068] Furthermore, the control method also includes calculating the absolute difference between the current valve control command and the valve control command of the previous cycle. When the absolute difference is within a preset range, the current valve control command is output to control the variable air volume valve; when the absolute difference is outside the preset range, the command of the previous cycle remains unchanged. For example, the preset range can be set to 2%–10% opening angle. When the absolute difference is within the preset range, the current valve control command is output to drive the actuator; when the absolute difference exceeds the preset range, the command of the previous cycle remains unchanged, and only the internal command register is updated, thereby suppressing high-frequency micro-jitter.

[0069] For example, if the current valve control command increases slightly and the absolute difference is outside the preset range, the controller determines it as noise disturbance rather than a real change in demand and actively blocks the command. This avoids frequent start-stop of the actuator, increased mechanical wear, abnormal wear of the valve sealing surface, and the resulting risk of air leakage. Understandably, this control only applies to the final output command. All other processes, such as sensing acquisition, anomaly detection, data repair, and model output, continue to operate normally, ensuring that the system's status perception and fault diagnosis capabilities are not affected. This balances the dual objectives of control silence and system observability.

[0070] like Figure 5 As shown, based on the method of the above embodiments, this embodiment provides a variable air volume valve control system. Exemplarily, the variable air volume valve control system 100 includes: The data acquisition module 110 is used to acquire multi-source sensor data during the operation of the variable air volume valve.

[0071] The verification module 120 is used to verify multi-source sensor data in order to identify abnormal data.

[0072] Replacement module 130 is used to replace abnormal data based on historical reliable data and generate repaired sensing data.

[0073] Input module 140 is used to input the repaired sensor data into the trained air volume fusion calculation model and output the fused air volume value.

[0074] The generation module 150 is used to obtain the target air volume setpoint and generate valve control commands to control the variable air volume valve based on the air volume deviation between the fused air volume value and the target air volume setpoint.

[0075] It is understood that the system in this embodiment corresponds to the control method in the above embodiments, and the options in the above embodiments are also applicable to this embodiment, so they will not be described again here.

[0076] This application also provides an electronic device, exemplary of which includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to enable the device to perform the functions of the various modules in the above-described variable air volume valve control method or the above-described variable air volume valve control system.

[0077] The processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including at least one of a Central Processing Unit (CPU), Graphics Processing Unit (GPU), Network Processor (NP), Digital Signal Processor (DSP), Application-Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0078] Memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), and Electrically Erasable Programmable Read-Only Memory (EEPROM). Memory is used to store computer programs, and the processor can execute these programs upon receiving execution instructions.

[0079] This application also provides a computer-readable storage medium for storing computer programs used in the aforementioned terminal devices. For example, the computer-readable storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0080] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that, as an alternative implementation, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0081] In addition, the functional modules or units in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0082] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a smartphone, personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application.

[0083] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for controlling a variable air volume valve, characterized in that, The method includes: Collect multi-source sensor data during the operation of the variable air volume valve; The multi-source sensor data is verified to identify abnormal data; The abnormal data is replaced based on historical reliable data to generate repaired sensor data; The repaired sensor data is input into the trained air volume fusion calculation model, and the fused air volume value is output. Obtain the target air volume setpoint, and based on the air volume deviation between the fused air volume value and the target air volume setpoint, generate a valve control command to control the variable air volume valve.

2. The variable air volume valve control method according to claim 1, characterized in that, The process of verifying the collected multi-source sensor data and identifying abnormal data includes: A sliding time window containing a preset number of continuous sampling points is constructed to calculate the standardized value of each detection signal of the multi-source sensor data within the sliding time window; Assign a first weighting coefficient to each of the detection signals, and calculate a weighted anomaly index based on the standardized value of each of the detection signals and the corresponding first weighting coefficient; Map the weighted abnormality index to a health score; If the health score is outside the preset health threshold range, the multi-source sensor data within the sliding time window is determined to be an abnormal data group. The detection signal whose standardized value in the abnormal data group is outside the preset standardized value range is determined to be abnormal data.

3. The variable air volume valve control method according to claim 2, characterized in that, The process of replacing the abnormal data with historical reliable data to generate repaired sensor data includes: Filter multi-source sensor data samples whose health scores fall within the preset health threshold range within a historical time window; A second weighting coefficient is assigned to each of the multi-source sensing data samples that is of the same type as the abnormal data, and the second weighting coefficient decreases exponentially with the increase of time distance. Based on each of the detected signal samples and the corresponding second weighting coefficient, a weighted average operation is performed on the abnormal data to obtain a replacement value; The abnormal data is replaced with the replacement value to generate the repaired sensing data.

4. The variable air volume valve control method according to claim 2, characterized in that, The training of the air volume fusion calculation model includes: The feedforward neural network model was initially trained using an offline dataset to obtain the air volume fusion calculation model; When the health score of the current multi-source sensor data is within the preset health threshold and the corresponding air volume deviation is less than the preset deviation threshold, the current multi-source sensor data and the corresponding fused air volume value are saved as a fine-tuning sample. When the cumulative number of fine-tuned samples reaches the preset number of samples to form a sample set, the air volume fusion calculation model is incrementally updated and trained based on the sample set.

5. The variable air volume valve control method according to claim 1, characterized in that, The generation of valve control commands based on the airflow deviation between the fused airflow value and the target airflow setpoint includes: Calculate the rate of change between the target air volume setpoint and the target air volume setpoint of the previous period; When the rate of change exceeds a preset change threshold within a preset time, an intermediate target air volume sequence is generated based on a preset step size. Based on the current fused air volume value and the intermediate deviation air volume of each intermediate target value in the intermediate target air volume sequence, corresponding intermediate valve control commands are generated to control the variable air volume valve in sequence.

6. The variable air volume valve control method according to claim 5, characterized in that, After generating the corresponding intermediate valve control commands to control the variable air volume valve sequentially, the method further includes: Based on each of the intermediate target values ​​and the corresponding fused air volume values, calculate the tracking error corresponding to each of the intermediate target values; When the tracking error exceeds the first preset error threshold for a preset number of consecutive times, the generation time of the remaining intermediate target values ​​in the intermediate target air volume sequence is extended. When the tracking error is negative and the absolute value of the tracking error is greater than the second preset error threshold, the next intermediate target value of the intermediate target air volume sequence is reduced.

7. The variable air volume valve control method according to claim 1, characterized in that, After obtaining the target airflow setpoint and generating a valve control command based on the airflow deviation between the fused airflow value and the target airflow setpoint, the control method further includes: Calculate the absolute difference between the current valve control command and the valve control command of the previous cycle; When the absolute difference is within a preset range, the current valve control command is output to control the variable air volume valve; When the absolute difference is outside the preset range of change, the instruction of the previous cycle remains unchanged.

8. A variable air volume valve control system, characterized in that, include: The data acquisition module is used to collect multi-source sensor data during the operation of the variable air volume valve; The verification module is used to verify the multi-source sensor data in order to identify abnormal data. The replacement module is used to replace the abnormal data based on historical reliable data and generate repaired sensor data; The input module is used to input the repaired sensor data into the trained air volume fusion calculation model and output the fused air volume value. The generation module is used to obtain the target air volume setpoint and, based on the air volume deviation between the fused air volume value and the target air volume setpoint, generate valve control commands to control the variable air volume valve.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the variable air volume valve control method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed on a processor, implements the steps of the variable air volume valve control method according to any one of claims 1-7.

Images