Method and system for self-adaptive adjustment of air curtain machine air speed based on environmental temperature and humidity

Abstract

This application provides a method and system for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, belonging to the field of equipment control. The method performs differential operations and vector construction on the external temperature and humidity change sequences at building entrances to obtain an environmental gradient feature vector; it then classifies risks based on the environmental gradient feature vector to obtain the risk level of protected windows; based on the risk level of protected windows and real-time speed feedback data of the air curtain machine motor, it calculates the target wind speed adjustment amount and generates a motor control signal to drive the motor to perform the wind speed adjustment action; it calculates barrier strength data based on the speed feedback data, calculates the time difference between the barrier strength data and the predicted time of external air intrusion, and generates a continuous operation control command based on the time difference. This application, by sensing changes in the environmental gradient and predicting the risk of protected windows, completes wind speed adjustment before substantial changes in temperature and humidity occur, eliminating the protected window period caused by post-event feedback mechanisms.

Description

Technical Field

[0001] This application relates to the field of equipment control technology, specifically to a method and system for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity. Background Technology

[0002] In the field of building entrance environmental control, air curtains are important devices for maintaining a stable indoor environment by forming a high-speed airflow barrier to isolate indoor and outdoor air exchange. However, the control strategies of existing air curtains generally rely on a post-event feedback mechanism, that is, the controller only issues a command to adjust the air speed after the sensor detects a significant change in the ambient temperature and humidity.

[0003] This passive response mode has an inherent time lag problem. After changes in ambient temperature and humidity, the wind speed adjustment often lags behind the actual environmental changes, resulting in a protective gap before the adjustment is completed, thus affecting the indoor-outdoor air isolation effect.

[0004] The information disclosed in the background section is only for enhancing the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] In view of this, this application provides a method and system for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, which can solve the technical problem that the existing air curtain machine control strategy relies on post-event feedback, resulting in wind speed adjustment lagging behind environmental changes and creating a protection gap.

[0006] In a first aspect, embodiments of this application provide a method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity, the method comprising: By performing difference operations and vector construction on the air temperature and humidity change sequences outside the building entrance, an environmental gradient feature vector is obtained. Risk classification is performed based on the environmental gradient feature vector to obtain the risk level of the protective open window at the building entrance; The target air curtain speed adjustment amount is calculated based on the risk level of the protective window at the building entrance and the real-time speed feedback data of the air curtain machine motor. The air curtain machine motor control signal is generated based on the target air curtain speed adjustment amount to drive the air curtain machine motor to perform the speed adjustment action. The barrier strength data is calculated based on the rotational speed feedback data during the wind speed adjustment process, and the time difference between the barrier strength data and the predicted time of external air intrusion is calculated. Based on the time difference, a continuous operation control command for the air curtain machine is generated.

[0007] Secondly, embodiments of this application provide a system for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity. The system includes: The gradient extraction module is used to perform difference operations and vector construction on the air temperature change sequence and humidity change sequence outside the building entrance to obtain the environmental gradient feature vector. The risk classification module is used to classify risks based on the environmental gradient feature vector to obtain the risk level of the protective windows at the building entrances and exits. The wind speed adjustment module is used to calculate the target wind speed adjustment amount of the air curtain based on the risk level of the protective window at the building entrance and the real-time speed feedback data of the air curtain motor, and generate a control signal for the air curtain motor based on the target wind speed adjustment amount to drive the air curtain motor to perform wind speed adjustment action. The barrier monitoring module is used to calculate barrier strength data based on the rotational speed feedback data during the wind speed adjustment process, and to calculate the time difference between the barrier strength data and the predicted time of external air intrusion, and to generate a continuous operation control command for the air curtain machine based on the time difference.

[0008] This application provides a method and system for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity. By performing differential calculations on the temperature and humidity change sequences of the external air at building entrances and exits to extract environmental gradient feature vectors, the quantitative expression of environmental change trends is used as input for risk classification. This allows the air curtain machine control system to predict the risk level of protected windows before substantial fluctuations in temperature and humidity occur, fundamentally breaking the time-series dependency between environmental changes and equipment response in traditional post-event feedback mechanisms. Based on this, the target air curtain wind speed adjustment amount is calculated by combining real-time speed feedback data from the air curtain machine motor, and a motor control signal is generated. This shifts the triggering time of the wind speed adjustment action to the stage of environmental gradient anomalies, rather than waiting for a significant shift in temperature and humidity before responding. Furthermore, by continuously calculating barrier strength data from the speed feedback data during the wind speed adjustment process and combining it with the predicted time of external air intrusion for time difference calculation, a closed-loop monitoring mechanism is formed. This generates continuous operation control commands in a timely manner before the barrier strength is insufficient to resist external air intrusion, ensuring the continuous effectiveness of the indoor and outdoor air isolation barrier. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1This is a flowchart illustrating a method for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, provided in an exemplary embodiment of this application.

[0011] Figure 2 This is a flowchart illustrating a method for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, provided in another exemplary embodiment of this application.

[0012] Figure 3 This is a flowchart illustrating a method for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, provided in another exemplary embodiment of this application.

[0013] Figure 4 This is a flowchart illustrating a method for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, provided in another exemplary embodiment of this application.

[0014] Figure 5 This is a flowchart illustrating a method for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, provided in another exemplary embodiment of this application. Detailed Implementation

[0015] The technical solutions of 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, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of a described feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. It should also be understood that, as used in this specification and the appended claims, the term "and / or" refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0017] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0018] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0019] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0020] To illustrate the technical solution of this application, specific embodiments are described below.

[0021] To address the problems mentioned above, this application provides a method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity. Figure 1 The method shown is for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity. This method may include the following steps: Step S10: Perform differential operations and vector construction on the air temperature change sequence and humidity change sequence outside the building entrance to obtain the environmental gradient feature vector; Step S20: Perform risk classification based on the environmental gradient feature vector to obtain the risk level of the protective open window at the building entrance; Step S30: Calculate the target air curtain speed adjustment amount based on the risk level of the protective window at the building entrance and the real-time speed feedback data of the air curtain machine motor, and generate an air curtain machine motor control signal based on the target air curtain speed adjustment amount to drive the air curtain machine motor to perform the speed adjustment action. Step S40: Calculate the barrier strength data based on the rotational speed feedback data during the wind speed adjustment process, calculate the time difference between the barrier strength data and the predicted time of external air intrusion, and generate a continuous operation control command for the air curtain machine based on the time difference.

[0022] In this embodiment, building entrances and exits are the locations with the most frequent indoor-outdoor air exchange. Changes in external air temperature and humidity directly affect whether the air curtain can effectively isolate indoor and outdoor air. However, the absolute values ​​of temperature and humidity are not sufficient to reflect the urgency of environmental changes; two moments with similar absolute values ​​may have drastically different environmental change trends. Relying solely on absolute temperature and humidity values ​​to trigger adjustments will result in the system failing to respond promptly in the initial stages of sudden environmental changes. Therefore, this embodiment calculates the temperature and humidity differences between adjacent time points by performing differential operations on the temperature and humidity change sequences, using the rate characteristics of environmental change rather than absolute values ​​as the input for subsequent risk assessment. Differential operations can directly reflect the magnitude of temperature and humidity changes per unit time. The faster the temperature drops, the larger the temperature drop gradient characteristic; the faster the humidity rises, the larger the humidity rise gradient characteristic. Together, they constitute an environmental gradient feature vector that can quantify the intensity of sudden changes in the external environment, providing a reliable feature basis for subsequent risk level prediction.

[0023] Based on the above description, this embodiment inputs the extracted environmental gradient feature vector into the risk classification model to determine the risk level of protective gaps. The risk classification model is trained using historical environmental gradient data and corresponding risk-labeled samples. By learning the distribution characteristics of the environmental gradient feature vector, it establishes a mapping relationship between the gradient change magnitude and the risk level of protective gaps. The risk level classification reflects the likelihood of protective gaps appearing in the air curtain barrier under the current environmental gradient change trend. A higher level indicates a more urgent risk of cold, humid external air intruding into the room, requiring faster wind speed adjustment. By using the risk level, rather than absolute temperature and humidity values, as the trigger for adjustment, the system can complete the protective response before a substantial shift in the environment occurs.

[0024] Based on this, this embodiment calculates the target air curtain speed adjustment amount required under the current operating condition by combining the risk level assessment result with the real-time speed feedback data of the air curtain machine motor. The real-time speed feedback data of the motor is introduced in the calculation because there is an objective mechanical response delay between the motor receiving the control command and the actual speed reaching the target value. If this delay is not considered and the target air speed is directly mapped to the risk level, the generated control signal will deviate from the motor's current actual operating state, leading to inaccurate adjustment results. By combining the real-time speed feedback data to compensate for the response delay, the real-time state of the motor is incorporated into the feedback loop of the process control. The resulting target air curtain speed adjustment amount accurately reflects the target speed increment that the motor needs to achieve under the current state. Furthermore, a corresponding motor control signal is generated based on the target air curtain speed adjustment amount to drive the motor to perform the air speed adjustment action.

[0025] For example, at the entrance of a shopping mall, the outside temperature drops continuously in a short period of time. The temperature drop gradient feature obtained by differential calculation continues to increase. Based on this, the risk classification model determines that the current risk level of the protective window has reached a high level. The system then combines the current real-time speed of the motor to trigger the calculation of the target wind speed, generates a motor control signal to drive the motor to accelerate, and completes the reinforcement of the wind curtain barrier before the indoor temperature drops significantly. This avoids the situation in the traditional post-event response mode where cold air has already invaded the room before adjustment begins.

[0026] During the wind speed adjustment process, this embodiment continuously collects motor speed feedback data. By calculating the error between the speed feedback data and the target speed, the stability of the motor's current acceleration process is continuously evaluated through process control. Then, combined with the motor's operating status data, the strength data of the air barrier currently formed by the air curtain machine at the building entrance is calculated. The barrier strength data reflects the actual blocking ability of the air curtain airflow on the exchange of indoor and outdoor air at the current speed, and is a direct basis for judging whether the current wind speed is sufficient to cope with external air intrusion. After obtaining the barrier strength data, this embodiment further calculates the time difference based on the predicted time of external air intrusion. If the calculated time difference meets the preset early protection response conditions, it indicates that the current barrier strength will be at risk of failure before external air intrusion. The system then generates a continuous operation control command for the air curtain machine, actively maintaining or further increasing the air curtain barrier strength through process control to ensure that there are no gaps in protection.

[0027] For example, at the entrance of a building, the system calculates based on the current barrier strength data that the existing barrier is expected to become ineffective in 900 seconds, while the predicted intrusion time of the strong external wind is 1200 seconds from the current time. The difference between the two is 300 seconds, which meets the preset early protection response conditions. Based on this, the system generates a continuous operation control command, increases the operating frequency of the air curtain machine and locks it in continuous operation mode, and completes the barrier reinforcement before the external airflow actually arrives, forming a complete closed-loop protection.

[0028] In one embodiment, such as Figure 2 As shown, step S10 specifically includes the following steps: S11: Calculate the difference between temperature values ​​at adjacent time points in the air temperature change sequence outside the building entrance and extract the result with the largest absolute value of the difference as the temperature decrease gradient feature. S12: Calculate the difference between humidity values ​​at adjacent time points in the humidity change sequence outside the building entrance and extract the maximum positive difference as the humidity rise gradient feature. S13: Construct a vector based on the temperature decrease gradient feature and the humidity increase gradient feature to obtain the environmental gradient feature vector.

[0029] In this embodiment, temperature and humidity sensors are deployed outside the building entrance to continuously collect external air temperature and humidity values ​​at a fixed sampling frequency, forming temperature and humidity change sequences over time. The absolute values ​​of temperature and humidity at a single sampling moment cannot reflect the rate of environmental change. However, when the door is suddenly opened, the instantaneous influx of cold, humid outdoor air manifests as a rapid drop in temperature and a rapid rise in humidity within a short period. This rate of change is crucial information for assessing the urgency of the air intrusion threat. Therefore, this embodiment performs first-order forward difference operations on both the temperature and humidity change sequences, calculating the numerical difference between every two adjacent time points in the sequence to quantify the magnitude of temperature and humidity changes within a unit sampling interval.

[0030] Based on the above explanation, for a temperature change sequence, this embodiment calculates the temperature difference between each pair of adjacent sampling points in the sequence, obtaining a set of difference sequences. Since the temperature decreases when cold air rushes in, the difference is negative; the larger the absolute value, the more drastic the temperature drop within that sampling interval. This embodiment extracts the difference with the largest absolute value from the difference sequence as the temperature drop gradient feature. This feature directly quantifies the instantaneous rate of the most drastic temperature drop within the observation window, enabling the capture of the most intense temperature shock moments during the cold air intrusion process.

[0031] For example, the system collected a temperature change sequence of 5.2, 4.8, 4.1, 3.5, and 3.0 degrees Celsius at a sampling frequency of once per second over the past five seconds. The difference between adjacent sampling points was calculated to obtain a difference sequence of -0.4, -0.7, -0.6, and -0.5. The difference with the largest absolute value was 0.7. The system extracted this value as the temperature decrease gradient feature within the current sampling window.

[0032] For the humidity change sequence, this embodiment also calculates the humidity difference for each pair of adjacent sampling points to obtain a humidity difference sequence. Unlike temperature, humidity increases when cold, humid air from outside intrudes, and the difference is positive; a larger positive difference indicates a faster increase in humidity. This embodiment extracts the maximum positive difference from the difference sequence as a feature of the humidity rise gradient, reflecting the instantaneous rate of the most dramatic increase in humidity within the observation window. The reason for extracting the maximum positive difference instead of the average difference is that the air intrusion process often concentrates within a short time window after the door is opened, and the average value would smooth out the characteristics of this peak moment, while the maximum positive difference can accurately capture the moment when the humidity impact is most intense.

[0033] For example, the humidity change sequence corresponding to the above temperature sequence is 60%, 65%, 72%, 78%, 85%. The difference sequence obtained by subtracting these values ​​is 5, 7, 6, 7. The maximum positive difference is 7, and the system extracts this value as the humidity gradient feature.

[0034] After obtaining the temperature decrease gradient features and humidity increase gradient features, this embodiment combines them to construct an environmental gradient feature vector. This vector uses the temperature decrease gradient feature as the first component and the humidity increase gradient feature as the second component, forming a two-dimensional feature representation, which serves as the input to the risk classification model in subsequent step S20. The reason for combining the two gradient features into a vector instead of using them independently is that temperature decrease and humidity increase are physically two accompanying manifestations of the event of external cold and humid air intrusion. Their joint distribution can more comprehensively reflect the overall intensity of the intrusion event. Relying solely on one dimension may lead to misjudgment due to sensor noise or local environmental disturbances. For example, if only the temperature decrease gradient is large while the humidity remains relatively constant, it may only be due to the influence of local ventilation equipment rather than cold and humid air intrusion. When both gradient features are simultaneously large, the basis for judging the intrusion event is more sufficient.

[0035] In one embodiment, after constructing the environmental gradient feature vector based on the temperature decrease gradient feature and the humidity increase gradient feature, the method further includes: Anomaly detection is performed on the environmental gradient feature vector to obtain an abnormal environmental gradient value vector, and the corresponding abnormal timestamp is extracted based on the abnormal environmental gradient value vector; Based on the abnormal timestamp, extract the corresponding abnormal speed value sequence from the real-time speed feedback data of the air curtain machine motor; Spectral analysis is performed on the abnormal speed value sequence to obtain the speed amplitude sequence. The maximum amplitude value in the speed amplitude sequence is compared with a preset amplitude threshold. If the maximum amplitude value exceeds the preset amplitude threshold, it is determined that the air curtain machine motor is in an overload operation state, and an air curtain machine motor shutdown control command is generated. The air curtain machine motor shutdown control command is used to trigger the air curtain machine motor to stop.

[0036] In this embodiment, when the air curtain machine responds to sudden changes in the external environment, the motor needs to rapidly increase its speed within a short period to strengthen the air curtain barrier. During this process, the load on the motor will be significantly higher than under normal operating conditions. If the motor continues to operate under conditions exceeding its rated load for an extended period, the winding temperature will continue to rise, potentially leading to insulation damage or even motor burnout. Therefore, after constructing the environmental gradient feature vector, it is necessary to perform overload detection on the motor's current operating status to promptly trigger shutdown protection when an overload risk occurs, thus preventing equipment damage.

[0037] Based on the above description, this embodiment first performs anomaly detection on the environmental gradient feature vector, identifying abnormal gradient values ​​that deviate from the normal distribution range. This embodiment uses the isolated forest algorithm to perform anomaly detection. The basic principle of the isolated forest algorithm is that abnormal samples are sparsely distributed in the feature space, making them easier to isolate by random segmentation trees compared to normal samples, requiring fewer segmentation steps. Specifically, using historical environmental gradient feature vector samples as the training set, an isolated forest composed of multiple isolated trees is constructed. Each isolated tree recursively segments samples by randomly selecting segmentation features and segmentation values ​​in the feature space until each sample is isolated individually. For the currently input environmental gradient feature vector, the average path length required for it to be isolated in each isolated tree is calculated. The shorter the path length, the easier the sample is to isolate, and the higher the degree of abnormality. The average path length is converted into an anomaly score, and samples with anomaly scores exceeding a preset anomaly score threshold are determined to be abnormal environmental gradient value vectors. The preset abnormal score threshold is determined based on the statistical distribution of abnormal scores of a large number of historical gradient samples collected under normal operating conditions. The 95th percentile of the abnormal score distribution of normal samples is taken as the threshold. Samples below the threshold are classified as normal, and samples above the threshold are judged as abnormal, thus balancing the false negative rate and the false positive rate.

[0038] After identifying the abnormal environmental gradient value vector, this embodiment extracts the corresponding collection time of each abnormal sample in the time series, obtaining a set of abnormal timestamps. The abnormal timestamps mark the specific times when abnormal fluctuations in the environmental gradient occur, which are also the time points when the motor may be subjected to abnormal loads. Using the abnormal timestamps as indexes, the speed data within the corresponding time period is extracted from the real-time speed feedback data of the air curtain machine motor, obtaining an abnormal speed value sequence. The extraction range extends forward and backward by several sampling points centered on the abnormal timestamp to completely cover the entire process of the motor responding to abnormal gradient events.

[0039] For example, if the system detects that the abnormal score of the environmental gradient feature vector is 0.91 at a certain moment, which exceeds the preset abnormal score threshold of 0.88, it determines that the moment is an abnormal timestamp, and then extracts the speed value sequence within 10 seconds before and after the moment from the motor speed feedback data as the abnormal speed value sequence.

[0040] After obtaining the abnormal speed value sequence, this embodiment performs spectral analysis to convert the time-domain speed signal into a frequency-domain representation, obtaining a speed amplitude sequence. Spectral analysis can separate and display the amplitudes of different frequency components in the speed signal. Under overload conditions, the speed fluctuation of the motor will produce significant amplitude anomalies in specific frequency components. These anomalies are often submerged in the overall speed trend in the time-domain signal and are difficult to observe directly, but can be clearly identified through spectral analysis. This embodiment extracts the maximum amplitude value from the speed amplitude sequence and compares it with a preset amplitude threshold. The preset amplitude threshold is determined based on the statistical distribution of the maximum amplitude value of the speed amplitude sequence when the motor is running within its rated load range. The average of the maximum amplitude values ​​under normal operating conditions plus twice the standard deviation is taken as the threshold. Exceeding this threshold indicates that the speed fluctuation has exceeded the normal load range. If the maximum amplitude value exceeds the preset amplitude threshold, the air curtain machine motor is determined to be in an overload operating state. The system generates an air curtain machine motor shutdown control command, which is sent to the motor driver via the control bus to trigger the motor shutdown, thereby avoiding equipment damage caused by continuous overload.

[0041] In one embodiment, such as Figure 3 As shown, step S20 specifically includes the following steps: S21: Normalize the environmental gradient feature vector to obtain a normalized environmental gradient feature vector, and calculate the split gain value of the decision tree branch node based on the normalized environmental gradient feature vector. S22: Construct multiple decision trees based on the split gain values ​​of the decision tree branch nodes, and statistically analyze the sample distribution density of the leaf nodes of the multiple decision trees to obtain the sample distribution density values ​​of the leaf nodes; S23: Based on the sample distribution density values ​​of the leaf nodes, classify the prediction results of the multiple decision trees to obtain the risk level of the protective void at the building entrance.

[0042] In this embodiment, the environmental gradient feature vector consists of two components: a temperature decrease gradient feature and a humidity increase gradient feature. These two components have different dimensions and numerical ranges. The temperature gradient is measured in degrees Celsius per sampling interval, while the humidity gradient is measured as a percentage per sampling interval. If the original values ​​are directly input into the classification model, the component with the larger numerical range will dominate the node splitting calculation, suppressing the contribution of the other component and leading to a bias in the classification result. Therefore, this embodiment first normalizes the environmental gradient feature vector before performing risk classification to eliminate the influence of the difference in dimensions.

[0043] Specifically, this embodiment employs a maximum-minimum normalization method to process the temperature decrease gradient feature and the humidity increase gradient feature separately. The maximum and minimum values ​​of each component in historical samples are used as the normalization boundaries to map the original feature values ​​to a unified range of 0 to 1. The normalization formula is: the normalized value equals the original value minus the historical minimum value, then divided by the difference between the historical maximum and the historical minimum value. The historical maximum and minimum values ​​are determined statistically based on a large number of normal and abnormal operating condition samples collected over a period of time. The samples cover building entrance and exit environmental data under different seasons and weather conditions to ensure that the normalization boundaries are sufficiently representative.

[0044] For example, the historical temperature gradient has a maximum value of 2.0 and a minimum value of 0, and the current temperature decreasing gradient feature is 0.7, which is normalized to 0.35; the historical humidity gradient has a maximum value of 10% and a minimum value of 0, and the current humidity increasing gradient feature is 7%, which is normalized to 0.70. Thus, the normalized environmental gradient feature vector is (0.35, 0.70).

[0045] After normalization, this embodiment uses the normalized environmental gradient feature vector as input to calculate the split gain value of the decision tree branch node. The split gain value measures the improvement in sample purity after dividing the current node's samples into two subsets using a certain threshold of a certain feature as the split point. The specific calculation process is as follows: First, calculate the information entropy of the sample set contained in the current node. The formula for calculating information entropy is the sum of the negative logarithms of the proportion of samples of each category, reflecting the degree of disorder in the distribution of sample categories at the current node. Then, using the normalized feature value as the candidate split threshold, divide the sample set into two subsets, calculate the information entropy of each subset, and sum them using the proportion of the number of samples in each subset to the total number of samples in the parent node as weights to obtain the conditional entropy. Finally, subtract the conditional entropy from the information entropy of the current node to obtain the split gain value corresponding to the split point. The larger the split gain value, the stronger the ability of the split point to distinguish sample categories.

[0046] For example, the current node contains 100 historical samples, including 30 high-risk samples and 70 low-risk samples, with an information entropy of 0.88. Using a temperature gradient normalization value of 0.3 as the splitting threshold, the samples are divided into two groups, resulting in a conditional entropy of 0.55 and a splitting gain of 0.33. Based on the principle of maximum information gain, the system selects the threshold with the largest splitting gain from all candidate splitting thresholds to perform node splitting, recursively until the splitting gain value falls below a preset stopping threshold. The preset stopping threshold is determined based on the trade-off between the model's classification accuracy on the validation set and the tree depth. It is set to the gain value at which the validation set accuracy no longer significantly improves; in this embodiment, it is 0.01. Splitting stops when the value falls below this to avoid overfitting.

[0047] Based on the above description, this embodiment constructs multiple decision trees in parallel using the aforementioned node splitting rules. During the construction of each decision tree, a subset of samples is randomly extracted with replacement from the training sample set. At each node split, a subset of features is randomly selected from all features as candidate splitting features. By introducing dual randomness in both samples and features, the differences between decision trees are maintained, reducing the overall variance of the model. After construction, all leaf nodes of each decision tree are traversed, the number of training samples falling into each leaf node is counted, and the volume of the feature interval covered by that leaf node is calculated. The sample distribution density value of the leaf node is obtained by dividing the number of samples by the feature interval volume. The sample distribution density value reflects the degree of clustering of samples within the feature interval corresponding to the leaf node. A higher density indicates a more concentrated set of samples falling into that interval, and the more reliable the classification result corresponding to that leaf node.

[0048] After obtaining the sample distribution density values ​​of each leaf node, this embodiment feeds the current input normalized environmental gradient feature vector into all decision trees. Each decision tree routes the input vector to the corresponding leaf node according to the splitting rules, and outputs the risk level category represented by that leaf node. The outputs of all decision trees are collected, and the number of votes for each risk level category is weighted by the sample distribution density values ​​of the corresponding leaf nodes of each decision tree. The risk level category with the highest number of votes is taken as the final risk level of the building entrance / exit protective window. The sample distribution density value is introduced as the voting weight, rather than simple majority voting, because the prediction results falling into high-density leaf nodes come from the training sample region where the feature distribution is more concentrated, resulting in higher prediction reliability. Giving it higher weight allows the final voting result to more accurately reflect the risk distribution pattern of the feature interval where the current input vector is located.

[0049] In one embodiment, the step of normalizing the environmental gradient feature vector to obtain a normalized environmental gradient feature vector, and calculating the split gain value of the decision tree branch node based on the normalized environmental gradient feature vector, includes: The maximum and minimum gradient values ​​are determined based on historical environmental gradient data. The maximum and minimum normalization of each feature value in the environmental gradient feature vector are calculated to obtain the normalized environmental gradient feature vector. The initial information entropy is calculated based on the sample category distribution contained in the current node in the normalized environmental gradient feature vector. The samples are divided using the feature threshold as the split point. The conditional entropy is calculated based on the sample category distribution after the division. The difference between the initial information entropy and the conditional entropy is calculated to obtain the split gain value of the decision tree branch node. The split gain value of the decision tree branch node is compared according to the principle of maximum information gain. If the split gain value is lower than the preset stopping threshold, the node splitting is stopped.

[0050] In this embodiment, the two components of the environmental gradient feature vector—the temperature decreasing gradient feature and the humidity increasing gradient feature—differ in magnitude. The temperature gradient is in degrees Celsius, while the humidity gradient is in percentage. If directly fed into the decision tree for node splitting calculation, the component with the larger value will naturally have an advantage in gain calculation, causing the splitting result to over-rely on one component while neglecting the discriminative ability of the other. Therefore, this embodiment first performs max-min normalization on the two components before calculating the splitting gain.

[0051] Specifically, the maximum and minimum values ​​of each component in the historical environmental gradient data are used as the normalization boundary. The current feature value is mapped using the following formula: the normalized value equals the current feature value minus the historical minimum value, divided by the difference between the historical maximum and the historical minimum value. The historical maximum and minimum values ​​are statistically obtained from a large number of normal and abnormal operating condition samples collected in the past. The samples need to cover different seasons and weather conditions to ensure that the normalization boundary can represent the complete distribution range of feature values ​​in actual operation. After normalization, both components are mapped to the interval between 0 and 1, eliminating dimensional differences, resulting in the normalized environmental gradient feature vector.

[0052] For example, the historical maximum temperature gradient is 2.0 and the minimum is 0, the current temperature gradient feature is 0.5, and after normalization, it becomes 0.25; the historical maximum humidity gradient is 5% and the minimum is 0, the current humidity gradient feature is 2%, and after normalization, it becomes 0.40. Thus, the normalized environmental gradient feature vector is (0.25, 0.40).

[0053] After obtaining the normalized environmental gradient feature vector, this embodiment uses this vector as input to calculate the split gain value of the decision tree branch node. The calculation process of the split gain value is as follows: First, count the number of samples of each risk level category in the training samples contained in the current node. Suppose that the current node has a total of N samples, and the number of samples of the k-th risk level is Nk. Then the proportion of each category is Nk divided by N. The initial information entropy H of the current node is calculated by the following formula: H is equal to the sum of the proportion of each category multiplied by the negative logarithm of that proportion to the base 2. That is, summing over all categories k, and each term being a negative Nk divided by N multiplied by the base 2 logarithm of Nk divided by N. The initial information entropy reflects the degree of disorder in the distribution of sample categories in the current node. The higher the entropy value, the higher the degree of mixing of samples of each category and the lower the purity of the node.

[0054] Furthermore, using the value of a certain component in the normalized feature vector as the candidate splitting threshold t, the N samples of the current node are divided into two subsets: samples with feature values ​​less than t are assigned to the left subset, and samples with feature values ​​greater than or equal to t are assigned to the right subset. The number of samples in the left subset is NL, and the number of samples in the right subset is NR. The information entropy HL and HR of the left and right subsets are calculated separately, and then weighted and summed using the proportion of the number of samples in each subset to the total number of samples in the parent node as weights, to obtain the conditional entropy: the conditional entropy equals NL divided by N multiplied by HL plus NR divided by N multiplied by HR. The final splitting gain value equals the initial information entropy H minus the conditional entropy.

[0055] For example, the current node contains 100 samples, including 30 high-risk samples and 70 low-risk samples, with an initial information entropy of 0.88. Using a temperature gradient normalization value of 0.3 as the splitting threshold, after splitting, the left subset has 40 samples and the right subset has 60 samples, resulting in a conditional entropy of 0.55 and a splitting gain of 0.33.

[0056] After traversing all candidate split thresholds, the system selects the threshold with the largest split gain value to perform the actual split, i.e., the maximum information gain principle. After splitting, the above calculation process is recursively performed on each child node until the maximum split gain value of a node is lower than the preset stopping threshold, at which point the splitting stops. The preset stopping threshold is set based on the following: statistically analyzing the relationship between the model classification accuracy and the decision tree depth corresponding to different stopping thresholds on the validation set. When the stopping threshold continues to decrease but the validation set accuracy no longer improves or even begins to decline, the gain value corresponding to that inflection point is determined as the preset stopping threshold. A stopping threshold that is too low will cause the tree to continue growing until the samples are completely pure, resulting in overfitting to the training data and a decrease in generalization ability on new samples; a stopping threshold that is too high will cause the tree to stop growing prematurely, resulting in overly coarse classification boundaries and missing important splitting information. In this embodiment, the preset stopping threshold is set to 0.01. At this threshold, the validation set classification accuracy reaches a stable level, and further decreasing the threshold no longer brings a substantial improvement in accuracy.

[0057] In one embodiment, such as Figure 4 As shown, step S30 specifically includes the following steps: S31: Based on the risk level of the protective window and the real-time speed feedback data of the air curtain motor, the response delay of the air curtain motor is compensated and calculated to obtain the target air curtain speed adjustment amount. S32: Calculate the duty cycle based on the target air curtain wind speed adjustment amount to obtain the pulse width modulation duty cycle parameter, and generate the air curtain machine motor control signal based on the pulse width modulation duty cycle parameter; S33: Detect the safe operating status of the air curtain machine motor, obtain the safety detection result, correct the air curtain machine motor control signal according to the safety detection result, obtain the corrected air curtain machine motor control signal, and drive the air curtain machine motor to perform wind speed adjustment action according to the corrected air curtain machine motor control signal.

[0058] In this embodiment, the risk level of the protective window reflects the urgency of external air intrusion under the current environmental gradient change trend. Different risk levels correspond to different target wind speed requirements; the higher the risk level, the greater the required strength of the air curtain barrier, and the higher the corresponding target wind speed. However, the risk level itself only describes the intensity of the protection requirement and is not directly equivalent to the speed increment that the motor needs to output. There is an objective mechanical response delay in the process of the motor accelerating from the current speed to the target speed. This delay is closely related to the motor's current actual speed: the lower the current speed, the longer it takes for the motor to accelerate to the target speed, and the greater the response delay; the closer the current speed is to the target speed, the smaller the required compensation amount. If this delay is not considered and the target wind speed is directly mapped to the risk level, there will be a deviation between the generated control command and the actual operating state of the motor, resulting in inaccurate wind speed adjustment results. Therefore, in this embodiment, when calculating the target air curtain wind speed adjustment amount, the risk level of the protective window is combined with the real-time speed feedback data of the air curtain machine motor to compensate for the response delay.

[0059] Specifically, this embodiment first extracts the upper and lower limits of the corresponding wind speed range from a preset risk level threshold range parameter table based on the risk level of the protected window. This parameter table is calibrated according to the minimum wind curtain barrier strength actually required under different risk levels. The wind speed range corresponding to each level is determined by experimentally measuring the blocking effect of the wind curtain barrier on outdoor cold and humid air at different wind speeds. After confirming that the classification confidence probability value corresponding to the current risk level exceeds the preset classification confidence threshold, subsequent compensation calculations are triggered. The preset classification confidence threshold is determined statistically based on the confidence distribution of the risk classification model on the validation set. The upper bound of the confidence distribution of misclassified samples is taken as the threshold. Classification results below this threshold are unreliable and should not trigger wind speed adjustment actions to avoid unnecessary motor acceleration due to classification errors. In this embodiment, the preset classification confidence threshold is 0.85.

[0060] Furthermore, in this embodiment, the current speed value is extracted from the real-time speed feedback data of the air curtain machine motor. The current speed value is divided by the rated speed of the motor to obtain the base speed ratio, which is then multiplied by a preset attenuation constant to obtain the response delay compensation coefficient of the air curtain machine motor. The preset attenuation constant reflects the degree of attenuation of the motor's response capability under the current load conditions. Its value is determined based on the fitting of the measured acceleration response curve of the motor in different speed ranges. In this embodiment, it is set to 0.9. Taking a current speed of 1200 rpm and a rated speed of 1500 rpm as an example, the base speed ratio is 0.8. Multiplying it by the attenuation constant of 0.9 yields a response delay compensation coefficient of 0.72.

[0061] After obtaining the response delay compensation coefficient, this embodiment performs a weighted summation with the normalized real-time speed feedback data. The response delay compensation coefficient is assigned a weight of 0.4, and the normalized speed feedback data is assigned a weight of 0.6, with the sum of the two weights being 1. The weighted summation result is the target air curtain speed adjustment amount. The basis for the weight allocation is: in actual operation, the current speed state of the motor contributes more directly to the final adjustment amount, so the speed feedback data is given a higher weight; the compensation coefficient, as a correction to the response delay, contributes relatively less and is given a lower weight. For example, multiplying the response delay compensation coefficient 0.72 by 0.4 and adding the normalized speed feedback data 0.8 by 0.6 yields the target air curtain speed adjustment amount of 0.768.

[0062] Based on the above description, after obtaining the target air curtain wind speed adjustment amount, this embodiment compares it with a preset rated wind speed threshold. The preset rated wind speed threshold corresponds to the upper limit of the rated operating wind speed of the air curtain machine under normal air curtain barrier maintenance conditions. This threshold is calibrated by the air curtain machine manufacturer based on hardware parameters such as the equipment impeller design parameters, motor rated power, and outlet cross-sectional area, and is written into the control motherboard memory at the factory. Different models of air curtain machines have different preset rated wind speed thresholds, which can be configured according to the actual deployed equipment model. The preset rated wind speed threshold corresponds to the rated operating wind speed of the motor when normally maintaining the air curtain barrier. When the target adjustment amount exceeds this threshold, it indicates that the current operating condition requires the motor to accelerate further on top of the rated wind speed, triggering a high-load acceleration response. In this embodiment, a linear mapping algorithm is used to convert the target wind curtain speed adjustment amount into a pulse width modulation duty cycle parameter. The mapping relationship is as follows: the duty cycle is equal to the base duty cycle plus the ratio of the target adjustment amount to the maximum adjustment amount multiplied by the remaining duty cycle range. The base duty cycle corresponds to the initial value of the duty cycle at the rated speed of the motor, the maximum adjustment amount corresponds to the maximum speed increment allowed by the motor, and the remaining duty cycle range is 100% minus the base duty cycle.

[0063] For example, if the target air curtain speed adjustment is 2.5 meters per second, the maximum adjustment is 5.0 meters per second, and the basic duty cycle is 40%, then the duty cycle is equal to 40% plus 2.5 divided by 5.0 multiplied by 60%, resulting in a pulse width modulation duty cycle parameter of 70%. At a switching frequency of 10 kHz, the high-level duration is 70 microseconds and the low-level duration is 30 microseconds within a single 100-microsecond control cycle. Based on this, a continuous sequence of air curtain motor control signal pulse waveforms is generated, which drives the motor speed increase through the inverter bridge circuit.

[0064] After generating the motor control signal, this embodiment detects the motor's safe operating status to prevent damage during acceleration due to excessive temperature or current exceeding limits. Specifically, a platinum resistance temperature sensor collects the motor winding temperature monitoring value, and a Hall current sensor collects the motor phase current. A sliding window algorithm is used to calculate the current peak value to obtain the current peak limit protection value. The preset safe operating temperature is determined based on the heat resistance grade of the motor winding insulation material; exceeding this temperature will accelerate insulation aging. In this embodiment, 90 degrees Celsius is used. The preset current protection threshold is determined based on the motor's rated current and short-time overload capacity; in this embodiment, 15 amperes is used.

[0065] If the winding temperature monitoring value is lower than the preset safe operating temperature and the current peak limit protection value does not exceed the preset current protection threshold, the safety test result is considered passed, triggering the torque curve correction of the motor control signal. Specifically, the current torque output curve of the motor is fitted with a quadratic polynomial using the least squares method to calculate the deviation between the actual torque output and the ideal torque output, obtaining the torque fluctuation compensation coefficient. This compensation coefficient is multiplied by the current pulse width modulation duty cycle to obtain the corrected duty cycle. Based on this, the control signal pulse waveform sequence is reconstructed to obtain the corrected air curtain motor control signal. For example, if the initial duty cycle is 40% and the torque fluctuation compensation coefficient is 1.05, the corrected duty cycle is 42%. Based on this, the corrected pulse waveform sequence is generated and output to the drive circuit, driving the motor to smoothly accelerate from the current speed to the target speed, executing the wind speed adjustment action. If the safety test result is failed, the acceleration command is paused, and the adjustment process is retried after the winding temperature or current recovers to the safe range.

[0066] In one embodiment, such as Figure 5 As shown, step S40 specifically includes the following steps: S41: Compare the speed feedback data during the wind speed adjustment process with the target speed to obtain speed feedback closed-loop error data. Perform curve fitting on the speed feedback closed-loop error data and perform second derivative calculation to obtain the acceleration slope change rate. S42: If the rate of change of the acceleration slope is within the preset stable range, the operating status data of the air curtain machine motor is extracted to obtain the motor output torque and operating power data. S43: Substitute the motor output torque and operating power data into the fluid dynamics model to calculate the barrier strength data; S44: Perform time-domain analysis on the barrier strength data to obtain the barrier stability maintenance duration characteristics; S45: Based on external environmental data, predict the moment when external air breaks through the barrier to obtain the predicted moment of external air intrusion, and calculate the time difference between the barrier stability maintenance duration characteristic and the predicted moment of external air intrusion. S46: Compare the time difference with the preset advance protection response conditions. If the time difference meets the preset advance protection response conditions, generate a continuous operation control command for the air curtain machine.

[0067] Specifically, after the drive motor performs the wind speed adjustment action, the motor does not instantly reach the target speed, but undergoes an acceleration process. During this process, there is a continuous deviation between the actual speed and the target speed, and the trend of this deviation directly reflects whether the motor acceleration process is smooth and controllable. If the motor speed oscillates or the acceleration fluctuates violently during acceleration, it indicates a mismatch between the control signal and the current load state of the motor. In this case, the collected motor operating status data cannot accurately reflect the barrier strength under stable operating conditions, and it is necessary to wait for the acceleration process to stabilize before evaluating the barrier strength. Therefore, this embodiment first judges the stability of the acceleration process before deciding whether to proceed to the barrier strength calculation stage.

[0068] In this embodiment, the actual motor speed is continuously collected at a sampling frequency of 100 Hz. The difference between the actual speed and the target speed at each sampling moment is used to obtain the speed deviation value at that moment. The proportional-integral-differential algorithm is used to process the speed deviation value sequence. The proportional term reflects the instantaneous magnitude of the deviation at the current moment, the integral term sums up the historical deviations to eliminate steady-state error, and the differential term estimates the rate of change of the deviation to suppress overshoot. The weighted sum of the three terms yields the speed feedback closed-loop error data at each sampling moment.

[0069] After obtaining the speed feedback closed-loop error data sequence, this embodiment takes error data from several consecutive sampling periods and uses the least squares method to curve-fit the trend of error change over time, obtaining a smooth curve of error change over time. The least squares method solves for the coefficients of the optimal fitting polynomial by minimizing the sum of squares of the differences between the fitted curve and the error values ​​at each sampling point. Compared with the original error sequence, the obtained smooth curve filters out the interference of sampling noise and more accurately reflects the true trend of error change. Taking the second derivative of this smooth curve, the first derivative corresponds to the rate of change of error, i.e., acceleration, and the second derivative corresponds to the rate of change of acceleration, i.e., the rate of change of acceleration slope. This value reflects whether the acceleration tends to be stable during the acceleration process of the motor.

[0070] For example, the system takes error data from 10 consecutive sampling periods, fits it using the least squares method, and then performs a second derivative on the smooth curve to obtain the current acceleration slope change rate as 15 revolutions per second squared.

[0071] This embodiment compares the rate of change of acceleration slope with a preset stable interval. The boundaries of the preset stable interval are determined based on the statistical distribution of the rate of change of acceleration slope under normal acceleration conditions. The 5th and 95th percentiles of the distribution of the rate of change of acceleration slope in the normal operating condition sample are taken as the lower and upper boundaries of the interval. If the acceleration falls within this interval, it indicates that the smoothness of the current acceleration process is consistent with the normal operating condition, and the subsequent barrier strength assessment can proceed. If the acceleration falls outside this interval, it indicates that there are abnormal fluctuations in the motor acceleration process, and we continue to wait until the rate of change falls back to the stable interval. In this embodiment, the preset stable interval is set to -20 to +20 revolutions per second squared.

[0072] After confirming that the rate of change of acceleration slope is within a preset stable range, this embodiment extracts the current output torque and operating power data of the motor at a sampling frequency of 50 Hz. The output torque is estimated by the product of the motor phase current and the motor torque constant, and the operating power is calculated by the product of the bus voltage and the bus current. Both reflect the state of mechanical energy output by the motor to the impeller of the air curtain machine at the current speed.

[0073] Based on the above description, this embodiment substitutes the motor output torque and operating power data into the fluid dynamics model to calculate the strength of the air barrier formed by the current air curtain machine at the building entrance. The fluid dynamics model is based on Bernoulli's equation. Its basic principle is that the air curtain machine impeller converts the mechanical power output by the motor into air kinetic energy. After the airflow is ejected downwards at a certain speed, it forms dynamic pressure at the entrance / exit cross-section. This dynamic pressure is the source of the air curtain barrier's ability to block the exchange of indoor and outdoor air. The specific calculation process is as follows: First, the volumetric flow rate of the airflow is estimated based on the operating power and the air curtain machine impeller efficiency coefficient, which is provided by the air curtain machine's factory calibration data. Then, the outlet airflow velocity is obtained by dividing the volumetric flow rate by the cross-sectional area of ​​the air curtain machine's outlet. Finally, according to Bernoulli's equation, the dynamic pressure value of the air curtain barrier, i.e., the barrier strength data, is obtained by multiplying the square of the airflow velocity by the air density and then dividing by 2. For example, if the current motor operating power calculates an airflow velocity of 8 meters per second and the air density is taken as 1.2 kilograms per cubic meter, the calculated barrier strength data is 38.4 Pascals.

[0074] After obtaining the barrier strength data, this embodiment continuously collects the data at a sampling frequency of 10 Hz to form a barrier strength time series. A sliding window algorithm is then used to perform time-domain analysis on this series. The sliding window length is determined based on the shortest duration for which the barrier strength remains stable under typical operating conditions; in this embodiment, it is set to 600 seconds. Within each sliding window, the variance of the barrier strength series is calculated. If the variance is lower than a preset variance threshold, the barrier is considered to be in a stable state within the current window. The preset variance threshold is determined based on the distribution of the barrier strength series variance in historical normal operating data, taking the 95th quantile of the variance distribution under normal stable operating conditions; in this embodiment, it is set to 0.25. The system statistically analyzes the continuous duration for which the barrier strength variance remains below the preset variance threshold and extracts this duration as a barrier stability maintenance duration feature. For example, the system statistically analyzes the current barrier stability maintenance duration feature as 900 seconds.

[0075] After obtaining the characteristics of the barrier's stability duration, this embodiment predicts the moment when external air breaks through the barrier based on external environmental data. This embodiment uses a Long Short-Term Memory (LSTM) network model to process outdoor air pressure and wind direction change sequences to predict the moment when strong external winds reach the building entrance. The LTM network models long-term dependencies in the time-series data through a gating mechanism. The input layer receives air pressure and wind direction values ​​from the current moment and several past moments. A forget gate filters out information that needs to be retained from historical states. An input gate determines the update magnitude of the current input to the hidden state. An output gate generates the hidden state output for the current moment. Finally, a fully connected layer maps this to the time offset of the predicted external air intrusion moment, which is then added to the current moment to obtain the predicted external air intrusion moment.

[0076] For example, based on the current downward trend of air pressure and changes in wind direction, the Long Short-Term Memory Network predicts that the strong external wind will reach the building entrance 1200 seconds from the current moment, thus obtaining the predicted time of external air intrusion.

[0077] After obtaining the barrier stability maintenance duration characteristic and the predicted time of external air intrusion, this embodiment adds the barrier stability maintenance duration characteristic to the current time to obtain the predicted barrier failure time. Then, it subtracts the predicted barrier failure time from the predicted external air intrusion time to obtain the time difference. A positive time difference value indicates that the external air intrusion time is later than the barrier failure time, meaning that the barrier will fail before the external air intrusion, posing a risk of a protective gap. A negative or zero time difference value indicates that the barrier can still maintain effective protection when external air intrudes.

[0078] This embodiment compares the time difference with a preset early protection response condition, which is that the time difference is less than or equal to a preset time threshold. The preset time threshold is determined based on the longest response time required for the air curtain machine to complete barrier strength enhancement from receiving the continuous operation control command, ensuring sufficient time for barrier enhancement before actual barrier failure. In this embodiment, 400 seconds is used. If the time difference meets the preset early protection response condition, the system generates a continuous operation control command for the air curtain machine, increases the air curtain machine's operating frequency via the industrial control bus, and locks it in continuous operation mode, completing barrier enhancement ahead of schedule and eliminating the risk of a protection gap. For example, if the current time difference is 300 seconds, which is less than the preset time threshold of 400 seconds, the preset early protection response condition is met. The system generates a continuous operation control command, increases the air curtain machine's operating frequency from 35 Hz to 50 Hz, and locks it in operation.

[0079] In one embodiment, the speed feedback data during the wind speed adjustment process is compared with the target speed to obtain speed feedback closed-loop error data. Curve fitting and second derivative calculation are then performed on the speed feedback closed-loop error data to obtain the acceleration slope change rate, including: The difference between the speed feedback data and the target speed during the wind speed adjustment process is calculated, and the difference is processed by the proportional-integral-differential algorithm to obtain the speed feedback closed-loop error data. The least squares method was used to perform curve fitting on the speed feedback closed-loop error data to obtain an error smoothing curve; The second derivative of the error smoothing curve is performed to obtain the rate of change of the acceleration slope.

[0080] In this embodiment, the motor begins to accelerate after receiving the control signal generated in step S30. Due to the inertia of the motor rotor, the actual rotational speed cannot reach the target speed instantly, but gradually approaches it over a period of time. During this process, there is always a deviation between the actual rotational speed and the target rotational speed. Simply recording the deviation value at a certain moment cannot reflect whether the acceleration process is smooth; it is necessary to analyze the trend of the deviation over time. Therefore, this embodiment continuously reads the actual rotational speed of the motor at a sampling frequency of 100 Hz, and calculates the difference between the actual rotational speed and the target rotational speed at each sampling moment to obtain the rotational speed deviation value at that moment.

[0081] After obtaining the speed deviation values ​​at each sampling time, this embodiment uses a proportional-integral-derivative (PID) algorithm to process the deviation value sequence and obtain the speed feedback closed-loop error data. The PID algorithm consists of three components: the proportional component equals the current deviation value multiplied by the proportional coefficient Kp, directly reflecting the magnitude of the current deviation; the larger the deviation, the larger the proportional component. The integral component equals the cumulative sum of deviation values ​​at all historical times multiplied by the integral coefficient Ki, used to eliminate long-term steady-state deviations, ensuring the actual speed eventually converges completely to the target speed. The derivative component equals the difference between the current deviation value and the previous deviation value divided by the sampling interval, then multiplied by the derivative coefficient Kd, reflecting the rate of change of the deviation and used to suppress speed overshoot. The three components are added together to obtain the speed feedback closed-loop error data at the current time. The proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd are tuned according to the motor's mechanical and electrical time constants, ensuring that the closed-loop system can converge quickly without significant oscillations under step response.

[0082] For example, when the target speed command is 1000 rpm and the actual speed is 950 rpm, the deviation value at the current moment is 50 rpm. The weighted sum of the proportional component, integral component and differential component is used to obtain the speed feedback closed-loop error data at the current moment.

[0083] After obtaining the closed-loop error data sequence of the rotational speed feedback over a period of time, this embodiment takes error data from several consecutive sampling periods and uses the least squares method to curve-fit the trend of error change over time. The specific process of the least squares method is as follows: let the fitting polynomial be a function of time t, take each sampling time as the independent variable and the corresponding error value as the dependent variable, construct a system of equations, and obtain the optimal fitting polynomial by solving for the polynomial coefficients that minimize the sum of squares of the differences between the fitted value and the measured value. This embodiment uses a quadratic polynomial for fitting, that is, the fitting curve is in the form of at² + bt + c, where the coefficients a, b, and c are determined by solving the least squares method. The quadratic polynomial can describe the typical accelerated convergence characteristics of the error, which first decreases rapidly and then tends to stabilize. At the same time, compared with higher-order polynomials, it is less prone to overfitting, and the obtained error smoothing curve retains the main trend of error change while filtering out sampling noise.

[0084] After obtaining the error smoothing curve, this embodiment performs a second derivative operation. The first derivative of the fitted polynomial at² + bt + c is taken, yielding the first derivative 2at + b, which reflects the instantaneous rate of error change, i.e., acceleration. Further derivative of the first derivative yields the second derivative 2a, which is the rate of change of the acceleration slope. The second derivative is a constant, reflecting whether the overall trend of acceleration change is stable within the current fitting window: if the absolute value of the second derivative is small, it indicates that the acceleration changes smoothly within the current window, and the motor acceleration process tends to be stable; if the absolute value of the second derivative is large, it indicates that the acceleration is still changing rapidly, and the motor has not yet entered a stable acceleration state.

[0085] For example, the system takes error data from 10 consecutive sampling periods, and fits it using the least squares method to obtain a quadratic polynomial coefficient a of 7.5. Then the rate of change of acceleration slope is 2 multiplied by 7.5, which equals 15 revolutions per second squared. This value is then compared with the preset stable range of -20 to +20 revolutions per second squared to determine whether the current acceleration process meets the stable conditions for entering the barrier strength assessment.

[0086] This application also provides a system for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity. This system may include: The gradient extraction module is used to perform difference operations and vector construction on the air temperature change sequence and humidity change sequence outside the building entrance to obtain the environmental gradient feature vector. The risk classification module is used to classify risks based on the environmental gradient feature vector to obtain the risk level of the protective windows at the building entrances and exits. The wind speed adjustment module is used to calculate the target wind speed adjustment amount of the air curtain based on the risk level of the protective window at the building entrance and the real-time speed feedback data of the air curtain motor, and generate a control signal for the air curtain motor based on the target wind speed adjustment amount to drive the air curtain motor to perform wind speed adjustment action. The barrier monitoring module is used to calculate barrier strength data based on the rotational speed feedback data during the wind speed adjustment process, and to calculate the time difference between the barrier strength data and the predicted time of external air intrusion, and to generate a continuous operation control command for the air curtain machine based on the time difference.

[0087] It should be noted that the embodiments of the system for adaptively adjusting the wind speed of the air curtain machine based on ambient temperature and humidity provided in this application can be used to execute the processing flow of the embodiments of the method for adaptively adjusting the wind speed of the air curtain machine based on ambient temperature and humidity in the above embodiments. Its functions will not be repeated here, but can be referred to the detailed description of the above method embodiments.

[0088] As described above, the system for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity, provided in this application, obtains an environmental gradient feature vector by performing differential operations and vector construction on the external temperature and humidity change sequences at building entrances and exits. Risk classification is then performed based on the environmental gradient feature vector to obtain the risk level of protected windows. The target wind speed adjustment amount is calculated based on the protected window risk level and the real-time speed feedback data of the air curtain machine motor, generating a motor control signal to drive the motor to perform the wind speed adjustment action. Barrier strength data is calculated based on the speed feedback data, and the time difference between the barrier strength data and the predicted time of external air intrusion is calculated. A continuous operation control command is generated based on the time difference. This system senses changes in the environmental gradient and predicts the risk of protected windows, completing wind speed adjustment before substantial changes in temperature and humidity occur, eliminating the protected window period caused by post-event feedback mechanisms.

[0089] This application also provides an electronic device, which includes one or more processors and memory resources represented by a memory for storing instructions executable by the processor, such as application programs. The application programs stored in the memory may include one or more modules, each corresponding to a set of instructions. Furthermore, the processor is configured to execute instructions to perform the aforementioned method for adaptively adjusting the air curtain machine's fan speed based on ambient temperature and humidity.

[0090] The electronic device may also include a power supply component configured to perform power management of the electronic device, a wired or wireless network interface configured to connect the electronic device to a network, and an input / output (I / O) interface. The electronic device can be operated based on operating devices stored in memory, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0091] In one embodiment, a computer device, which may be a server, is also provided. The computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device stores data. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity.

[0092] In one embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity. The display unit of the computer device is used to form a visually visible image and may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0093] This application also provides a non-transitory computer-readable storage medium. When the instructions in the storage medium are executed by the processor of the aforementioned electronic device, the electronic device is able to execute a method for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity. The method includes: performing differential operations and vector construction on the temperature and humidity change sequences outside the building entrance to obtain an environmental gradient feature vector; classifying risks based on the environmental gradient feature vector to obtain the risk level of the protected windows at the building entrance; calculating a target air curtain wind speed adjustment amount based on the risk level of the protected windows at the building entrance and the real-time speed feedback data of the air curtain machine motor, and generating an air curtain machine motor control signal based on the target air curtain wind speed adjustment amount to drive the air curtain machine motor to perform wind speed adjustment actions; calculating barrier strength data based on the speed feedback data during the wind speed adjustment action, and calculating the time difference between the barrier strength data and the predicted time of external air intrusion, and generating a continuous operation control command for the air curtain machine based on the time difference.

[0094] This application may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information may be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to: phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0095] It should be noted that although the steps of the method for adaptively adjusting the air curtain machine speed based on ambient temperature and humidity in this application are described in a specific order in the accompanying drawings, this does not require or imply that these steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps, such as omitting certain steps, combining multiple steps into one step, and / or breaking down one step into multiple steps, should all be considered part of this application.

[0096] It should be understood that this application is not limited to the detailed structure and arrangement of the modules in the system for adaptively adjusting the wind speed of an air curtain machine based on ambient temperature and humidity as described in this specification. This application can have other embodiments and can be implemented and executed in various ways. The foregoing variations and modifications fall within the scope of this application. It should be understood that the disclosure and definition of this application extend to all alternative combinations of two or more individual features mentioned or apparent in the text and / or drawings. All these different combinations constitute multiple alternative aspects of this application. The embodiments described in this specification illustrate the best known mode for implementing this application and will enable those skilled in the art to utilize this application.

Claims

1. A method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity, characterized in that, include: By performing difference operations and vector construction on the air temperature and humidity change sequences outside the building entrance, an environmental gradient feature vector is obtained. Risk classification is performed based on the environmental gradient feature vector to obtain the risk level of the protective open window at the building entrance; The target air curtain speed adjustment amount is calculated based on the risk level of the protective window at the building entrance and the real-time speed feedback data of the air curtain machine motor. The air curtain machine motor control signal is generated based on the target air curtain speed adjustment amount to drive the air curtain machine motor to perform the speed adjustment action. The barrier strength data is calculated based on the rotational speed feedback data during the wind speed adjustment process, and the time difference between the barrier strength data and the predicted time of external air intrusion is calculated. Based on the time difference, a continuous operation control command for the air curtain machine is generated.

2. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 1, characterized in that, The environmental gradient feature vector obtained by performing difference operations and vector construction on the air temperature and humidity change sequences outside the building entrance includes: The temperature difference between adjacent time points in the air temperature change sequence outside the building entrance is calculated, and the result with the largest absolute value of the difference is extracted as the temperature decrease gradient feature. The humidity values ​​at adjacent time points in the sequence of air humidity changes outside the building entrance are calculated, and the maximum positive difference is extracted as the humidity rise gradient feature. An environmental gradient feature vector is obtained by constructing vectors based on the temperature decrease gradient feature and the humidity increase gradient feature.

3. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 2, characterized in that, After constructing the environmental gradient feature vector based on the temperature decrease gradient feature and the humidity increase gradient feature, the method further includes: Anomaly detection is performed on the environmental gradient feature vector to obtain an abnormal environmental gradient value vector, and the corresponding abnormal timestamp is extracted based on the abnormal environmental gradient value vector; Based on the abnormal timestamp, extract the corresponding abnormal speed value sequence from the real-time speed feedback data of the air curtain machine motor; Spectral analysis is performed on the abnormal speed value sequence to obtain the speed amplitude sequence. The maximum amplitude value in the speed amplitude sequence is compared with a preset amplitude threshold. If the maximum amplitude value exceeds the preset amplitude threshold, it is determined that the air curtain machine motor is in an overload operation state, and an air curtain machine motor shutdown control command is generated. The air curtain machine motor shutdown control command is used to trigger the air curtain machine motor to stop.

4. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 1, characterized in that, The risk classification based on the environmental gradient feature vector to obtain the risk level of the protective window at the building entrance includes: The environmental gradient feature vector is normalized to obtain a normalized environmental gradient feature vector, and the split gain value of the decision tree branch node is calculated based on the normalized environmental gradient feature vector. Multiple decision trees are constructed based on the split gain values ​​of the branch nodes of the decision tree. The sample distribution density of the leaf nodes of the multiple decision trees is statistically analyzed to obtain the sample distribution density values ​​of the leaf nodes. Based on the sample distribution density values ​​of the leaf nodes, the prediction results of the multiple decision trees are categorized and voted to obtain the risk level of the protective void at the building entrance.

5. The method for adaptively adjusting the air curtain machine speed based on ambient temperature and humidity according to claim 4, characterized in that, The step of normalizing the environmental gradient feature vector to obtain a normalized environmental gradient feature vector, and calculating the split gain value of the decision tree branch node based on the normalized environmental gradient feature vector, includes: The maximum and minimum gradient values ​​are determined based on historical environmental gradient data. The maximum and minimum normalization of each feature value in the environmental gradient feature vector are calculated to obtain the normalized environmental gradient feature vector. The initial information entropy is calculated based on the sample category distribution contained in the current node in the normalized environmental gradient feature vector. The samples are divided using the feature threshold as the split point. The conditional entropy is calculated based on the sample category distribution after the division. The difference between the initial information entropy and the conditional entropy is calculated to obtain the split gain value of the decision tree branch node. The split gain value of the decision tree branch node is compared according to the principle of maximum information gain. If the split gain value is lower than the preset stopping threshold, the node splitting is stopped.

6. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 1, characterized in that, The step of calculating the target air curtain speed adjustment amount based on the risk level of the protective windows at the building entrance and the real-time speed feedback data of the air curtain machine motor, and generating an air curtain machine motor control signal based on the target air curtain speed adjustment amount to drive the air curtain machine motor to perform the air speed adjustment action includes: Based on the risk level of the protective window and the real-time speed feedback data of the air curtain motor, the response delay of the air curtain motor is compensated and calculated to obtain the target air curtain speed adjustment amount. The duty cycle is calculated based on the target air curtain wind speed adjustment amount to obtain the pulse width modulation duty cycle parameter, and the air curtain machine motor control signal is generated based on the pulse width modulation duty cycle parameter. The safe operating status of the air curtain machine motor is detected, and the safety detection result is obtained. The control signal of the air curtain machine motor is corrected according to the safety detection result to obtain the corrected control signal of the air curtain machine motor. The air curtain machine motor is driven to perform wind speed adjustment action according to the corrected control signal of the air curtain machine motor.

7. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 1, characterized in that, The calculation of barrier strength data based on rotational speed feedback data during the wind speed adjustment process includes: The rotational speed feedback data during the wind speed adjustment process is compared with the target rotational speed to obtain the rotational speed feedback closed-loop error data. The rotational speed feedback closed-loop error data is then curve-fitted and a second derivative is performed to obtain the acceleration slope change rate. If the rate of change of the acceleration slope is within the preset stable range, the operating status data of the air curtain machine motor is extracted to obtain the motor output torque and operating power data. The barrier strength data is obtained by substituting the motor output torque and operating power data into the fluid dynamics model.

8. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 7, characterized in that, The speed feedback data during the wind speed adjustment process is compared with the target speed to obtain speed feedback closed-loop error data. The speed feedback closed-loop error data is then curve-fitted and a second derivative is performed to obtain the acceleration slope change rate, including: The difference between the speed feedback data and the target speed during the wind speed adjustment process is calculated, and the difference is processed by the proportional-integral-differential algorithm to obtain the speed feedback closed-loop error data. The least squares method was used to perform curve fitting on the speed feedback closed-loop error data to obtain an error smoothing curve; The rate of change of acceleration slope is obtained by performing a second derivative operation on the error smoothing curve.

9. The method for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity according to claim 1, characterized in that, The step of calculating the time difference between the barrier strength data and the predicted time of external air intrusion, and generating a continuous operation control command for the air curtain machine based on the time difference, includes: Time-domain analysis was performed on the barrier strength data to obtain the barrier stability maintenance duration characteristics; The timing of external air intrusion is predicted based on external environmental data to obtain the predicted timing of external air intrusion. The time difference between the predicted timing of external air intrusion and the characteristics of the barrier stability maintenance duration is calculated. The time difference is compared with the preset advance protection response conditions. If the time difference meets the preset advance protection response conditions, a continuous operation control command for the air curtain machine is generated.

10. A system for adaptively adjusting the airflow speed of an air curtain machine based on ambient temperature and humidity, characterized in that, include: The gradient extraction module is used to perform difference operations and vector construction on the air temperature change sequence and humidity change sequence outside the building entrance to obtain the environmental gradient feature vector. The risk classification module is used to classify risks based on the environmental gradient feature vector to obtain the risk level of the protective windows at the building entrances and exits. The wind speed adjustment module is used to calculate the target wind speed adjustment amount of the air curtain based on the risk level of the protective window at the building entrance and the real-time speed feedback data of the air curtain motor, and generate a control signal for the air curtain motor based on the target wind speed adjustment amount to drive the air curtain motor to perform wind speed adjustment action. The barrier monitoring module is used to calculate barrier strength data based on the rotational speed feedback data during the wind speed adjustment process, and to calculate the time difference between the barrier strength data and the predicted time of external air intrusion, and to generate a continuous operation control command for the air curtain machine based on the time difference.

Images