Cooking fume backward flow prediction method and device

Abstract

The invention discloses an oil smoke backward flow prediction method and device, and belongs to the technical field of kitchen appliances. The method comprises the following steps: acquiring environmental parameters of a public flue; the environment parameters comprise at least two of the pressure difference between the public flue and a room where the range hood is located, the target object concentration of the public flue and the temperature of the public flue; extracting features of the various environmental parameters, and performing feature fusion on the features of the various environmental parameters based on the time sequence relevance of the various environmental parameters to obtain feature vectors; and inputting the feature vector into a trained prediction model, and determining the oil smoke backflow condition of the public flue according to a prediction result output by the prediction model. According to the embodiment of the invention, the relevance between different environmental parameters and the rule that the parameters change along with time are fully considered, so that the feature vector can more accurately reflect the trend of backward flowing of the oil smoke, the accuracy of backward flowing prediction of the oil smoke is improved, a basis can be provided for taking preventive measures in time, and pollution of backward flowing of the oil smoke to the indoor environment is reduced.

Description

Technical Field

[0001] This application belongs to the field of kitchen appliance technology, and in particular relates to a method and device for predicting backflow of cooking fumes. Background Technology

[0002] In multi-story and high-rise residential buildings, in order to achieve intensive pipeline layout and rational space utilization, the exhaust pipes of the range hoods usually need to be connected to the public flue set up by the building, so that the oil fumes generated by each household can be centrally discharged outdoors through the public flue.

[0003] However, when the range hood is not in operation, external cooking fumes can easily flow back into the room through the shared exhaust duct, causing backflow of cooking fumes. This phenomenon not only pollutes indoor air quality, affecting the freshness and comfort of the living environment, but also has a significant negative impact on the user experience. Currently, there is still no effective method to accurately predict the phenomenon of backflow of cooking fumes. Summary of the Invention

[0004] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method and apparatus for predicting backflow of cooking fumes, thereby improving the accuracy of backflow prediction.

[0005] Firstly, this application provides a method for predicting backflow of cooking fumes, including: Obtain environmental parameters of the common flue; the environmental parameters include at least two of the following: pressure difference between the common flue and the room where the exhaust fan is located, concentration of the target substance in the common flue, and temperature of the common flue; Features of various environmental parameters are extracted, and feature fusion is performed on the features of various environmental parameters based on the temporal correlation of various environmental parameters to obtain feature vectors; The feature vector is input into the trained prediction model, and the backflow of oil fumes in the public flue is determined based on the prediction results output by the prediction model.

[0006] According to the oil fume backflow prediction method of this application, environmental parameters of the public flue are obtained. These environmental parameters include at least two of the following: pressure difference between the public flue and the room where the range hood is located, target concentration in the public flue, and temperature of the public flue. Features of various environmental parameters are extracted, and feature fusion is performed based on the temporal correlation of these parameters to obtain a feature vector. The feature vector is then input into a trained prediction model, and the oil fume backflow situation in the public flue is determined based on the prediction results output by the model. This application's embodiments, through different environmental parameters, can obtain environmental features related to oil fume backflow from different dimensions. By extracting features of various environmental parameters and performing feature fusion based on their temporal correlation to obtain a feature vector, the correlation between different environmental parameters and the laws governing parameter changes over time are fully considered. This allows the feature vector to more accurately reflect the trend of oil fume backflow. The model's data analysis and prediction capabilities are used to output prediction results and determine the oil fume backflow situation, improving the accuracy of oil fume backflow prediction and providing a basis for timely preventative measures, thereby reducing the pollution caused by oil fume backflow to the indoor environment.

[0007] According to one embodiment of this application, environmental parameters of a public flue are obtained according to at least one of the following methods: Micro pressure sensors are installed in the common flue and the room where the range hood is located. The pressure difference is calculated based on the pressure of the common flue and the room where the range hood is located, based on the pressure of the common flue and the room where the range hood is located. The public flue is equipped with a target object monitoring sensor to obtain the concentration of target objects in the public flue. The target object concentration includes one or more of the following: VOCs concentration, particulate matter concentration, and fatty acid concentration. Temperature sensors are installed in the public flue to obtain the temperature of the public flue.

[0008] In this embodiment, various environmental parameters are collected using different sensors, which improves the accuracy of environmental parameter acquisition.

[0009] According to one embodiment of this application, features of various environmental parameters are extracted, including: Given environmental parameters including the pressure difference between the public flue and the room where the exhaust fan is located, the pressure difference and the rate of change of the pressure difference within the target time window are defined as the characteristics of the pressure difference. Given the target concentration in the environmental parameter common flue, the target concentration, as well as the integral and / or gradient of the target concentration within the target time window, are determined as characteristics of the target concentration. Given the environmental parameter of the temperature of the public flue, the temperature difference between the public flue and the indoor temperature is defined as the characteristic of the temperature of the public flue.

[0010] In this embodiment, by using the pressure difference and the rate of change within the target time window as features, the static pressure state and dynamic evolution trend of the pressure difference can be obtained; by determining the target concentration and the concentration integral and / or gradient within the target time window as features, where the concentration integral can quantify the degree of oil fume accumulation within a specific period, and the concentration gradient reflects the rate of change of oil fume concentration, combined with the target concentration, the accumulation state and change pattern of oil fume in the common flue can be determined; by using the temperature difference between the common flue and the indoor environment as a temperature parameter feature, the thermal difference between the common flue and the indoor environment can be reflected, which is helpful in determining the direction of oil fume flow; through the above feature extraction methods, the influence of environmental parameters on oil fume backflow can be analyzed from different dimensions, improving the accuracy of oil fume backflow prediction.

[0011] According to one embodiment of this application, feature fusion is performed on the features of various environmental parameters based on the temporal correlation of various environmental parameters to obtain a feature vector, including: Calculate the time-series correlation parameters between the characteristics of various environmental parameters; the time-series correlation parameters include at least one of the time difference of abrupt changes in various environmental parameters and the sequence correlation coefficient of the characteristics of various environmental parameters. The features of time-series correlation parameters and various environmental parameters are concatenated to obtain a feature vector.

[0012] In this embodiment, by calculating the temporal correlation parameters between various environmental parameter features, it is possible to obtain the sequential logic of abnormal changes in different parameters, the synergy and correlation of feature changes over time, so that the feature vector can characterize the precursory pattern of oil fume backflow, making the feature vector more consistent with the actual characterization of oil fume backflow, and further improving the accuracy of oil fume backflow prediction.

[0013] According to one embodiment of this application, the prediction model is trained in the following manner: A training sample set is constructed, which includes positive samples and negative samples. The negative samples are the feature vectors corresponding to the environmental parameters collected when backflow of oil fumes occurs in the public flue, and the positive samples are the feature vectors corresponding to the environmental parameters collected in non-backflow of oil fumes scenarios. The prediction model is trained using a training sample set.

[0014] In this embodiment, the prediction model is trained by constructing a training sample set containing positive and negative samples, which can cover both backflow and non-backflow scenarios of oil fume. During the training process, the prediction model can learn the differences in feature vectors between positive and negative samples, so that the trained prediction model can accurately grasp the feature distribution patterns under different scenarios and improve the prediction ability of oil fume backflow.

[0015] According to one embodiment of this application, the prediction result includes a probability value of backflow of cooking fumes; The backflow of cooking fumes in the public flue is determined based on the prediction results output by the prediction model, including: If the probability of backflow of cooking fumes exceeds the target threshold, it is determined that there is a risk of backflow of cooking fumes in the public flue.

[0016] In this embodiment, the backflow probability value of oil fume output by the prediction model is used to determine the backflow situation of oil fume in the public flue. The backflow probability value can reflect the likelihood of backflow of oil fume, further improving the accuracy of backflow prediction.

[0017] According to one embodiment of this application, the method further includes: If it is determined that there is a risk of backflow of cooking fumes into the public exhaust duct, turn on the exhaust fan.

[0018] In this embodiment, when it is determined that there is a risk of backflow of oil fumes in the public flue, backflow prevention control is implemented by turning on the range hood and activating its exhaust function to enhance the exhaust of indoor air, thereby reducing the occurrence of backflow of oil fumes.

[0019] According to one embodiment of this application, the method further includes: Receive feedback from users regarding the prediction results; the feedback includes valid predictions and invalid predictions. If the feedback information is a valid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a positive sample; if the feedback information is an invalid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a negative sample. The prediction model is incrementally trained using an incremental training set.

[0020] In this embodiment, user feedback information includes two cases: valid prediction and invalid prediction. When the feedback information is a valid prediction, it means that the output of the prediction model matches the actual situation. The corresponding environmental parameter feature vector is added as a positive sample to the incremental training set, which can enhance the model's ability to predict features of similar scenarios. When the feedback information is an invalid prediction, it means that the prediction result does not match the actual situation. The corresponding environmental parameter feature vector is added as a negative sample to the incremental training set, which can specifically correct the model's judgment bias, so as to continuously adjust and optimize its own parameters and structure to better adapt to the user's specific home environment.

[0021] According to one embodiment of this application, the method further includes: Statistics on the backflow of cooking fumes into public exhaust ducts at different times of the day; The same period of time during which there is a risk of backflow of cooking fumes for N consecutive days is defined as the period when backflow is prone to occur. Turn on the smoke generator when the current time is within the target duration of the period when backflow is likely to occur.

[0022] In this embodiment, by statistically analyzing the backflow situation at different times of the day, the occurrence pattern of backflow can be extracted from historical data. By turning on the range hood in advance when the current time is a target time away from the backflow-prone period, a stable negative pressure environment can be established between the indoor and public flues before the risk of backflow occurs, thus reducing the risk of oil fume backflow.

[0023] Secondly, this application provides a device for predicting backflow of cooking fumes, comprising: The acquisition module is used to acquire environmental parameters of the common flue; the environmental parameters include at least two of the following: pressure difference between the common flue and the room where the smoke machine is located, concentration of the target substance in the common flue, and temperature of the common flue; The extraction module is used to extract features of various environmental parameters and perform feature fusion on the features of various environmental parameters based on the temporal correlation of various environmental parameters to obtain feature vectors; The prediction module is used to input feature vectors into a trained prediction model and determine the backflow of fumes in the public flue based on the prediction results output by the prediction model.

[0024] According to the oil fume backflow prediction device of this application, environmental parameters of a common flue are acquired. These environmental parameters include at least two of the following: pressure difference between the common flue and the room where the range hood is located, target concentration in the common flue, and temperature of the common flue. Features of various environmental parameters are extracted, and feature fusion is performed based on the temporal correlation of these parameters to obtain a feature vector. The feature vector is then input into a trained prediction model, and the oil fume backflow situation in the common flue is determined based on the prediction results output by the model. This embodiment of the application can acquire environmental features related to oil fume backflow from different dimensions through different environmental parameters. By extracting features of various environmental parameters and performing feature fusion based on the temporal correlation of these parameters to obtain a feature vector, the correlation between different environmental parameters and the laws governing parameter changes over time are fully considered. This allows the feature vector to more accurately reflect the trend of oil fume backflow. By utilizing the model's data analysis and prediction capabilities to output prediction results and determine the oil fume backflow situation, the accuracy of oil fume backflow prediction is improved. This provides a basis for timely preventive measures and reduces the pollution caused by oil fume backflow to the indoor environment.

[0025] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the oil fume backflow prediction method as described in the first aspect above.

[0026] Fourthly, this application provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the oil fume backflow prediction method as described in the first aspect above.

[0027] The above-described one or more technical solutions in the embodiments of this application have at least one of the following technical effects: According to the oil fume backflow prediction method of this application, environmental parameters of the public flue are obtained. These environmental parameters include at least two of the following: pressure difference between the public flue and the room where the range hood is located, target concentration in the public flue, and temperature of the public flue. Features of various environmental parameters are extracted, and feature fusion is performed based on the temporal correlation of these parameters to obtain a feature vector. The feature vector is then input into a trained prediction model, and the oil fume backflow situation in the public flue is determined based on the prediction results output by the model. This application's embodiments, through different environmental parameters, can obtain environmental features related to oil fume backflow from different dimensions. By extracting features of various environmental parameters and performing feature fusion based on their temporal correlation to obtain a feature vector, the correlation between different environmental parameters and the laws governing parameter changes over time are fully considered. This allows the feature vector to more accurately reflect the trend of oil fume backflow. The model's data analysis and prediction capabilities are used to output prediction results and determine the oil fume backflow situation, improving the accuracy of oil fume backflow prediction and providing a basis for timely preventative measures, thereby reducing the pollution caused by oil fume backflow to the indoor environment.

[0028] In some embodiments, various environmental parameters are collected using different sensors, which improves the accuracy of environmental parameter acquisition.

[0029] In some embodiments, by using pressure difference and the rate of change within a target time window as features, the static pressure state and dynamic evolution trend of pressure difference can be obtained; by determining the concentration of the target substance and the concentration integral and / or gradient within the target time window as features, where the concentration integral can quantify the degree of oil fume accumulation within a specific period, and the concentration gradient reflects the rate of change of oil fume concentration, combined with the concentration of the target substance, the accumulation state and change pattern of oil fume in the common flue can be determined; by using the temperature difference between the common flue and the indoor environment as a temperature parameter feature, the thermal difference between the common flue and the indoor environment can be reflected, which is helpful in determining the direction of oil fume flow; through the above feature extraction methods, the influence of environmental parameters on oil fume backflow can be analyzed from different dimensions, improving the accuracy of oil fume backflow prediction.

[0030] In some embodiments, by calculating the temporal correlation parameters between various environmental parameter features, it is possible to obtain the sequential logic of abnormal changes in different parameters, the synergy and correlation of feature changes over time, so that the feature vector can characterize the precursor patterns of oil fume backflow, making the feature vector more consistent with the actual characterization of oil fume backflow, and further improving the accuracy of oil fume backflow prediction.

[0031] In some embodiments, the prediction model is trained by constructing a training sample set containing positive and negative samples, which can cover both backflow and non-backflow scenarios of oil fume. During the training process, the prediction model can learn the differences in feature vectors between positive and negative samples, so that the trained prediction model can accurately grasp the feature distribution patterns under different scenarios and improve the prediction ability of oil fume backflow.

[0032] In some embodiments, the backflow probability value of oil fume output by the prediction model is used to determine the backflow situation of oil fume in the public flue. The backflow probability value can reflect the likelihood of backflow of oil fume, further improving the accuracy of backflow prediction.

[0033] In some embodiments, when a risk of backflow of cooking fumes is identified in the public flue, backflow prevention control is implemented by turning on the range hood and activating its exhaust function to enhance the exhaust of indoor air, thereby reducing the occurrence of backflow of cooking fumes.

[0034] In some embodiments, user feedback information includes two cases: valid prediction and invalid prediction. When the feedback information is a valid prediction, it means that the output of the prediction model matches the actual situation. Adding the corresponding environmental parameter feature vector as a positive sample to the incremental training set can enhance the model's ability to predict features of similar scenarios. When the feedback information is an invalid prediction, it means that the prediction result does not match the actual situation. Adding the corresponding environmental parameter feature vector as a negative sample to the incremental training set can specifically correct the model's judgment bias, enabling continuous adjustment and optimization of its parameters and structure to better adapt to the user's specific home environment.

[0035] In some embodiments, by statistically analyzing backflow at different times of the day, the occurrence pattern of backflow can be identified from historical data. By turning on the range hood in advance when the current time is a target time away from the backflow-prone period, a stable negative pressure environment can be established between the indoor and public flues before the risk of backflow occurs, thus reducing the risk of oil fume backflow.

[0036] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0037] The above and / or additional aspects and advantages of this application will become apparent and readily understood in conjunction with the following description of the embodiments in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the oil fume backflow prediction method provided in the embodiments of this application; Figure 2 This is a schematic diagram illustrating the relationship between the pressure difference between the public flue and the room where the range hood is located and the backflow of cooking fumes, provided in an embodiment of this application. Figure 3 This is a schematic diagram illustrating the relationship between the concentration of the target substance in the public flue and the backflow of cooking fumes, provided in an embodiment of this application. Figure 4 This is a schematic diagram illustrating the relationship between the temperature of the public flue and the backflow of cooking fumes, provided in an embodiment of this application. Figure 5 This is a schematic diagram of the structure of the oil fume backflow prediction device provided in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0038] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0039] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0040] The method, apparatus, and electronic equipment for predicting backflow of oily fumes provided in this application will be described in detail below with reference to the accompanying drawings and through specific embodiments and application scenarios.

[0041] The oil fume backflow prediction method provided in this application embodiment can be executed by an electronic device or a functional module or entity in an electronic device that can implement the oil fume backflow prediction method. For example, the electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, servers, etc. The oil fume backflow prediction method provided in this application embodiment is described below using an electronic device as the execution subject.

[0042] A range hood, also known as a kitchen exhaust hood, is an essential piece of equipment in the kitchen used to remove cooking fumes and exhaust gases, keeping the kitchen air fresh. A range hood typically includes components such as a fan system, exhaust duct, control panel, air inlet, and air outlet. The air inlet is where the range hood absorbs cooking fumes and exhaust gases; it is usually located at the bottom or side of the range hood, close to the stove, to capture the fumes produced during cooking. The air outlet is located at the top or rear of the range hood, expelling the absorbed fumes and exhaust gases outdoors through the exhaust duct. The fan system, as the core power component of the range hood, consists of a motor and an impeller. When the motor drives the impeller to rotate at high speed, a negative pressure area is created inside the fan. This pressure difference draws the cooking fumes into the machine and then exhausts them outdoors through the air outlet and exhaust duct. The control panel allows users to interact with the range hood. Users can control functions such as power on / off, fan speed adjustment, and lighting through physical buttons or a touch interface on the control panel.

[0043] A shared flue is a ventilation duct system in multi-story or high-rise buildings designed to collect and centrally discharge cooking fumes from the kitchen exhaust fans of each household. It ensures unobstructed exhaust from multiple kitchens and prevents fumes from spreading within the building. Shared flues are typically constructed of fire-resistant reinforced concrete or specialized fire-resistant panels. Some shared flues also include baffles, check valves, or auxiliary ventilation devices on specific floors to optimize airflow. When a household turns on its exhaust fan, the fumes enter the shared flue at a certain velocity and pressure, mixing with the existing airflow to form a complex airflow. When multiple households turn on their exhaust fans simultaneously, the airflows from each fan interact within the shared flue, potentially causing localized pressure increases and airflow turbulence. If the exhaust fan's suction pressure is insufficient to overcome the resistance and back pressure within the flue, backflow of fumes can occur, meaning that the fumes in the shared flue, under pressure differential, flow back into the individual kitchens through the exhaust fan's outlet.

[0044] Backflow of cooking fumes not only pollutes indoor air quality, affecting the freshness and comfort of the living environment, but also has a significant negative impact on user experience. Currently, the most common prevention measure is to install check valves, which use a one-way opening structure to prevent airflow from flowing backward. However, due to limitations in valve structure, sealing performance, and the accumulation of grease after long-term use, traditional check valves still cannot achieve a complete seal in actual use, resulting in a small amount of cooking fume leakage, and the anti-backflow effect is not ideal. Currently, there is still no effective method to accurately predict the phenomenon of cooking fume backflow.

[0045] like Figure 1 As shown, the oil fume backflow prediction method of this application embodiment includes steps 110, 120 and 130.

[0046] Step 110: Obtain environmental parameters of the common flue; the environmental parameters include at least two of the following: the pressure difference between the common flue and the room where the exhaust fan is located, the concentration of the target substance in the common flue, and the temperature of the common flue.

[0047] In this embodiment of the application, the environmental parameters are parameters related to the environment of the public flue, such as the pressure difference between the public flue and the room where the exhaust fan is located, the concentration of the target substance in the public flue, and the temperature of the public flue.

[0048] Pressure difference is a physical quantity that directly reflects the direction of oil fume flow. Pressure difference = kitchen indoor pressure - pressure in the common exhaust duct. According to fluid mechanics principles, the direction of oil fume flow is determined by the pressure gradient. When the range hood is working normally, the negative pressure generated by the power system makes the kitchen indoor pressure slightly higher than the pressure in the common exhaust duct. At this time, the pressure difference is positive, and oil fumes flow from the kitchen through the range hood and exhaust pipes into the common exhaust duct under the influence of this pressure difference. However, when the common exhaust duct experiences airflow congestion due to multiple households simultaneously ventilating, or when local high pressure is caused by structural design flaws in the duct, the pressure in the common exhaust duct will rise and exceed the kitchen indoor pressure. At this time, the pressure difference becomes negative, and oil fumes, driven by the reverse pressure difference, will flow back into the kitchen from the common exhaust duct, forming backflow of oil fumes. Therefore, collecting pressure difference data can help analyze the changing trend of oil fume flow direction, thus aiding in assessing the risk of backflow.

[0049] In some embodiments, micro pressure sensors can be installed in the common flue and the room where the range hood is located, respectively. The pressure difference is calculated based on the pressure of the common flue and the room where the range hood is located, based on the pressure detected by the micro pressure sensors.

[0050] like Figure 2 As shown in one example, a micro-pressure sensor detects the air pressure P1 in the common flue and the indoor air pressure P2, and calculates the pressure difference between the flue and the indoor space. If the pressure difference is greater than or equal to 0, it is determined that there is no backflow phenomenon. If the pressure difference is less than 0, it is determined that backflow is suspected.

[0051] The target concentration is the concentration of substances related to cooking fumes, such as VOCs concentration, particulate matter concentration, fatty acid concentration, etc.

[0052] Among them, the concentration of VOCs (Volatile Organic Compounds) in public flues is the main chemical component of cooking fumes, mainly originating from the organic compounds such as hydrocarbons, aldehydes, and ketones produced by the thermal decomposition and volatilization of food oils during cooking. The VOCs concentration in public flues is positively correlated with the amount of cooking fumes emitted; when the amount of cooking fumes emitted increases, the VOCs concentration will rise accordingly; when the amount of cooking fumes emitted decreases or the flue is well-ventilated, the VOCs concentration will decrease.

[0053] Because cooking fumes carry a large amount of inhalable particulate matter during their formation and emission, the greater the amount of cooking fumes accumulated in the flue, the more obvious the upward trend in particulate matter concentration.

[0054] Fatty acids mainly originate from the high-temperature cracking reaction of oils and fats, and changes in fatty acid concentration can reflect the degree of pollution from cooking fumes.

[0055] Before backflow of cooking fumes occurs, high concentrations of cooking fumes in the public flue will accumulate at the flue interface, causing the concentrations of target substances at the corresponding locations, such as VOCs, particulate matter, and fatty acids, to show an abnormal upward trend. Therefore, VOCs concentration, particulate matter concentration, and fatty acid concentration can indirectly reflect the degree of cooking fume accumulation and backflow risk in the public flue.

[0056] Of course, the target concentration can also be the concentration of other substances related to cooking fumes, and this application does not limit this.

[0057] In some embodiments, a target substance monitoring sensor is installed in the public flue to obtain the concentration of the target substance in the public flue. Depending on the concentration of the target substance to be monitored, the target substance monitoring sensor may include different types of sensors. For example, if it is necessary to monitor the concentration of VOCs, the target substance monitoring sensor includes a VOCs sensor to obtain the VOCs concentration of the public flue.

[0058] If monitoring particulate matter concentration is required, the target analyte monitoring sensor includes particulate matter sensors such as light scattering sensors and optical particle counting sensors, used to acquire the concentrations of PM2.5, PM10, etc. If monitoring fatty acid concentration is required, the target analyte monitoring sensor includes fatty acid electrochemical sensors; if monitoring other target analyte concentrations is required, other types of sensors are used. Figure 3 As shown, when the concentration of the target substance in the public flue rises sharply and exceeds the set threshold (which is determined by actual experiments), it is determined that there may be oil fumes in the backflow gas, and it is suspected to be backflow with an oily odor.

[0059] The temperature of a shared flue is a physical quantity reflecting the energy state of the oil fume airflow within it. Since oil fumes are a product of high-temperature cooking, their temperature is typically higher than ambient temperature. Therefore, temperature changes within the shared flue are directly related to the amount of oil fume emitted. When a large amount of high-temperature oil fume enters the shared flue, the temperature rises rapidly; conversely, when the amount of oil fume emitted decreases or the airflow velocity increases, the temperature gradually decreases. Furthermore, temperature changes also affect the density and pressure characteristics of the gas within the shared flue. According to the ideal gas law, in a flue with a fixed volume, an increase in temperature leads to an increase in gas pressure, and this pressure increase is one of the causes of oil fume backflow. Therefore, by collecting temperature parameters from the shared flue, we can not only indirectly determine the oil fume load within the flue but also analyze the influence of temperature on the pressure within the flue in conjunction with pressure difference parameters, further improving the accuracy of oil fume backflow prediction.

[0060] In some embodiments, a temperature sensor is installed in the common flue to obtain the temperature of the common flue. For example... Figure 4 As shown, if the temperature of the public flue is higher than the indoor ambient temperature for a short period of time, it indicates that there is residual heat in the flue, and it is suspected that the residual heat is flowing back in.

[0061] In this embodiment, various environmental parameters are collected using different sensors, which improves the accuracy of environmental parameter acquisition.

[0062] It should be noted that at least two of the environmental parameters obtained—the pressure difference between the public flue and the room where the range hood is located, the concentration of the target substance in the public flue, and the temperature of the public flue—are required. This is because a single parameter can only reflect one dimension of the characteristics related to backflow of oil fumes, while a combination of multiple parameters can achieve complementary characteristics. For example, the pressure difference parameter directly reflects the flow direction, but it may be misjudged due to instantaneous airflow disturbances. Combining it with abnormal changes in the concentration of the target substance can verify whether the pressure anomaly is caused by actual oil fume accumulation. The temperature parameter can help determine whether the cause of the pressure change is an increase in oil fume load or fluctuations in ambient temperature.

[0063] Step 120: Extract the features of various environmental parameters, and perform feature fusion on the features of various environmental parameters based on the temporal correlation of various environmental parameters to obtain feature vectors.

[0064] In this embodiment, the characteristics of environmental parameters are indicators extracted from the original environmental parameters that reflect the essential attributes of the parameters and their correlation with backflow of cooking fumes. The characteristics of environmental parameters may include basic statistical characteristics, trend characteristics, and abrupt change characteristics. Basic statistical characteristics are statistics on the parameter values ​​within a specific time window, including but not limited to mean, standard deviation, maximum and minimum values; trend characteristics are indicators reflecting the direction and rate of change of the parameter over time, including linear trend coefficients and sliding window growth rates; abrupt change characteristics are indicators identifying sudden changes in parameter values, including the magnitude of the change at the time of the change and the duration of the change.

[0065] Different types of environmental parameters correspond to different characteristics. For example, the characteristics of pressure difference can include the average value, fluctuation range, and rate of change over a period of time; the characteristics of target concentration can include the peak value, duration, and trend of concentration; and the characteristics of temperature can include the highest temperature, lowest temperature, average temperature, temperature difference with the room, and slope of temperature change.

[0066] After extracting features from various environmental parameters, this application further considers the temporal correlation between these features to achieve feature fusion. Temporal correlation refers to the interrelationship and influence of different environmental parameters over time. For example, changes in pressure difference may affect the distribution of target substance concentration, while changes in temperature may be correlated with changes in both pressure difference and target substance concentration. Therefore, fusion techniques are needed to uncover the correlation patterns of different parameter features over time, forming a feature vector that comprehensively reflects the synergistic effects of multiple parameters.

[0067] In some embodiments, the features of various extracted environmental parameters can be arranged in chronological order to form a multi-dimensional time series dataset. Then, by constructing a time series model, such as an autoregressive moving average model or a long short-term memory network, the correlation and dependency of different features over time can be analyzed. Through the time series model, the temporal correlation between different features can be captured, and the features and temporal correlations of various environmental parameters can be fused into a comprehensive high-dimensional feature vector. This feature vector contains both the independent features of various environmental parameters and incorporates the temporal correlation patterns between parameters.

[0068] To better understand the composition of feature vectors, the following example of the feature vector obtained after fusion is given using actual data from a specific monitoring time.

[0069]

[0070] in, This represents the feature vector, consisting of three elements in dimensions 1-3. The characteristics of the pressure difference are represented by the average value, fluctuation, and rate of change of the pressure difference between the public flue and the indoor space; the three elements of dimensions 4-6. The characteristics of the target analyte concentration are represented by the maximum concentration, duration of high concentration, and rate of change; the three elements in dimensions 7-9. For temperature characteristics, the three elements represent the average temperature, maximum temperature, and rate of change within the common flue; the three elements of the 10th-12th dimensions. The temporal correlation characteristics of different environmental parameters are represented by the correlation coefficients between pressure difference and target concentration, temperature and target concentration, and pressure difference relative to target concentration, respectively.

[0071] Step 130: Input the feature vector into the trained prediction model, and determine the backflow of oil fumes in the public flue based on the prediction results output by the prediction model.

[0072] In this embodiment, the prediction model is a mathematical model built based on machine learning or deep learning algorithms. It is capable of classifying or regressing the newly input feature vector by learning the correlation between historical feature data and corresponding backflow results of cooking fumes. For example, the prediction model can adopt a Support Vector Machine (SVM), Random Forest, or Neural Network architecture, or a hybrid neural network architecture of Convolutional Neural Network (CNN) + Gated Recurrent Unit (GRU).

[0073] In some embodiments, the prediction model may also be a lightweight model suitable for embedded devices, such as a small fully connected neural network. The architecture of the prediction model is not limited in the embodiments of this application.

[0074] The prediction model is pre-trained on a large amount of historical data, including feature vectors of different environmental parameters and the corresponding actual occurrences of backflow of cooking fumes. During training, the prediction model learns the complex mapping relationship between feature vectors and backflow of cooking fumes, thus enabling it to accurately predict the probability of backflow of cooking fumes when given new feature vector inputs.

[0075] After the feature vectors are input into the trained prediction model, the feature enhancement layer of the prediction model performs convolution operations on the feature vectors to extract spatial correlation information of different dimensions in the feature vectors. For example, it correlates the pressure difference trend with the abrupt change in target concentration to generate composite features reflecting the coordinated changes of parameters. Then, it is downsampled by the pooling layer to reduce computational redundancy. The output of the feature enhancement layer is fed into the feature fusion and inference layer of the prediction model. The feature fusion and inference layer can capture the long-term and short-term temporal dependencies in the features. For example, it can identify the impact of temperature changes on the current backflow risk and achieve deep fusion of spatial and temporal features. The fused feature representation is then transformed into low-dimensional features adapted to the prediction task through nonlinear mapping by the fully connected layer.

[0076] The low-dimensional features output by the fully connected layer are fed into the output layer of the prediction model. The output layer can use an activation function, such as the Sigmoid function, to output a backflow risk probability value between 0 and 1 as the prediction result. The larger the value, the higher the backflow risk.

[0077] According to the oil fume backflow prediction method of this application, environmental parameters of the public flue are obtained. These environmental parameters include at least two of the following: pressure difference between the public flue and the room where the range hood is located, target concentration in the public flue, and temperature of the public flue. Features of various environmental parameters are extracted, and feature fusion is performed based on the temporal correlation of these parameters to obtain a feature vector. The feature vector is then input into a trained prediction model, and the oil fume backflow situation in the public flue is determined based on the prediction results output by the model. This application's embodiments, through different environmental parameters, can obtain environmental features related to oil fume backflow from different dimensions. By extracting features of various environmental parameters and performing feature fusion based on their temporal correlation to obtain a feature vector, the correlation between different environmental parameters and the laws governing parameter changes over time are fully considered. This allows the feature vector to more accurately reflect the trend of oil fume backflow. The model's data analysis and prediction capabilities are used to output prediction results and determine the oil fume backflow situation, improving the accuracy of oil fume backflow prediction and providing a basis for timely preventative measures, thereby reducing the pollution caused by oil fume backflow to the indoor environment.

[0078] In some embodiments, features of various environmental parameters are extracted, including: Given environmental parameters including the pressure difference between the public flue and the room where the exhaust fan is located, the pressure difference and the rate of change of the pressure difference within the target time window are defined as the characteristics of the pressure difference. Given the target concentration in the environmental parameter common flue, the target concentration, as well as the integral and / or gradient of the target concentration within the target time window, are determined as characteristics of the target concentration. Given the environmental parameter of the temperature of the public flue, the temperature difference between the public flue and the indoor temperature is defined as the characteristic of the temperature of the public flue.

[0079] In this embodiment, the feature extraction of the pressure difference between the public flue and the room where the smoke machine is located can be achieved by determining the instantaneous value of the pressure difference as a feature, or by determining the rate of change of the pressure difference within a target time window as a feature.

[0080] Specifically, the rate of change of pressure difference within the target time window can be represented by the first derivative of pressure difference over time, or it can be calculated using the formula: Rate of change = (Pressure difference at the end of the window - Pressure difference at the beginning of the window) / Window length. When the rate of change is positive, it indicates that the pressure in the public flue is increasing relative to the indoor pressure, and even if the current instantaneous value is negative, the risk of backflow is continuously increasing. When the rate of change is negative and the absolute value is large, it indicates that the pressure difference between the indoor and flue is widening, and the smoke exhaust environment is stabilizing. Combining the instantaneous value and the rate of change as a characteristic of pressure difference allows for a dual representation of the current risk status and future risk trends.

[0081] In some embodiments, the target time window can be 3s, 10s, 1min, etc., which can be preset as needed or dynamically determined according to the current time. For example, during periods of frequent cooking activities, the target time window can be set shorter in order to reflect the rapid changes in pressure difference more promptly; while during periods of relatively little cooking activity, the target time window can be appropriately extended.

[0082] For feature extraction of target concentration in public flues, the instantaneous value of the target concentration can be determined as a feature, or the integral and / or gradient of the target concentration within the target time window can be determined as a feature of the target concentration.

[0083] The integral of the target substance concentration within the target time window is the area under the target substance concentration-time curve within that window. Physically, it represents the total cumulative amount of cooking fumes in the flue during the window period, reflecting the intensity of continuous fume emissions. For example, during peak dinner hours, multiple households cooking together can cause the target substance concentration integral to rise rapidly, and this integral reflects the cumulative effect of cooking fumes in the flue.

[0084] The gradient of the target substance concentration within the target time window, i.e., the rate of change of the target substance concentration over time, can be used to determine the abrupt change characteristics of the concentration. When the target substance concentration gradient suddenly increases at a certain moment, it indicates that a large amount of oily fumes have rushed into the public flue in a short period of time. This abrupt change can easily cause airflow turbulence in the flue, leading to a local pressure increase and potentially causing oily fume backflow. In practical applications, integral or gradient features can be used alone, or both, depending on the required prediction accuracy.

[0085] For the characteristic extraction of temperature in a shared flue, the temperature difference between the shared flue and the indoor temperature can be defined as the characteristic of the shared flue's temperature. The temperature of a shared flue is easily affected by factors such as outdoor ambient temperature and heat dissipation from the flue structure. For example, in winter when outdoor temperatures are low, the flue temperature may be generally low, and even with a large volume of cooking fumes, the temperature may not differ significantly from periods without fumes. Defining the temperature difference between the shared flue and the indoor temperature as the characteristic of the shared flue's temperature can reduce the interference of ambient temperature and reflect the temperature increase caused by cooking fume emissions. The indoor temperature can be obtained through a temperature sensor built into the range hood or through a temperature sensor installed indoors.

[0086] In this embodiment, by using the pressure difference and the rate of change within the target time window as features, the static pressure state and dynamic evolution trend of the pressure difference can be obtained; by determining the target concentration and the concentration integral and / or gradient within the target time window as features, where the concentration integral can quantify the degree of oil fume accumulation within a specific period, and the concentration gradient reflects the rate of change of oil fume concentration, combined with the target concentration, the accumulation state and change pattern of oil fume in the common flue can be determined; by using the temperature difference between the common flue and the indoor environment as a temperature parameter feature, the thermal difference between the common flue and the indoor environment can be reflected, which is helpful in determining the direction of oil fume flow; through the above feature extraction methods, the influence of environmental parameters on oil fume backflow can be analyzed from different dimensions, improving the accuracy of oil fume backflow prediction.

[0087] In some embodiments, feature fusion is performed on the features of various environmental parameters based on the temporal correlation of various environmental parameters to obtain a feature vector, including: Calculate the time-series correlation parameters between the characteristics of various environmental parameters; the time-series correlation parameters include at least one of the time difference of abrupt changes in various environmental parameters and the sequence correlation coefficient of the characteristics of various environmental parameters. The features of time-series correlation parameters and various environmental parameters are concatenated to obtain a feature vector.

[0088] The occurrence of backflow of cooking fumes is usually related to the temporal changes of parameters such as pressure difference, target substance concentration, and temperature. For example, a continuous increase in the temperature of a public flue precedes the accumulation of target substance concentration, and a sudden change in target substance concentration may induce abnormal fluctuations in pressure difference. Therefore, simply superimposing the characteristics of various environmental parameters cannot capture the correlation between them. Thus, the temporal correlation of various environmental parameters needs to be considered during feature fusion.

[0089] Specifically, the first step is to calculate the temporal correlation parameters between the features of various environmental parameters. Temporal correlation parameters are indicators that quantify the temporal dependencies between the features of various environmental parameters, such as the time difference of abrupt changes in various environmental parameters and the sequence correlation coefficients of the features of various environmental parameters. The time difference between abrupt changes in various environmental parameters refers to the time interval between the occurrence of abrupt changes in the characteristics of different environmental parameters. This allows us to understand the relative temporal order of changes in different environmental parameters, thereby analyzing whether abrupt changes in various environmental parameters are related to backflow of cooking fumes. For example, if the abrupt change in temperature occurs 20 seconds earlier than the abrupt change in target substance concentration, and the abrupt change in target substance concentration occurs 15 seconds earlier than the abrupt change in pressure difference, this combination of time differences is consistent with the precursors of increased risk of backflow of cooking fumes. If the abrupt change in pressure difference precedes the abrupt changes in temperature and target substance concentration, it may be caused by other factors such as abnormal flue structure, rather than by the risk of backflow of cooking fumes.

[0090] In this embodiment, abrupt changes in the feature sequences of each environmental parameter can be detected. For example, abrupt changes can be identified by calculating the first or second derivative of the feature sequences. When the absolute value of the derivative exceeds a preset threshold, a mutation is considered to have occurred. Then, the timestamps of the abrupt changes in the feature sequences of different environmental parameters are compared to calculate the time difference.

[0091] The serial correlation coefficient of various environmental parameters is a statistical indicator that measures the degree of linear correlation between time series of environmental parameter characteristics. The serial correlation coefficient ranges from [-1, 1]. A positive value indicates that the two types of characteristics are positively correlated, a negative value indicates that they are negatively correlated, and the larger the absolute value, the higher the degree of correlation.

[0092] The role of the serial correlation coefficient is to quantify the synergistic effect of various parameter characteristics over time. For example, during the stage of increased risk of backflow of cooking fumes, the characteristic sequence of target concentration and the characteristic sequence of pressure difference are usually positively correlated (an increase in target concentration is accompanied by an increase in pressure difference towards a positive value), and the characteristic sequence of temperature difference is also positively correlated with the characteristic sequence of target concentration (an increase in temperature difference reflects an increase in cooking fume load, which in turn leads to an increase in target concentration). During the normal exhaust stage, the characteristic sequence of pressure difference and the characteristic sequence of target concentration may be negatively correlated (when the target concentration increases, the suction effect of the exhaust fan is enhanced, and the negative value of the pressure difference increases). The serial correlation coefficient can help distinguish between normal conditions and the risk of backflow of cooking fumes.

[0093] In some embodiments, for any two environmental parameter feature sequences, standardization can first be performed to eliminate the influence of dimensions and magnitudes. Then, the covariance and standard deviation between the standardized sequences are calculated to obtain their correlation coefficient, which is the time-series correlation parameter between the features of the two environmental parameters. Of course, other methods can also be used to calculate the time-series correlation parameter, such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc., and this application embodiment does not limit this.

[0094] After calculating the time-series correlation parameters of various environmental parameters, the time-series correlation parameters can be concatenated with the features of various environmental parameters to obtain the feature vector.

[0095] Specifically, the features of various environmental parameters and time-series correlation parameters can be arranged and concatenated in a certain order. For example, the concatenation can be done by simple vector concatenation, or by introducing weighted average or other concatenation methods according to specific needs. The resulting feature vector can include the direct features of environmental parameters and the time-series correlation between various environmental parameters.

[0096] In one example, the feature vector is represented as follows:

[0097] in, This represents the feature vector, consisting of two elements in the first and second dimensions. The pressure difference is characterized by the pressure difference between the public flue and the indoor area, and the rate of change of the pressure difference, respectively; the two elements of the 3rd and 4th dimensions. The characteristics of the target concentration are represented by VOCs concentration and the integral of the target concentration within the target time window, respectively; the elements of the 5th dimension. The characteristic of temperature represents the temperature difference between the public flue temperature and the indoor temperature; the three elements of dimensions 6-8. The temporal correlation characteristics of different environmental parameters are represented by the time difference between the pressure difference and the target concentration, the serial correlation coefficient between the pressure difference and the target concentration, and the serial correlation coefficient between the temperature difference and the target concentration, respectively.

[0098] In this embodiment, by calculating the temporal correlation parameters between various environmental parameter features, it is possible to obtain the sequential logic of abnormal changes in different parameters, the synergy and correlation of feature changes over time, so that the feature vector can characterize the precursory pattern of oil fume backflow, making the feature vector more consistent with the actual characterization of oil fume backflow, and further improving the accuracy of oil fume backflow prediction.

[0099] In some embodiments, the prediction model is trained as follows: A training sample set is constructed, which includes positive samples and negative samples. The negative samples are the feature vectors corresponding to the environmental parameters collected when backflow of oil fumes occurs in the public flue, and the positive samples are the feature vectors corresponding to the environmental parameters collected in non-backflow of oil fumes scenarios. The prediction model is trained using a training sample set.

[0100] In this embodiment, considering that the prediction scenario of backflow of cooking fumes needs to be adapted to embedded devices such as range hood controllers, the prediction model is preferably a lightweight model suitable for embedded devices, such as a small fully connected neural network. Small fully connected neural networks have the advantages of small parameter size, fast inference speed, and low hardware resource consumption, which can reduce the dependence on the computing power of embedded devices.

[0101] In this embodiment, a training sample set containing positive and negative samples needs to be constructed. Negative samples are feature vectors corresponding to environmental parameters collected when backflow of cooking fumes occurs in a public flue. Negative samples reflect the characteristics and environmental state at the time of backflow and can be labeled with a true label "1" to indicate a backflow state. Positive samples are feature vectors corresponding to environmental parameters collected in non-backflow scenarios. Positive samples provide characteristics and environmental states under normal operating conditions, helping the prediction model distinguish between normal and backflow situations. Positive samples can be labeled with a true label "0" to indicate a non-backflow state.

[0102] It should be noted that, in order to improve the diversity and representativeness of the training sample set, in practice, positive and negative samples can be obtained by collecting data at different times and under different environmental conditions.

[0103] During training, the parameters of the prediction model, such as weights and biases in the neural network, need to be initialized first. Then, the training sample set is input into the prediction model. The prediction model will perform forward propagation calculation based on the current parameters and output the predicted probability of backflow of cooking fumes. For example, the prediction model will output a probability value between 0 and 1, representing the possibility of backflow of cooking fumes.

[0104] In this embodiment, a loss function can be used to measure the difference between the prediction results of the prediction model and the actual situation. For example, binary cross-entropy can be used as the loss function, and its mathematical expression is as follows:

[0105] Where L represents the binary cross-entropy loss function, y is the true label of the sample, and p is the predicted probability of backflow of oil fumes corresponding to the sample output by the prediction model. The smaller the value of the loss function, the closer the prediction result of the prediction model is to the actual situation.

[0106] By calculating the gradient of the loss function with respect to the parameters of the prediction model, and using an optimization algorithm (such as Adam) for backpropagation, the parameters of the prediction model are updated, so that the value of the loss function continuously decreases. After the loss function satisfies a certain number of iterations or convergence conditions, the training ends, and the trained prediction model is obtained.

[0107] In this embodiment, the prediction model is trained by constructing a training sample set containing positive and negative samples, which can cover both backflow and non-backflow scenarios of oil fume. During the training process, the prediction model can learn the differences in feature vectors between positive and negative samples, so that the trained prediction model can accurately grasp the feature distribution patterns under different scenarios and improve the prediction ability of oil fume backflow.

[0108] In some embodiments, the prediction results include the probability value of backflow of cooking fumes; The backflow of cooking fumes in the public flue is determined based on the prediction results output by the prediction model, including: If the probability of backflow of cooking fumes exceeds the target threshold, it is determined that there is a risk of backflow of cooking fumes in the public flue.

[0109] In this embodiment, the prediction result is output in the form of a backflow probability value of cooking fumes. The backflow probability value ranges from [0, 100%]. The larger the backflow probability value, the more likely backflow of cooking fumes is to occur.

[0110] For example, when the model outputs a backflow probability of 85%, it indicates that the current environmental state has a high degree of feature matching with historical backflow scenarios, and the risk of backflow is high; when the probability is 50%, it indicates that the current state is at the critical point between backflow and non-backflow; when the probability is 10%, it indicates that the current environmental state conforms to the normal smoke exhaust pattern, and the risk of backflow is low.

[0111] To accurately determine whether there is a risk of backflow of cooking fumes, the probability value of backflow of cooking fumes can be compared with a target threshold. When the probability value of backflow of cooking fumes is greater than the target threshold, it is considered that there is a risk of backflow of cooking fumes in the public flue. When the probability value of backflow of cooking fumes is less than or equal to the target threshold, it is considered that there is no risk of backflow of cooking fumes in the public flue.

[0112] The target threshold can be preset, for example, it can be 70%, 75%, 80%, 90%, etc., but this application embodiment does not limit it.

[0113] In this embodiment, the backflow probability value of oil fume output by the prediction model is used to determine the backflow situation of oil fume in the public flue. The backflow probability value can reflect the likelihood of backflow of oil fume, further improving the accuracy of backflow prediction.

[0114] In some embodiments, the method further includes: If it is determined that there is a risk of backflow of cooking fumes into the public exhaust duct, turn on the exhaust fan.

[0115] In some embodiments, if it is determined that there is a risk of backflow of cooking fumes into the public exhaust duct, a warning message can be issued, for example, by prompting the user with a voice message, or by prompting the user with a text message, a pop-up window in a mobile application, etc.

[0116] In this embodiment, backflow prevention control logic can also be activated, for example, by turning on the range hood, which further reduces the pressure in the kitchen relative to the public flue, thereby reducing the risk of backflow of fumes.

[0117] To reduce the risk of backflow of cooking fumes during prolonged periods of ineffective operation, the risk status can be continuously monitored. If no backflow risk is confirmed for several consecutive cycles, the range hood should be stopped or its operating speed reduced. Alternatively, to further reduce the risk of backflow, if a backflow risk is confirmed for several consecutive cycles, the operating speed can be increased.

[0118] In this embodiment, when it is determined that there is a risk of backflow of oil fumes in the public flue, backflow prevention control is implemented by turning on the range hood and activating its exhaust function to enhance the exhaust of indoor air, thereby reducing the occurrence of backflow of oil fumes.

[0119] In some embodiments, the method further includes: Receive feedback from users regarding the prediction results; the feedback includes valid predictions and invalid predictions. If the feedback information is a valid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a positive sample; if the feedback information is an invalid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a negative sample. The prediction model is incrementally trained using an incremental training set.

[0120] In this embodiment, the prediction model also has simple online learning capabilities. Users can input feedback information via a mobile application regarding the accuracy of the prediction results. The feedback information includes valid predictions and invalid predictions. A valid prediction means that the prediction model's prediction results match the actual situation; for example, the prediction model predicts a risk of backflow of cooking fumes, and actual monitoring confirms that backflow of cooking fumes has occurred. An invalid prediction means that the prediction model's prediction results do not match the actual situation; for example, the prediction model predicts a risk of backflow of cooking fumes, but backflow of cooking fumes has not actually occurred.

[0121] In this embodiment, after receiving user feedback, the corresponding environmental parameter feature vector can be processed according to the type of feedback and added to the incremental training set. Specifically, when the feedback is a valid prediction, the environmental parameter feature vector corresponding to the prediction result can be added to the incremental training set as a positive sample. When the feedback is an invalid prediction, the corresponding feature vector is added to the incremental training set as a negative sample.

[0122] As users continue to provide feedback, the incremental training set will gradually accumulate more positive and negative samples, thus enabling the incremental training set to reflect the performance of the prediction model in real-time.

[0123] Incremental training is a dynamic model update method that allows a model to optimize and adjust itself based on new data without retraining the entire dataset. In this embodiment, the prediction model can be incrementally trained using an incremental training set at preset intervals or at any specified time, allowing the prediction model to adjust its parameters based on positive and negative samples in the incremental training set.

[0124] In this embodiment, user feedback information includes two cases: valid prediction and invalid prediction. When the feedback information is a valid prediction, it means that the output of the prediction model matches the actual situation. The corresponding environmental parameter feature vector is added as a positive sample to the incremental training set, which can enhance the model's ability to predict features of similar scenarios. When the feedback information is an invalid prediction, it means that the prediction result does not match the actual situation. The corresponding environmental parameter feature vector is added as a negative sample to the incremental training set, which can specifically correct the model's judgment bias, so as to continuously adjust and optimize its own parameters and structure to better adapt to the user's specific home environment.

[0125] In some embodiments, the method further includes: Statistics on the backflow of cooking fumes into public exhaust ducts at different times of the day; The same period of time during which there is a risk of backflow of cooking fumes for N consecutive days is defined as the period when backflow is prone to occur. Turn on the smoke generator when the current time is within the target duration of the period when backflow is likely to occur.

[0126] Considering that the occurrence of backflow of cooking fumes is affected by factors such as residents' cooking habits and daily routines, the phenomenon tends to exhibit time-based clustering. For example, during the breakfast (7:00-8:30) and dinner (18:00-19:30) periods when residents cook together, the amount of cooking fumes emitted into the public exhaust duct increases dramatically, airflow turbulence intensifies, and the risk of backflow is higher than at other times.

[0127] In this embodiment, a 24-hour day can be divided into multiple statistical periods at fixed time intervals. For example, each 30 minutes or hour can be divided into a statistical period. Based on the prediction results of the prediction model, the risk of backflow of oil fumes can be statistically analyzed for each statistical period of the day.

[0128] Statistical results can be used to identify the same time period where there is a risk of backflow of cooking fumes over N consecutive days. N can be 3 days, 5 days, 7 days, etc. For example, if statistical data shows that the probability of backflow of cooking fumes exceeds the target threshold between 7 pm and 8 pm for 7 consecutive days, indicating a risk of backflow, then this period is identified as a high-risk period for backflow.

[0129] After identifying the periods when backflow is likely to occur, the anti-backflow control logic can be activated based on the relationship between the current time and these periods. Specifically, the smoke generator will automatically turn on when the current time is 30 minutes away from the target duration of the potential backflow period. The target duration can be preset, such as 10 minutes, 20 minutes, or 30 minutes. For example, if the target duration is set to 30 minutes, the smoke generator will automatically start when it detects that there are 30 minutes left before the potential backflow period.

[0130] In this embodiment, by statistically analyzing the backflow situation at different times of the day, the occurrence pattern of backflow can be extracted from historical data. By turning on the range hood in advance when the current time is a target time away from the backflow-prone period, a stable negative pressure environment can be established between the indoor and public flues before the risk of backflow occurs, thus reducing the risk of oil fume backflow.

[0131] The oil fume backflow prediction method provided in this application can be executed by an oil fume backflow prediction device. This application uses an oil fume backflow prediction device to execute the oil fume backflow prediction method as an example to illustrate the oil fume backflow prediction device provided in this application.

[0132] This application also provides an oil fume backflow prediction device.

[0133] like Figure 5 As shown, the oil fume backflow prediction device includes: The acquisition module 510 is used to acquire environmental parameters of the common flue; the environmental parameters include at least two of the following: pressure difference between the common flue and the room where the smoke machine is located, concentration of the target substance in the common flue, and temperature of the common flue. The extraction module 520 is used to extract features of various environmental parameters and perform feature fusion on the features of various environmental parameters based on the temporal correlation of various environmental parameters to obtain feature vectors; The prediction module 530 is used to input the feature vector into the trained prediction model and determine the backflow of oil fumes in the public flue based on the prediction results output by the prediction model.

[0134] According to the oil fume backflow prediction device of this application, environmental parameters of a common flue are acquired. These environmental parameters include at least two of the following: pressure difference between the common flue and the room where the range hood is located, target concentration in the common flue, and temperature of the common flue. Features of various environmental parameters are extracted, and feature fusion is performed based on the temporal correlation of these parameters to obtain a feature vector. The feature vector is then input into a trained prediction model, and the oil fume backflow situation in the common flue is determined based on the prediction results output by the model. This embodiment of the application can acquire environmental features related to oil fume backflow from different dimensions through different environmental parameters. By extracting features of various environmental parameters and performing feature fusion based on the temporal correlation of these parameters to obtain a feature vector, the correlation between different environmental parameters and the laws governing parameter changes over time are fully considered. This allows the feature vector to more accurately reflect the trend of oil fume backflow. By utilizing the model's data analysis and prediction capabilities to output prediction results and determine the oil fume backflow situation, the accuracy of oil fume backflow prediction is improved. This provides a basis for timely preventive measures and reduces the pollution caused by oil fume backflow to the indoor environment.

[0135] In some embodiments, the acquisition module 510 is further configured to: Environmental parameters of the public flue can be obtained using at least one of the following methods: Micro pressure sensors are installed in the common flue and the room where the range hood is located. The pressure difference is calculated based on the pressure of the common flue and the room where the range hood is located, based on the pressure of the common flue and the room where the range hood is located. The public flue is equipped with a target object monitoring sensor to obtain the concentration of target objects in the public flue. The target object concentration includes one or more of the following: VOCs concentration, particulate matter concentration, and fatty acid concentration. Temperature sensors are installed in the public flue to obtain the temperature of the public flue.

[0136] In some embodiments, the extraction module 520 is further configured to: Given environmental parameters including the pressure difference between the public flue and the room where the exhaust fan is located, the pressure difference and the rate of change of the pressure difference within the target time window are defined as the characteristics of the pressure difference. Given the target concentration in the environmental parameter common flue, the target concentration, as well as the integral and / or gradient of the target concentration within the target time window, are determined as characteristics of the target concentration. Given the environmental parameter of the temperature of the public flue, the temperature difference between the public flue and the indoor temperature is defined as the characteristic of the temperature of the public flue.

[0137] In some embodiments, the extraction module 520 is further configured to: Calculate the time-series correlation parameters between the characteristics of various environmental parameters; the time-series correlation parameters include at least one of the time difference of abrupt changes in various environmental parameters and the sequence correlation coefficient of the characteristics of various environmental parameters. The features of time-series correlation parameters and various environmental parameters are concatenated to obtain a feature vector.

[0138] In some embodiments, the prediction model is trained as follows: A training sample set is constructed, which includes positive samples and negative samples. The negative samples are the feature vectors corresponding to the environmental parameters collected when backflow of oil fumes occurs in the public flue, and the positive samples are the feature vectors corresponding to the environmental parameters collected in non-backflow of oil fumes scenarios. The prediction model is trained using a training sample set.

[0139] In some embodiments, the prediction module 530 is further configured to: If the probability of backflow of cooking fumes exceeds the target threshold, it is determined that there is a risk of backflow of cooking fumes in the public flue.

[0140] In some embodiments, the prediction module 530 is further configured to: If it is determined that there is a risk of backflow of cooking fumes into the public exhaust duct, turn on the exhaust fan.

[0141] In some embodiments, the prediction module 530 is further configured to: Receive feedback from users regarding the prediction results; the feedback includes valid predictions and invalid predictions. If the feedback information is a valid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a positive sample; if the feedback information is an invalid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a negative sample. The prediction model is incrementally trained using an incremental training set.

[0142] In some embodiments, the prediction module 530 is further configured to: Statistics on the backflow of cooking fumes into public exhaust ducts at different times of the day; The same period of time during which there is a risk of backflow of cooking fumes for N consecutive days is defined as the period when backflow is prone to occur. Turn on the smoke generator when the current time is within the target duration of the period when backflow is likely to occur.

[0143] The backflow prediction device for cooking fumes in this application embodiment can be an electronic device or a component within an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), embedded device, range hood, etc. This application embodiment does not specifically limit the specific device.

[0144] The oil fume backflow prediction device in this application embodiment can be a device with an operating system. This operating system can be a Microsoft (Windows) operating system, an Android operating system, an iOS operating system, or other possible operating systems, such as an embedded operating system; this application embodiment does not specifically limit the specific operating system.

[0145] In some embodiments, such as Figure 6 As shown, this application embodiment also provides an electronic device 600, including a processor 601, a memory 602, and a computer program stored in the memory 602 and executable on the processor 601. When the program is executed by the processor 601, it implements the various processes of the above-described oil fume backflow prediction method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0146] It should be noted that the electronic devices in the embodiments of this application include the aforementioned mobile electronic devices and non-mobile electronic devices.

[0147] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described oil fume backflow prediction method embodiments and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0148] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0149] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for predicting backflow of cooking fumes.

[0150] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0151] This application also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described oil fume backflow prediction method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0152] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0153] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0154] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0155] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0156] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0157] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for predicting backflow of cooking fumes, characterized in that, include: Obtain environmental parameters of the common flue; the environmental parameters include at least two of the following: pressure difference between the common flue and the room where the exhaust fan is located, concentration of the target substance in the common flue, and temperature of the common flue; Features of various environmental parameters are extracted, and feature fusion is performed on the features of various environmental parameters based on their temporal correlation. Obtain the feature vector; The feature vector is input into a trained prediction model, and the backflow of oil fumes in the public flue is determined based on the prediction results output by the prediction model.

2. The method according to claim 1, characterized in that, Environmental parameters of the public flue can be obtained using at least one of the following methods: Micro pressure sensors are installed in the common flue and the room where the range hood is located. The pressure difference is calculated based on the pressure of the common flue and the room where the range hood is located, which are detected by the micro pressure sensors. The public flue is equipped with a target object monitoring sensor to obtain the target object concentration in the public flue. The target object concentration includes one or more of VOCs concentration, particulate matter concentration, and fatty acid concentration. A temperature sensor is installed in the public flue to obtain the temperature of the public flue.

3. The method according to claim 1, characterized in that, The extraction of features from various environmental parameters includes: When the environmental parameters include the pressure difference between the common flue and the room where the exhaust fan is located, the pressure difference and the rate of change of the pressure difference within the target time window are determined as the characteristics of the pressure difference; In the case of the target concentration in the public flue of the environmental parameters, the target concentration, and the integral and / or gradient of the target concentration within the target time window are determined as characteristics of the target concentration; Given the temperature of the public flue in the environmental parameters, the temperature difference between the public flue and the indoor temperature is defined as a characteristic of the temperature of the public flue.

4. The method according to claim 1, characterized in that, The feature fusion is performed on the features of various environmental parameters based on the temporal correlation of these environmental parameters. The feature vector is obtained, including: Calculate the temporal correlation parameters between the features of each of the environmental parameters; the temporal correlation parameters include at least one of the time difference of abrupt changes in each of the environmental parameters and the sequence correlation coefficients of the features of each of the environmental parameters. The feature vector is obtained by concatenating the features of the time-series correlation parameters and the various environmental parameters.

5. The method according to claim 1, characterized in that, The prediction model is trained in the following manner: A training sample set is constructed, which includes positive samples and negative samples. The negative samples are feature vectors corresponding to environmental parameters collected when backflow of oil fumes occurs in a public flue, and the positive samples are feature vectors corresponding to environmental parameters collected in non-backflow of oil fumes scenarios. The prediction model is trained using the training sample set.

6. The method according to claim 1, characterized in that, The prediction results include the probability value of backflow of cooking fumes; Determining the backflow of oil fumes in the public flue based on the prediction results output by the prediction model includes: If the probability value of backflow of cooking fumes is greater than the target threshold, it is determined that there is a risk of backflow of cooking fumes in the public flue.

7. The method according to claim 1, characterized in that, The method further includes: If it is determined that there is a risk of backflow of oil fumes into the public flue, the exhaust fan shall be turned on.

8. The method according to claim 1, characterized in that, The method further includes: Receive feedback from users regarding the prediction results; the feedback includes valid predictions and invalid predictions. If the feedback information is a valid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a positive sample; if the feedback information is an invalid prediction, the feature vector corresponding to the environmental parameter is added to the incremental training set as a negative sample. The prediction model is incrementally trained using the incremental training set.

9. The method according to claim 1, characterized in that, The method further includes: Statistics on the backflow of cooking fumes in the public flue at different times each day; The same period of time during which there is a risk of backflow of cooking fumes for N consecutive days is defined as the period when backflow is prone to occur. The smoke machine is turned on when the current time is within the target time period of the backflow-prone period.

10. A device for predicting backflow of cooking fumes, characterized in that, include: An acquisition module is used to acquire environmental parameters of the common flue; the environmental parameters include at least two of the following: the pressure difference between the common flue and the room where the exhaust fan is located, the concentration of the target substance in the common flue, and the temperature of the common flue; The extraction module is used to extract features from various types of environmental parameters, and to perform feature fusion on the features of various types of environmental parameters based on the temporal correlation of these environmental parameters. Obtain the feature vector; The prediction module is used to input the feature vector into a trained prediction model and determine the backflow of oil fumes in the public flue based on the prediction results output by the prediction model.

Images