Methods, devices, and electronic equipment for determining the setting information of fresh air systems

Abstract

This invention provides a method, apparatus, and electronic device for determining the setting information of a fresh air system. The method involves acquiring real-time information about the space where the fresh air system is located, including environmental information; inputting this real-time information into a classification prediction model to obtain predicted operating status information (whether the fresh air system is running or not); inputting the real-time information into a regression prediction model to obtain predicted setting information; and determining the setting information of the fresh air system based on the predicted operating status information and the predicted setting information. The training steps of the classification prediction model include acquiring training samples, including sample information in a sample space and the expected setting information of the fresh air system corresponding to the sample information. The sample information includes at least environmental information. The training samples are mapped onto a high-dimensional ellipsoid, and the distance is measured using cosine to ensure the training samples are distributed ellipsoidally. The classification prediction model is then trained using an SVM classification algorithm and Softmax loss.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a method, apparatus, and electronic device for determining setting information of a fresh air system. Background Technology

[0002] In related technologies, fresh air systems in indoor venues such as shopping malls and hotels are manually set to turn on, off, or at different frequencies based on human experience and perception. The set strategies are generally not changed, resulting in high energy consumption and energy waste. Summary of the Invention

[0003] This invention provides a method, apparatus, and electronic device for determining the setting information of a fresh air system, in order to solve the problem that the setting of the opening, closing, or frequency of fresh air systems in the prior art leads to high energy consumption and energy waste.

[0004] This invention provides a method for determining the setting information of a fresh air device, comprising:

[0005] Obtain real-time information about the space where the fresh air device is located, including at least environmental information;

[0006] The real-time information is input into the classification prediction model to obtain the operating status information of the fresh air equipment output by the classification prediction model. The predicted operating status information includes whether the fresh air equipment is running or not.

[0007] The real-time information is input into the regression prediction model to obtain the prediction setting information of the fresh air equipment output by the regression prediction model;

[0008] Based on the predicted operating status information and the predicted setting information, the setting information of the fresh air equipment is determined;

[0009] The training steps of the classification prediction model include:

[0010] Acquire training samples, which include sample information of the sample space and expected setting information of the fresh air equipment corresponding to the sample information, wherein the sample information includes at least environmental information;

[0011] The training samples are mapped onto a high-dimensional ellipsoid, and the side length of the triangle on the ellipsoid is calculated using the cosine function as a distance metric, so that the training samples are distributed on an ellipsoid.

[0012] Based on the training samples distributed in an elliptic shape, the classification prediction model is obtained by training the Support Vector Machine (SVM) classification algorithm and the Softmax loss function.

[0013] According to the method for determining the setting information of a fresh air device provided by the present invention, the training steps of the regression prediction model include:

[0014] Obtain the training samples;

[0015] Based on the training samples, the regression prediction model is trained using a multiple linear regression prediction algorithm and a gradient descent method.

[0016] According to the method for determining the setting information of a fresh air device provided by the present invention, the environmental information includes at least one of the following:

[0017] The temperature, humidity, carbon dioxide concentration, PM2.5 concentration, negative pressure, and flow of people in the space are all measured.

[0018] According to the method for determining the setting information of a fresh air device provided by the present invention, the fresh air device includes a fresh air unit, a return air unit, and an exhaust fan, and the setting information of the fresh air device includes at least one of the following:

[0019] The set frequency of the fresh air unit, the set frequency of the return air unit, and the set frequency of the exhaust fan.

[0020] According to the method for determining the setting information of a fresh air device provided by the present invention, after obtaining the training samples, the method further includes:

[0021] The training samples are preprocessed, and the preprocessing includes processing one or more of the following:

[0022] Missing values ​​in the training samples;

[0023] Negative values ​​in the training samples;

[0024] The abnormally large values ​​in the training samples;

[0025] The abnormally small values ​​in the training samples.

[0026] According to the method for determining the setting information of a fresh air device provided by the present invention, after obtaining the training samples, the method further includes:

[0027] The training samples are normalized.

[0028] The present invention also provides a device for determining setting information of a fresh air system, comprising:

[0029] The acquisition module is used to acquire real-time information about the space where the fresh air device is located, and the real-time information includes at least environmental information;

[0030] The first input module is used to input the real-time information into the classification prediction model to obtain the predicted operating status information of the fresh air equipment output by the classification prediction model, wherein the operating status information includes operating and not operating.

[0031] The second input module is used to input the real-time information into the regression prediction model to obtain the prediction setting information of the fresh air equipment output by the regression prediction model.

[0032] The determination module is used to determine the setting information of the fresh air equipment based on the predicted operating status information and the predicted setting information;

[0033] The training steps of the classification prediction model include:

[0034] Acquire training samples, which include sample information of the sample space and expected setting information of the fresh air equipment corresponding to the sample information, wherein the sample information includes at least environmental information;

[0035] The training samples are mapped onto a high-dimensional ellipsoid, and the side length of the triangle on the ellipsoid is calculated using the cosine function as a distance metric, so that the training samples are distributed on an ellipsoid.

[0036] Based on the training samples distributed in an elliptic shape, the classification prediction model is obtained by training the Support Vector Machine (SVM) classification algorithm and the Softmax loss function.

[0037] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-mentioned method for determining the setting information of the fresh air device.

[0038] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned method for determining the setting information of a fresh air device.

[0039] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned method for determining the setting information of a fresh air device.

[0040] This invention provides a method, apparatus, and electronic device for determining the setting information of a fresh air system. The method acquires real-time information about the space where the fresh air system is located. This information is then input into two prediction models: a classification prediction model and a regression prediction model. The two models predict the optimal operating status and setting information of the fresh air system for customer comfort. The classification prediction model maps its training onto a high-dimensional ellipsoid, resulting in better intra-class compactness and inter-class diversity of the training sample features in the metric space. Based on the ellipsoidal distribution of the training samples, the model is trained using an SVM classification algorithm and a Softmax loss function, making the training results more accurate. By setting the fresh air system in real-time based on the predicted operating status and setting information, customer comfort can be improved and energy consumption reduced. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0042] Figure 1 This is a flowchart illustrating the method for determining the setting information of a fresh air device provided by the present invention;

[0043] Figure 2 This is a schematic diagram of the training sample data provided by the present invention;

[0044] Figure 3 This is a schematic diagram of the automatic control process of the fresh air system provided by the present invention;

[0045] Figure 4 This is a schematic diagram of the setting information determination device for the fresh air equipment provided by the present invention;

[0046] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0048] First, let's introduce the following:

[0049] A fresh air system works by using specialized equipment to supply fresh air into a sealed room from one side and exhausting it to the outside from the other side, creating a "fresh air flow field" within the room, thus meeting the need for indoor ventilation.

[0050] A fresh air system mainly consists of multiple fresh air units, multiple return air units, multiple exhaust fans, ductwork, waste air vents, and fresh air inlets. Fresh air systems are classified into: unidirectional flow fresh air systems, bidirectional flow fresh air systems, and total heat exchange fresh air systems. Total heat exchange fresh air systems achieve simultaneous and equal-volume conversion of indoor exhaust air and fresh air. During ventilation, the fresh air recovers most of the heat (cold) and moisture from the exhaust air through the total heat exchanger, minimizing the energy loss from exhaust ventilation.

[0051] The start-up and operation frequency of fresh air units, return air units, and exhaust fans affect indoor comfort and power consumption. The minimum energy consumption of a fresh air system is affected by the following factors: 1. Indoor and outdoor temperature; 2. Indoor and outdoor humidity; 3. Indoor and outdoor carbon dioxide concentration; 4. Indoor and outdoor fine particulate matter (PM2.5); 5. Indoor negative pressure; 6. Indoor occupancy; 7. The operation and frequency of multiple fresh air units; 8. The operation and frequency of multiple return air units; 9. The operation and frequency of multiple exhaust fans.

[0052] Currently, the on / off status and frequency of various devices in hotel and shopping mall ventilation systems are determined by hotel managers based on experience. These settings are generally not changed, resulting in high energy consumption and significant energy waste in hotel and shopping mall systems.

[0053] To address the aforementioned problems, this invention provides a method, apparatus, and electronic device for determining the setting information of a fresh air system, which can automatically adjust the opening, closing, or frequency of fresh air systems in shopping malls or hotels, thereby reducing energy consumption.

[0054] Figure 1 This is a flowchart illustrating the method for determining the setting information of a fresh air device provided by the present invention, as shown below. Figure 1 As shown, the method for determining the setting information of the fresh air device includes the following steps:

[0055] Step 100: Obtain real-time information about the space where the fresh air device is located, including at least environmental information;

[0056] Optionally, since indoor comfort is affected by environmental factors such as temperature, humidity, and carbon dioxide concentration, and the settings of the fresh air equipment required to achieve optimal comfort vary for different environments, in order to adjust the settings of the fresh air equipment in real time to adjust indoor comfort, the real-time environmental information of the space where the fresh air equipment is located can be obtained first.

[0057] Alternatively, the environmental information can be any information that reflects the environment of the space where the fresh air equipment is located, such as the temperature and humidity of the shopping mall or hotel where the fresh air equipment is located.

[0058] Optionally, real-time information about the space where the fresh air equipment is located can be obtained through manual measurement, management system, or Internet of Things devices, such as temperature sensors, data acquisition boards, multi-channel control boards, data acquisition devices, power sensors, frequency converters, etc. The present invention does not limit this.

[0059] For example, the temperature information of the space where the fresh air equipment is located can be obtained in real time through a temperature sensor in the space where the fresh air equipment is located.

[0060] Step 110: Input the real-time information into the classification prediction model to obtain the operating status information of the fresh air equipment output by the classification prediction model. The predicted operating status information includes whether the equipment is running or not.

[0061] Optionally, after obtaining real-time information about the space where the fresh air equipment is located, the management system can input the real-time information about the space where the fresh air equipment is located into the classification prediction model to obtain the predicted operating status information output by the classification prediction model.

[0062] Optionally, the classification prediction model can be any model that can output prediction settings after inputting real-time information of the space where the fresh air device is located; this invention does not limit this.

[0063] Optionally, the predicted operating status information can be information output by a classification prediction model that can be used to determine the operating status of the fresh air equipment, such as whether the fresh air equipment is operating or not.

[0064] Step 120: Input the real-time information into the regression prediction model to obtain the prediction setting information of the fresh air equipment output by the regression prediction model;

[0065] Optionally, after the management system obtains the real-time information of the space where the fresh air equipment is located, it can input the real-time information of the space where the fresh air equipment is located into the regression prediction model to obtain the prediction setting information output by the regression prediction model.

[0066] Optionally, the regression prediction model can be any model that can output predicted setting information after inputting real-time information of the space where the fresh air device is located; the present invention does not limit this.

[0067] Optionally, the prediction setting information can be information output by a regression prediction model that can be used to determine the setting information of the fresh air equipment, such as the frequency and power of the fresh air equipment.

[0068] Step 130: Based on the predicted operating status information and the predicted setting information, determine the setting information of the fresh air equipment;

[0069] The training steps of the classification prediction model include:

[0070] Acquire training samples, which include sample information of the sample space and expected setting information of the fresh air equipment corresponding to the sample information, wherein the sample information includes at least environmental information;

[0071] The training samples are mapped onto a high-dimensional ellipsoid, and the side length of the triangle on the ellipsoid is calculated using the cosine function as a distance metric, so that the training samples are distributed on an ellipsoid.

[0072] Based on the training samples distributed in an elliptic shape, the classification prediction model is obtained by training the Support Vector Machine (SVM) classification algorithm and the Softmax loss function.

[0073] Optionally, in order to obtain training samples, some spaces with fresh air equipment can be selected as sample spaces. Sample information of the sample space and expected setting information of the fresh air equipment corresponding to the sample information can be obtained. Using the sample information and expected setting information of the fresh air equipment, an SVM classification algorithm and a softmax loss function can be used to train a classification prediction model.

[0074] Optionally, the desired setting information of the fresh air equipment corresponding to the sample information can be the fresh air equipment setting information that makes people in the space feel the most comfortable in the sample space environment.

[0075] For example, if turning on 6 fresh air devices in a shopping mall corresponding to a certain sample information would make people in the mall feel the most comfortable, then the expected setting information for the fresh air devices corresponding to that sample information could include turning on 6 fresh air devices.

[0076] Figure 2 This is a schematic diagram of the training sample data provided by the present invention, such as... Figure 2 As shown, environmental information from multiple sample spaces and the corresponding expected settings for fresh air equipment can be obtained as training samples.

[0077] Alternatively, the Softmax Loss function has good convergence and representation vector, but the data of fresh air equipment has the characteristic of data clustering at the center point. Direct classification on the original data leads to poor results or even inseparability.

[0078] Optionally, considering the inseparable clustering of fresh air data points, training samples can be mapped onto a high-dimensional ellipsoid. The cosine function is used to calculate the side length of the triangle on the ellipsoid as a distance metric, i.e., the distance and angle between each training sample and the center of the ellipsoid are calculated. This makes the training samples distributed on an ellipsoid, representing the solution effect of the fresh air equipment in a high dimension, while the kernel function is calculated in a low dimension. This solves the problem of inseparable fresh air equipment data and makes the features of the training samples have better intra-class compactness and inter-class diversity in the metric space.

[0079] Optionally, sample information from multiple sample spaces, along with the corresponding expected settings of the fresh air equipment, can be used as training samples. The training samples are mapped onto a high-dimensional ellipsoid, and the distance is measured using cosine to ensure that the training samples are distributed on an ellipsoid. The first classification prediction model is then trained using the SVM classification algorithm and the Softmax loss function.

[0080] Optionally, after training the classification prediction model, real-time information of the space where the fresh air equipment is located can be input into the classification prediction model to obtain predicted operating status information.

[0081] Optionally, the operating status of the fresh air equipment can be set to the predicted operating status by sending a notification through an IoT device based on the predicted operating status information of the fresh air equipment.

[0082] Optionally, the setting information of the fresh air device can be transmitted via IoT devices based on the predicted setting information of the fresh air device.

[0083] The method for determining the setting information of a fresh air system provided by this invention obtains real-time information about the space where the fresh air system is located. This information is then input into two prediction models: a classification prediction model and a regression prediction model. The two models predict the optimal operating status and setting information of the fresh air system for customer comfort. The classification prediction model maps its training onto a high-dimensional ellipsoid, resulting in better intra-class compactness and inter-class diversity of the training sample features in the metric space. Based on the ellipsoidal distribution of the training samples, the model is trained using an SVM classification algorithm and a Softmax loss function, making the training results more accurate. By setting the fresh air system in real-time based on the predicted operating status and setting information, customer comfort can be improved and energy consumption reduced.

[0084] Optionally, the training steps of the regression prediction model include:

[0085] Obtain the training samples;

[0086] Based on the training samples, the regression prediction model is obtained by training a multiple linear regression prediction algorithm.

[0087] Optionally, in order to obtain training samples, some spaces with fresh air equipment can be selected as sample spaces, sample information of the sample space and expected setting information of the fresh air equipment corresponding to the sample information can be obtained, and a regression prediction model can be trained using the sample information and the expected setting information of the fresh air equipment using a multiple linear regression prediction algorithm.

[0088] Optionally, the desired setting information of the fresh air equipment corresponding to the sample information can be the fresh air equipment setting information that makes people in the space feel the most comfortable in the sample space environment.

[0089] For example, in a shopping mall corresponding to certain sample information, setting the frequency of the fresh air equipment to 60Hz can make people in the mall feel most comfortable. Therefore, the expected setting information of the fresh air equipment corresponding to this sample information can include setting the frequency of the fresh air equipment to 60Hz.

[0090] Optionally, based on the sample information in the sample space and the expected setting information of the fresh air equipment corresponding to the sample information, a linear regression algorithm with a targeted modification of the loss function can be used to train the regression prediction model. The training objective is to find the model parameters W, i.e., to find w1, w2, w3...wn, and to minimize the loss function, i.e., to minimize the sum of errors between all true values ​​and predicted values.

[0091] Alternatively, gradient descent can be used, which involves iterative training in the opposite direction of the gradient of the loss function at the current point, with a specified step size, to find the optimal solution.

[0092] Optionally, sample information from multiple sample spaces, along with the corresponding expected settings of the fresh air system, can be used as training samples to train a regression prediction model using a multiple linear regression prediction algorithm.

[0093] Optionally, after training the regression prediction model, the real-time information of the space where the fresh air equipment is located can be input into the regression prediction model to obtain the prediction setting information.

[0094] The method for determining the setting information of fresh air equipment provided by the present invention trains a regression prediction model through a multiple linear regression prediction algorithm and a gradient descent method, so that the fresh air equipment setting information that optimizes customer comfort is more accurate as predicted by the trained regression prediction model.

[0095] Optionally, the environmental information includes at least one of the following:

[0096] The temperature, humidity, carbon dioxide concentration, PM2.5 concentration, negative pressure, and flow of people in the space are all measured.

[0097] Optionally, since the indoor and outdoor temperatures of a space can affect the comfort of people in the space, the temperature of the space can be adjusted by adjusting the fresh air equipment to achieve the best comfort. Therefore, the indoor and outdoor temperatures of the space where the fresh air equipment is located can be obtained, and the fresh air equipment setting information predicted by the classification model and regression model can be obtained based on the temperature. The fresh air equipment can then be adjusted based on the setting information to achieve the best comfort.

[0098] Optionally, since the indoor and outdoor humidity of a space can affect the comfort of people in the space, the humidity of the space can be adjusted by adjusting the fresh air equipment to achieve the best comfort. Therefore, the indoor and outdoor humidity of the space where the fresh air equipment is located can be obtained, and the fresh air equipment setting information predicted by the classification model and regression model can be obtained based on the humidity. The fresh air equipment can be adjusted based on the setting information to achieve the best comfort.

[0099] Optionally, since the indoor and outdoor carbon dioxide concentrations of a space can affect the comfort of people in the space, the carbon dioxide concentration of the space can be adjusted by adjusting the fresh air equipment to achieve the best comfort. Therefore, the indoor and outdoor carbon dioxide concentrations of the space where the fresh air equipment is located can be obtained, and the fresh air equipment setting information predicted by the classification model and regression model can be obtained based on the carbon dioxide concentration. The fresh air equipment can be adjusted based on the setting information to achieve the best comfort.

[0100] Optionally, since the indoor and outdoor PM2.5 concentrations of a space can affect the comfort of people in the space, the PM2.5 concentration in the space can be adjusted by adjusting the fresh air equipment to achieve the best comfort. Therefore, the indoor and outdoor PM2.5 concentrations of the space where the fresh air equipment is located can be obtained, and the fresh air equipment setting information predicted by the classification model and regression model can be obtained based on the PM2.5 concentration. The fresh air equipment can be adjusted based on the setting information to achieve the best comfort.

[0101] Optionally, since the indoor negative pressure of the space can affect the comfort of the people in the space, the indoor negative pressure of the space can be adjusted by adjusting the fresh air equipment to achieve the best comfort. Therefore, the indoor negative pressure of the space where the fresh air equipment is located can be obtained, and the fresh air equipment setting information predicted by the classification model and regression model can be obtained based on the indoor negative pressure. The fresh air equipment can be adjusted based on the setting information to achieve the best comfort.

[0102] Optionally, since the indoor traffic flow of the space affects the comfort of the people in the space, and the traffic flow directly affects factors such as indoor temperature, humidity, and carbon dioxide concentration that affect the comfort of the people in the space, the fresh air equipment can be adjusted to achieve the best comfort. Therefore, the indoor traffic flow of the space where the fresh air equipment is located can be obtained, and the fresh air equipment setting information predicted by the classification model and regression model can be obtained based on the indoor negative pressure. The fresh air equipment can be adjusted based on the setting information to achieve the best comfort.

[0103] The method for determining the setting information of fresh air equipment provided by this invention predicts the setting information of fresh air equipment based on factors that affect the comfort of people in the space, such as temperature, humidity, carbon dioxide concentration, PM2.5 concentration, negative pressure, and flow of people. Since the prediction results of the regression model are affected by multiple factors in the space, the setting information of fresh air equipment predicted by the regression prediction model can be closer to the optimal setting for the comfort of people in the space, thereby achieving dynamic, optimal, energy-saving, automated, and real-time adjustment and reducing energy waste.

[0104] Optionally, the fresh air equipment includes a fresh air unit, a return air unit, and an exhaust fan, and the setting information of the fresh air equipment includes at least one of the following:

[0105] The set frequency of the fresh air unit, the set frequency of the return air unit, and the set frequency of the exhaust fan.

[0106] Optionally, the equipment in the fresh air system mainly includes a fresh air unit, a return air unit, and an exhaust fan. By adjusting the on / off state and frequency of the fresh air unit, return air unit, and exhaust fan, the indoor comfort level can be adjusted and energy consumption reduced.

[0107] Optionally, the real-time collected environmental information can be input into the classification prediction model to obtain the operating status information of multiple fresh air units, multiple return air units, and multiple exhaust fans. For example, there are 10 fresh air units, of which 4 are running and 6 are not running; there are 8 return air units, of which 2 are running and 6 are not running; and there are 6 exhaust fans, of which 3 are running and 3 are not running.

[0108] Optionally, the real-time collected environmental information can be input into the regression prediction model to obtain the prediction setting information for multiple fresh air units, multiple return air units, and multiple exhaust fans. For example, the set frequency for multiple fresh air units is 50Hz, the set frequency for multiple return air units is 60Hz, and the frequency for multiple exhaust fans is 70Hz.

[0109] The method for determining the setting information of fresh air equipment provided by the present invention can achieve the effect of automatically adjusting the indoor comfort and reducing energy consumption by automatically adjusting the on / off state and frequency of the fresh air unit, return air unit and exhaust fan.

[0110] Optionally, after obtaining the training samples, the method further includes:

[0111] The training samples are preprocessed, and the preprocessing includes processing one or more of the following:

[0112] Missing values ​​in the training samples;

[0113] Negative values ​​in the training samples;

[0114] The abnormally large values ​​in the training samples;

[0115] The abnormally small values ​​in the training samples.

[0116] Optionally, the obtained training samples may contain abnormal or missing data. In order to correct the abnormal or missing data in the collected training samples, they can be preprocessed.

[0117] Optionally, the training samples may be preprocessed in any of the following ways:

[0118] 1. Missing values ​​in the training samples;

[0119] 2. Negative values ​​in the training samples;

[0120] 3. Abnormally large values ​​in the training samples;

[0121] 4. Abnormally small values ​​in the training samples;

[0122] 5. Missing values ​​and negative values ​​in the training samples;

[0123] 6. Missing values ​​and abnormally large values ​​in the training samples;

[0124] 7. Missing values ​​and outliers in the training samples;

[0125] 8. Negative values ​​and abnormally large values ​​in the training samples;

[0126] 9. Negative values ​​and abnormally small values ​​in the training samples;

[0127] 10. Abnormally large and abnormally small values ​​in the training samples;

[0128] 11. Missing values, negative values, and abnormally large values ​​in the training samples;

[0129] 12. Missing values, negative values, and abnormally small values ​​in the training samples;

[0130] 13. Missing values, abnormally large values, and abnormally small values ​​in the training samples;

[0131] 14. Negative values, abnormally large values, and abnormally small values ​​in the training samples;

[0132] 15. Missing values, negative values, abnormally large values, and abnormally small values ​​in the training samples.

[0133] Optionally, training samples collected over a period of time can be analyzed for their temporal patterns using artificial intelligence (AI), and training samples collected after that period can be preprocessed based on these patterns.

[0134] Alternatively, linear interpolation can be performed on the missing values ​​in the training samples to handle them, for example, by taking the median value of the vector distribution of the daily time feature values ​​analyzed by AI to supplement the missing values.

[0135] Optionally, all values ​​less than 0 can be set to 0 to handle negative values ​​in the training samples.

[0136] Optionally, the criterion for judging abnormally large values ​​can be a value that is N times greater than the average value of the previous training sample data, where N is a positive integer, or a value that is greater than a preset value. This invention does not limit this.

[0137] Optionally, the criterion for judging abnormally small values ​​may be a value less than 1 / M of the average value of the previous training sample data, where M is a positive integer, or a value less than a preset value. This invention does not limit this.

[0138] Optionally, outlier values ​​can be set to Not a Number (NaN) to handle outlier values ​​in the training samples.

[0139] The method for determining the setting information of a fresh air device provided by the present invention corrects abnormal or missing data in the collected training samples by preprocessing missing values, negative values, abnormally large values ​​or abnormally small values ​​in the training samples, and then trains the model based on the preprocessed training samples, which can improve the accuracy of model prediction.

[0140] Optionally, after obtaining the training samples, the method further includes:

[0141] The training samples are normalized.

[0142] Optionally, in order to make the obtained training samples usable for model training, the training samples can be normalized.

[0143] Optionally, the on / off states of fresh air equipment, such as fresh air units, return air units, or exhaust fans, can be normalized, for example, by using 1 to represent on and 0 to represent off.

[0144] Optionally, the frequency of fresh air equipment, such as fresh air unit, return air unit or exhaust fan, can be normalized, for example by using the specific Hertz value of the frequency of each equipment type, such as 100Hz, 150Hz or 200Hz.

[0145] Optionally, the temperature of the sample space, such as the indoor or outdoor temperature of the space, can be normalized, for example, by using the specific values ​​of the indoor temperature and the outdoor temperature as the normalized temperature value of the sample space.

[0146] Optionally, the humidity of the sample space, such as the indoor or outdoor humidity of the space, can be normalized, for example, by using the specific values ​​of indoor humidity and outdoor humidity as the normalized humidity values ​​of the sample space.

[0147] Optionally, the carbon dioxide concentration of the sample space, such as the indoor or outdoor carbon dioxide concentration of the space, can be normalized. For example, the specific values ​​of the indoor and outdoor carbon dioxide concentrations can be used as the normalized carbon dioxide concentration of the sample space.

[0148] Optionally, the PM2.5 concentration in the sample space, such as the PM2.5 concentration indoors or outdoors, can be normalized. For example, the specific values ​​of indoor PM2.5 concentration and outdoor PM2.5 concentration can be used as the normalized PM2.5 concentration of the sample space.

[0149] Optionally, the negative pressure of the sample space, such as the negative pressure of the room in which the space is located, can be normalized. For example, 0 can be used to represent that the indoor negative pressure is 0, 1 can be used to represent that the indoor negative pressure is negative, and 2 can be used to represent that the indoor negative pressure is positive.

[0150] Optionally, the flow of people in the sample space, such as the flow of people in the indoor area where the space is located, can be normalized. For example, 1 represents the cumulative flow of people in the sample space in five minutes < 10 people, 2 represents the cumulative flow of people in the sample space in five minutes >= 10 people and < 30 people, and 3 represents the cumulative flow of people in the sample space in five minutes >= 30 people.

[0151] Optionally, the indoor traffic flow in the sample space can be collected by a camera, obtained through video facial recognition, obtained manually, or by other means; the present invention does not limit this.

[0152] In one embodiment of the present invention, the space where the fresh air equipment is located, such as a shopping mall or a hotel, can be divided into multiple areas. One or more fresh air equipment can be installed in each area. Each area can respectively obtain the real-time environmental information of each area, including temperature, humidity, carbon dioxide concentration, PM2.5 concentration, negative pressure and human flow, and respectively input the real-time environmental information of each area into the classification prediction model trained by the SVM classification algorithm and the regression prediction model trained by the multiple linear regression prediction algorithm, and obtain the predicted operation status information and predicted setting information of each area output by the two regression prediction models, and the operation status information and setting information of the fresh air equipment in each area can be determined. Based on the operation status information and setting information of the fresh air equipment in each area, one or more fresh air equipment in each area, including fresh air fans, return air fans and exhaust fans, can be turned on, turned off, and the frequency can be adjusted to improve the comfort of customers in this area.

[0153] Figure 3 is a schematic diagram of the automatic control process of the fresh air system provided by the present invention. As Figure 3 shown, in one embodiment of the present invention, the automatic control of the fresh air system is realized through the Lianyuan governor, the Lianyuan data acquisition platform, the Lianyuan AI service platform, the Lianyuan AI cloud platform and the Internet of Things platform.

[0154] The method for determining the setting information of the fresh air equipment provided by the present invention can better be used for model training by normalizing the training samples, and make the setting information of the fresh air equipment predicted by the model more accurate.

[0155] Figure 4 is a schematic structural diagram of the device for determining the setting information of the fresh air equipment provided by the present invention, including an acquisition module 410, a first input module 420, a second input module 430 and a determination module 440, wherein:

[0156] The acquisition module 410 is used to acquire the real-time information of the space where the fresh air equipment is located, and the real-time information at least includes environmental information;

[0157] The first input module 420 is used to input the real-time information into the classification prediction model to obtain the predicted operation status information of the fresh air equipment output by the classification prediction model, and the operation status information includes operating the fresh air equipment or not operating the fresh air equipment;

[0158] The second determination module 430 is used to input the real-time information into the regression prediction model to obtain the predicted setting information of the fresh air equipment output by the regression prediction model;

[0159] The determination module 440 is used to determine the setting information of the fresh air equipment based on the predicted operation status information and the predicted setting information.

[0160] The training steps of the classification prediction model include:

[0161] Acquire training samples, which include sample information of the sample space and expected setting information of the fresh air equipment corresponding to the sample information, wherein the sample information includes at least environmental information;

[0162] The training samples are mapped onto a high-dimensional ellipsoid, and the side length of the triangle on the ellipsoid is calculated using the cosine function as a distance metric, so that the training samples are distributed on an ellipsoid.

[0163] Based on the training samples distributed in an elliptic shape, the classification prediction model is obtained by training the Support Vector Machine (SVM) classification algorithm and the Softmax loss function.

[0164] The device for determining the setting information of a fresh air system provided by this invention acquires real-time information about the space where the fresh air system is located. It inputs the environmental information of that space into two classification prediction models and a regression prediction model. The two regression prediction models then predict the optimal operating status and setting information of the fresh air system for customer comfort. The classification prediction model maps its training onto a high-dimensional ellipsoid, resulting in better intra-class compactness and inter-class diversity of the training sample features in the metric space. Based on the ellipsoidal distribution of the training samples, it is trained using an SVM classification algorithm and a Softmax loss function, making the model's training results more accurate. By setting the fresh air system in real-time based on the predicted operating status and setting information, customer comfort can be improved and energy consumption reduced.

[0165] It is understood that the setting information determination device for fresh air equipment provided by the present invention corresponds to the setting information determination method for fresh air equipment provided in the above embodiments. The relevant technical features of the setting information determination device for fresh air equipment provided by the present invention can be referred to the relevant technical features of the setting information determination method for fresh air equipment provided in the above embodiments, and will not be repeated here.

[0166] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call logical instructions in the memory 530 to execute a method for determining the setting information of a fresh air device. This method includes: acquiring real-time information about the space where the fresh air device is located, the real-time information including at least environmental information; inputting the real-time information into a classification prediction model to obtain predicted operating status information of the fresh air device output by the classification prediction model, the predicted operating status information including whether the fresh air device is running or not; inputting the real-time information into a regression prediction model to obtain predicted setting information of the fresh air device output by the regression prediction model; and determining the setting information of the fresh air device based on the predicted operating status information and the predicted setting information. The training step of the classification prediction model includes: acquiring training samples, the training samples including sample information of a sample space and expected setting information of the fresh air device corresponding to the sample information, the sample information including at least environmental information; mapping the training samples onto a high-latitude ellipsoid, using the cosine function to calculate the side length of the triangle on the ellipsoid as a distance metric, so that the training samples are distributed on an ellipsoid; and based on the ellipsoidally distributed training samples, using a support vector machine (SVM) classification algorithm and a softmax loss function. The classification prediction model is obtained by training with loss.

[0167] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0168] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the method for determining the setting information of a fresh air device provided by the above methods. The method includes: acquiring real-time information of the space where the fresh air device is located, the real-time information including at least environmental information; inputting the real-time information into a classification prediction model to obtain the predicted operating status information of the fresh air device output by the classification prediction model, the predicted operating status information including whether the fresh air device is running or not; and inputting the real-time information into a regression prediction model to obtain the regression prediction model. The output includes predicted setting information for the fresh air system; based on the predicted operating status information and the predicted setting information, the setting information for the fresh air system is determined; wherein, the training steps of the classification prediction model include: acquiring training samples, the training samples including sample information in the sample space and the expected setting information of the fresh air system corresponding to the sample information, the sample information including at least environmental information; mapping the training samples onto a high-latitude ellipsoid, using the cosine function to calculate the side length of the triangle on the ellipsoid as a distance metric, so that the training samples are distributed on an ellipsoid; based on the training samples distributed on an ellipsoid, the classification prediction model is trained using the support vector machine (SVM) classification algorithm and the softmax loss function.

[0169] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a method for determining the setting information of a fresh air device provided by the methods described above. This method includes: acquiring real-time information about the space where the fresh air device is located, the real-time information including at least environmental information; inputting the real-time information into a classification prediction model to obtain predicted operating status information of the fresh air device output by the classification prediction model, the predicted operating status information including whether the fresh air device is running or not; and inputting the real-time information into a regression prediction model to obtain predicted setting information of the fresh air device output by the regression prediction model. Based on the predicted operating status information and the predicted setting information, the setting information of the fresh air device is determined; wherein, the training steps of the classification prediction model include: acquiring training samples, the training samples including sample information of the sample space and the expected setting information of the fresh air device corresponding to the sample information, the sample information including at least environmental information; mapping the training samples onto a high-latitude ellipsoid, using the cosine function to calculate the side length of the triangle on the ellipsoid as a distance metric, so that the training samples are distributed on an ellipsoid; based on the training samples distributed on an ellipsoid, the classification prediction model is trained using the support vector machine (SVM) classification algorithm and the softmax loss function.

[0170] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0171] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0172] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of determining setting information of a fresh air device, characterized by, The method comprises: obtaining real-time information of a space where a fresh air equipment is located, the real-time information comprising at least environmental information; inputting the real-time information into a classification prediction model to obtain predicted running state information of the fresh air equipment output by the classification prediction model, the predicted running state information comprising running the fresh air equipment or not running the fresh air equipment; inputting the real-time information into a regression prediction model to obtain predicted setting information of the fresh air equipment output by the regression prediction model; determining setting information of the fresh air equipment based on the predicted running state information and the predicted setting information; wherein a training step of the classification prediction model comprises: obtaining training samples, the training samples comprising sample information of a sample space and expected setting information of a fresh air equipment corresponding to the sample information, the sample information comprising at least environmental information; mapping the training samples to a high- latitude ellipsoidal surface, and using a cosine function to calculate the edge length of an ellipsoidal surface triangle as a distance measure, so that the training samples are distributed in an ellipsoidal surface; training the classification prediction model by using a support vector machine (SVM) classification algorithm and a loss function (Softmax loss) based on the training samples distributed in the ellipsoidal surface.

2. The method according to claim 1, wherein The training step of the regression prediction model comprises: obtaining the training samples; training the regression prediction model by using a multivariate linear regression prediction algorithm and a gradient descent method based on the training samples.

3. The method according to claim 1, wherein The environmental information comprises at least one of the following: temperature, humidity, carbon dioxide concentration, PM2.5 concentration, negative pressure, and human flow of the space.

4. The method according to claim 1, wherein The fresh air equipment comprises a fresh air fan, a return air fan, and an exhaust fan, and the setting information of the fresh air equipment comprises at least one of the following: set frequency of the fresh air fan, set frequency of the return air fan, and set frequency of the exhaust fan.

5. The method according to claim 1 or 2, wherein After obtaining the training samples, the method further comprises: preprocessing the training samples, the preprocessing comprising processing one or more of the following: missing values in the training samples; negative values in the training samples; abnormally large values in the training samples; abnormally small values in the training samples.

6. The method according to claim 1 or 2, wherein After obtaining the training samples, the method further comprises: normalizing the training samples.

7. A setting information determination device of a fresh air device, characterized by comprising: The method comprises: an obtaining module configured to obtain real-time information of a space where a fresh air equipment is located, the real-time information comprising at least environmental information; a first input module configured to input the real-time information into a classification prediction model to obtain predicted running state information of the fresh air equipment output by the classification prediction model, the running state information comprising running the fresh air equipment or not running the fresh air equipment; a second input module configured to input the real-time information into a regression prediction model to obtain predicted setting information of the fresh air equipment output by the regression prediction model; a determining module configured to determine setting information of the fresh air equipment based on the predicted running state information and the predicted setting information; wherein a training step of the classification prediction model comprises: Obtaining training samples, the training samples including sample information of a sample space and expected setting information of a fresh air equipment corresponding to the sample information, the sample information at least including environment information; Mapping the training samples to a high latitude ellipsoidal sphere, using a cosine function to calculate a side length of an ellipsoidal sphere triangle as a distance measure, so as to make the training samples be ellipsoidal sphere distributed, and obtaining; Based on the training samples being ellipsoidal sphere distributed, a classification prediction model is trained through a support vector machine (SVM) classification algorithm and a loss function Softmax loss.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the setting information determination method of the fresh air equipment according to any one of claims 1 to 6 when executing the program. 9.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the setting information determination method of the fresh air equipment according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the setting information determination method of the fresh air equipment according to any one of claims 1 to 6.

Images