Air conditioning unit

Abstract

This invention provides an air conditioning device, comprising a data acquisition module and a control module. The data acquisition module is configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device. The control module is configured to establish a prediction model based on a Transformer architecture based on historical samples; and to predict the predicted set temperature for several future target times based on the established prediction model. Furthermore, when the user set temperature deviates from the predicted set temperature at a future target time, the user set temperature is corrected to the predicted set temperature. Historical samples are generated based on the environmental parameters, operating parameters, and / or user behavior parameters associated with corresponding timestamps and stored in a predetermined database. This application can achieve high-precision, stable, and adaptive prediction of future set temperatures, improving the intelligence level of the air conditioning device.

Description

Technical Field

[0001] This invention relates to the field of air conditioning technology, and more particularly to an air conditioning device. Background Technology

[0002] Intelligent control of residential air conditioners has always been one of the main research topics in related fields. Manufacturers hope that their products can automatically control based on users' usage habits, such as automatically turning on the air conditioner before the user arrives home to make the ambient temperature more comfortable, or automatically turning off the air conditioner after the user leaves home to avoid wasting electricity.

[0003] Existing technologies have disclosed relevant technical solutions to solve this problem, such as the technical solution disclosed in Chinese patent application (CN107504656A): "An air conditioning system self-learning control method, including an information acquisition module and an automatic module, the self-learning control method includes: Step S10: The information acquisition module acquires data on whether the user turns the air conditioning system on or off using a remote control or APP, and sends the acquired data to the automatic control module; Step S11: The automatic control module analyzes and processes the acquired data by learning the user's on / off habits, and identifies the on / off time periods of the air conditioning system; Step S12: The automatic control module processes the on / off time periods of the air conditioning system respectively to determine the on / off time of the air conditioning system, wherein determining the on / off time of the air conditioning system includes: recording the on / off time in all on / off time periods, and calculating the self-learning on / off time of the nth hour period; determining the off / off time of the air conditioning system includes: recording the off / off time in all off / off time periods, and calculating the self-learning off / off time of the mth hour period; Step S13: The automatic control module automatically controls the on / off time of the air conditioning system according to the determined on / off time of the air conditioning system."

[0004] The self-learning control method provided in the comparison document generates self-learned on / off times based on the average number of on / off cycles over 10 days. However, in reality, user operation and control of air conditioning systems, especially in a specific scenario, are discrete time series with indefinite time intervals. In addition to on / off control, it also includes mode control (cooling / heating), function control (sleep / voice / human-sensor / auxiliary heating), air supply control (airflow direction / speed), etc. Predictive biases based on traditional data statistics are large, and the level of intelligence is poor.

[0005] The information disclosed in this background section is only intended to enhance the understanding of the background technology of this application, and therefore may include prior art that is not known to those skilled in the art. Summary of the Invention

[0006] In view of the problems pointed out in the background art, the first aspect of this application provides an air conditioning device.

[0007] Specifically, the air conditioning device includes a data acquisition module and a control module; the data acquisition module is configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device; the control module is configured to establish a prediction model based on the Transformer architecture based on historical samples; and, based on the established prediction model, predict the predicted set temperature for several future target times; and, when the user set temperature at a future target time deviates from the predicted set temperature, correct the user set temperature to the predicted set temperature.

[0008] In one or more embodiments of this application, the historical samples are generated and stored in a predetermined database based on the environmental parameters, operating parameters and / or user behavior parameters associated with the corresponding timestamps.

[0009] For air conditioning devices, due to various events such as extreme weather (e.g., rapid temperature changes) or uncontrollable factors, the patterns in the observation sequence may change significantly over time. This results in several outliers that cannot be eliminated. The observation sequences corresponding to these outliers are then input into the prediction model, causing each vector to be pointed to during dot product processing without fully utilizing contextual information. This leads to local contextual confusion in the attention sublayer, causing potential optimization problems. To solve this problem, a causal convolutional layer is configured in the prediction model. The prediction model uses the output sequence of the causal convolutional layer to calculate the query matrix and key matrix, and uses the query matrix and key matrix to calculate the attention weights of the prediction model to handle the dependencies between features at different locations in the observation sequence. The observation sequence is generated based on one or more parameters in the historical samples and serves as the input sequence for the prediction model.

[0010] Since the environmental parameters, operating parameters, and user behavior parameters collected by the data acquisition module are fine-grained, long-term dependent time series, in order to reduce the modeling difficulty from a hardware perspective, in one or more embodiments of this application, a sparse mode is configured in the prediction model; the prediction model can be configured to selectively calculate attention weights in the sparse mode to enhance the attention weights of features at some positions in the observation sequence and reduce the attention weights of features at other positions in the observation sequence; the observation sequence is generated based on one or more parameters in the historical samples and serves as the input sequence of the prediction model.

[0011] In one or more embodiments of this application, the sparse pattern is formed based on the LogSparse SelfAttention strategy; the prediction model includes multiple sub-attention layers, each of which is configured with the sparse pattern; the prediction model is stacked with multiple sub-attention layers of the sparse pattern.

[0012] In one or more embodiments of this application, the operating parameters include the actual operating state of the air conditioning device; the actual operating state includes user-defined parameters.

[0013] In one or more embodiments of this application, to remove invalid data, the air conditioning device further includes a preprocessing module, which is configured to enter a waiting state when the data acquisition module acquires any one or more user-defined parameters, and determine whether the acquired parameters remain unchanged within a set time period; if the acquired parameters remain unchanged within the set time period, the acquired parameters are presumed to be valid user-defined parameters, and the valid user-defined parameters are configured to be associated with the corresponding timestamp and stored in a predetermined database as part of the historical sample.

[0014] To help the prediction model obtain more stable and accurate predictions, the control module is also configured with a correction model independent of the prediction model. The correction model is configured to generate a set temperature correction value. The control module is configured to associate the set temperature correction value with the corresponding timestamp and store it in a predetermined database as part of the historical samples.

[0015] In one or more embodiments of this application, the air conditioning device further includes: a training module configured to input a plurality of the observation sequences to train the prediction model to adjust the hyperparameters of the prediction model until the training model converges.

[0016] In one or more embodiments of this application, the air conditioning device further includes: a testing module, which tests the prediction model trained by the training module to evaluate whether the prediction result of the prediction model meets the set accuracy conditions.

[0017] In one or more embodiments of this application, the air conditioning device further includes: an evaluation module configured to correct the user-set temperature to the predicted set temperature when the user-set temperature deviates from the predicted set temperature at a future target time; and to collect whether an intervention set temperature output by the user is received within a set evaluation period; and to update the observation sequence and configure the training module to retrain the prediction model when the intervention set temperature is received.

[0018] A second aspect of this application provides an air conditioning device, including a data acquisition module and a control module; the data acquisition module is configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device; the control module is configured to establish a prediction model based on a Transformer architecture based on historical samples; and, based on the established prediction model, predict the predicted set temperature at a target future time corresponding to the restart of the air conditioning device; and, when the air conditioning device restarts, correct the set temperature stored during standby to the predicted set temperature; the historical samples are generated based on the environmental parameters, operating parameters, and / or user behavior parameters associated with corresponding timestamps and stored in a predetermined database.

[0019] This application enables highly accurate adaptive intelligent generation of predicted set temperatures.

[0020] Other features and advantages of the present invention will become clearer after reading the detailed embodiments of the invention in conjunction with the accompanying drawings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A schematic block diagram of the structure of an air conditioning device provided for one or more embodiments of this application;

[0023] Figure 2 A schematic block diagram of the structure of a data acquisition module in an air conditioning device provided for one or more embodiments of this application;

[0024] Figure 3 A schematic block diagram of the structure of a data acquisition module in an air conditioning device provided for one or more embodiments of this application;

[0025] Figure 4 A schematic block diagram of the structure of a data acquisition module in an air conditioning device provided for one or more embodiments of this application;

[0026] Figure 5 A schematic block diagram of the structure of a data acquisition module in an air conditioning device provided for one or more embodiments of this application;

[0027] Figure 6 A flowchart of a control module in an air conditioning device provided for one or more embodiments of this application;

[0028] Figure 7 A schematic diagram of the structure of a causal convolutional layer in an air conditioning device provided for one or more embodiments of this application;

[0029] Figure 8 A schematic diagram of the structure of a causal convolutional layer in an air conditioning device provided for one or more embodiments of this application;

[0030] Figure 9 The dot product curve corresponding to the causal convolution layer in the air conditioning device provided in one or more embodiments of this application;

[0031] Figure 10 A schematic diagram illustrating the principle of sparse mode in an air conditioning device provided for one or more embodiments of this application;

[0032] Figure 11 A schematic block diagram of the structure of an air conditioning device provided for one or more embodiments of this application;

[0033] Figure 12 A flowchart of a pretreatment module in an air conditioning apparatus provided for one or more embodiments of this application;

[0034] Figure 13 A schematic block diagram of the structure of the control module of the air conditioning device provided in one or more embodiments of this application;

[0035] Figure 14 A schematic block diagram of the structure of an air conditioning device provided for one or more embodiments of this application;

[0036] Figure 15 A schematic block diagram of the structure of an air conditioning device provided for one or more embodiments of this application;

[0037] Figure 16 A schematic block diagram of the structure of an air conditioning device provided for one or more embodiments of this application;

[0038] Figure 17 This is a flowchart of a control module in an air conditioning device provided for one or more embodiments of this application. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0040] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0041] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "multiple" means two or more.

[0042] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0043] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" the second feature includes the first feature directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0044] The following disclosure provides many different embodiments or examples for implementing various structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. In addition, examples of various specific processes and materials are provided in this invention, but those skilled in the art will recognize the application of other processes and / or the use of other materials.

[0045] One or more embodiments of the present invention provide an air conditioning device. The air conditioning device in this application performs a refrigeration cycle by using a compressor, a condenser, an expansion valve, and an evaporator. The refrigeration cycle includes a series of processes involving compression, condensation, expansion, and evaporation to cool or heat an indoor space.

[0046] Low-temperature, low-pressure refrigerant enters the compressor, which compresses it into a high-temperature, high-pressure refrigerant gas and discharges the compressed refrigerant gas. The discharged refrigerant gas flows into the condenser. The condenser condenses the compressed refrigerant into a liquid phase, and the heat is released to the surrounding environment through the condensation process.

[0047] The expansion valve expands the high-temperature, high-pressure liquid refrigerant that condenses in the condenser into a low-pressure liquid refrigerant. The evaporator evaporates the expanded refrigerant in the expansion valve and returns the low-temperature, low-pressure refrigerant gas to the compressor. The evaporator achieves its cooling effect by utilizing the latent heat of refrigerant evaporation to exchange heat with the material being cooled. Throughout this cycle, the air conditioning system regulates the temperature of the indoor space.

[0048] The outdoor unit of an air conditioning unit refers to the part of the refrigeration cycle that includes the compressor and the outdoor heat exchanger. The indoor unit of an air conditioning unit includes the indoor heat exchanger, and an expansion valve can be provided in either the indoor or outdoor unit.

[0049] Indoor and outdoor heat exchangers function as either condensers or evaporators. When the indoor heat exchanger is used as a condenser, the air conditioning unit functions as a heater in heating mode; when the indoor heat exchanger is used as an evaporator, the air conditioning unit functions as a cooler in cooling mode.

[0050] Figure 1 This is a schematic block diagram of the air conditioning device provided in this embodiment; as shown. Figure 1 The air conditioning unit shown can automatically predict and configure future set temperatures to create smarter, more efficient, and personalized air conditioning systems, resulting in a better user experience and energy efficiency.

[0051] like Figure 1As shown, the air conditioning device includes a data acquisition module 100 and a control module 200. The data acquisition module 100 is configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device. The control module 200 is configured to establish a prediction model based on a Transformer architecture based on historical samples; and, based on the established prediction model, predict the setpoint temperature for several future target times; and, when the user setpoint temperature deviates from the predicted setpoint temperature at a future target time, correct the user setpoint temperature to the predicted setpoint temperature. The historical samples are generated based on environmental parameters, operating parameters, and / or user behavior parameters associated with corresponding timestamps and stored in a predetermined database.

[0052] Figure 2 This is a schematic block diagram of a data acquisition module 100 provided for one or more embodiments of the present invention. In this embodiment, the data acquisition module 100 is configured to collect environmental parameters of an air conditioning device.

[0053] In one or more embodiments of this application, the environmental parameters include indoor environmental parameters and outdoor environmental parameters.

[0054] In one or more embodiments of this application, indoor environmental parameters include: indoor ambient temperature, indoor ambient humidity, and indoor light intensity.

[0055] In one or more embodiments of this application, the indoor environmental parameters further include: air quality parameters, such as the concentrations of carbon dioxide, formaldehyde, and volatile organic compounds (VOCs).

[0056] In one or more embodiments of this application, indoor environmental parameters may also include the number of people in the room and odor levels.

[0057] In one or more embodiments of this application, outdoor environmental parameters include: outdoor ambient temperature, outdoor ambient humidity, and weather parameters; wherein the weather parameters include: precipitation level, wind speed, wind direction, and solar radiation intensity, etc.

[0058] Indoor and outdoor ambient temperatures can be collected using temperature sensors, while indoor and outdoor humidity can be collected using humidity sensors.

[0059] Indoor light intensity can be acquired through light meters, spectroradiometers, light sensors, or cameras and image processing. Indoor light intensity is used to measure and assess the intensity and distribution of light in an indoor environment. For example, light intensity can be measured using a photosensitive sensor, which operates based on the photoelectric effect. For instance, the resistance of a photoresistor changes with light intensity; when light shines on a photosensitive material, it excites electrons within the material, thus changing the resistance. Besides photoresistors, photosensitive sensors can also be implemented using photodiodes, phototransistors, photocapacitors, and other similar devices.

[0060] The concentrations of carbon dioxide, formaldehyde, volatile organic compounds (VOCs), and odor levels can all be collected by corresponding sensors, while noise levels can be collected by a decibel meter.

[0061] Weather parameters can be obtained from the server.

[0062] Figure 3 is a schematic block diagram of the structure of a data acquisition module 100 provided in one or more embodiments of the present invention. In this embodiment, the data acquisition module 100 is configured to collect the operating parameters of an air conditioning device.

[0063] In one or more embodiments of this application, the operating parameters include the actual operating state of the air conditioning device; the actual operating state includes, but is not limited to, one or more of the following: set temperature, set fan speed, set working mode, timer setting, set humidity, set energy efficiency mode, fault status, filter status, and energy consumption status.

[0064] Figure 4 The diagram below is a schematic block diagram of the structure of a data acquisition module 100 provided for one or more embodiments of the present invention. In this real-time mode, the data acquisition module 100 is configured to collect user behavior parameters of the air conditioning device.

[0065] User behavior parameters for air conditioning devices refer to information such as users' behaviors, habits, and preferences when using air conditioning devices.

[0066] In one or more embodiments of the present invention, user behavior parameters include one or more of the following: setting temperature preference, setting wind direction preference, setting wind speed preference, adjusting frequency preference, setting mode preference, setting timing preference, indoor and outdoor temperature difference preference, time period preference, preference for coordination with other devices, calibration preference, and feedback preference.

[0067] Among them, the preferred temperature settings represent the user's temperature preferences for different time periods, such as daytime, nighttime, weekdays, and weekends; preferred wind direction settings represent the preferred wind direction settings for different time periods; preferred wind speed settings represent the preferred wind speed settings for different time periods; preferred adjustment frequency settings represent the frequency at which the user adjusts the set temperature; preferred mode settings represent the user's preferences for different operating modes (cooling, heating, fresh air, etc.) and whether to use energy-saving mode; preferred time settings represent whether the user uses the timer function to set the air conditioner to automatically turn on or off at a specific time; preferred indoor-outdoor temperature difference settings represent the user's preferences regarding the difference between indoor and outdoor temperatures, such as the on / off action or temperature adjustment action corresponding to the indoor-outdoor temperature difference; preferred time period settings represent the user's preferences for using the air conditioner during different time periods, such as daytime, nighttime, and rest time; preferred coordination with other devices settings represent whether the user uses the air conditioner while using other smart devices, such as using multiple devices in a smart home system simultaneously; preferred calibration settings represent whether the user prefers to adjust the set temperature based on actual comfort levels and the degree of calibration; and preferred feedback settings represent whether the user has provided feedback on the user experience of the air conditioning device, such as unsatisfactory control performance in certain scenarios.

[0068] Figure 5 This is a schematic diagram of the structure of a data acquisition module 100 provided for one or more embodiments of the present invention. In this embodiment, the data acquisition module 100 is configured to simultaneously collect environmental parameters, operating parameters and user behavior parameters of the air conditioning device.

[0069] In one or more embodiments of this application, each environmental parameter, operating parameter, and user behavior parameter collected by the data acquisition module 100 includes a corresponding timestamp. For user behavior parameters, the timestamp is the time when the user behavior parameter is generated, for example, configuring the corresponding preference calculation method on the server and using the time when the corresponding preference is received from the server as the timestamp of the user behavior parameter.

[0070] Historical samples are generated by associating environmental parameters, runtime parameters, and / or user behavior parameters with corresponding timestamps and stored in a predefined database.

[0071] The control module 200 is configured to establish a prediction model based on a Transformer architecture according to historical samples; and, based on the established prediction model, predict the setpoint temperature for several future target times; and, when the user-setpoint temperature at a future target time deviates from the predetermined setpoint temperature, correct the user-setpoint temperature to the predicted setpoint temperature; such as... Figure 6 As shown.

[0072] The Transformer architecture is a deep learning architecture for processing sequential data. Its core idea is to map the input sequence into a set of key-value pairs and use an attention mechanism to weight and summarize these key-value pairs to produce the output sequence. The Transformer consists of two main components: the encoder and the decoder.

[0073] The encoder maps the input sequence into a set of key-value pairs. Each input is embedded as a vector and then processed through multiple layers of sub-attention layers and a feedforward neural network. The sub-attention mechanism allows the encoder to automatically adjust weights based on other positions in the input sequence as it processes each input, thereby capturing global contextual information.

[0074] The decoder also includes sub-attention layers and feedforward neural networks. When generating sequence data, the decoder can not only refer to the encoded representation of the input sequence, but also perform attention calculations on the already generated parts of the sequence, thereby capturing information from the previously generated parts when generating the next element.

[0075] The following describes the prediction model based on the Transformer architecture, established from historical samples, in one or more embodiments of this application:

[0076] In mathematics, the target variable, i.e., the observed values ​​(e.g., the set temperature), can be represented by a set:

[0077]

[0078] The above formula includes a time series consisting of observations from time step 1 to time step t0, where N represents the number of observations in the set, i represents the index of the observation in the time series (which can be an integer from 1 to N), and t0 represents the maximum value of the time step. The time steps in the observation series start from 1 and gradually increase until t0. The definition of is:

[0079]

[0080] The time step is used to represent the position of each observation sequence in the time dimension. The time step can correspond to different time intervals, such as one second, one hour, etc.

[0081] Assuming the time step at the future target time is τ, the external sequence can also be represented by a set:

[0082]

[0083] That is, it includes the external sequence from time step 1 to time step t0+τ, N represents the number of external sequences in the set, i represents the index of the external sequence, which can be an integer from 1 to N, t0+τ represents the maximum value of the time step, and the time steps in the external sequence start from 1 and gradually increase until t0+τ.

[0084] The prediction objective is thus transformed into modeling and predicting the following conditional distributions:

[0085]

[0086] This represents the future observation sequence, from time step t0+1 to t0+τ;

[0087] This represents the past observation sequence, from time step 1 to t0;

[0088] This represents the external sequence, from time step 1 to t0+τ;

[0089] Φ represents the parameter set of the model, i.e., the learnable parameters.

[0090] Indicates a known past observation sequence external sequence Given the parameter set Φ, the future observation sequence The conditional probability density function;

[0091] This represents the product of probability density functions in the time step from t0+1 to t0+τ.

[0092] Based on the above model, we can further simplify it into a learning prediction model.

[0093] p(z t |z i，1：t-1 x 1：t ;Φ)

[0094] Among them, z t z represents the observation at time step t; i，1：t-1 Let x represent the sequence of past observations from time step 1 to time step t-1. 1：t Let p(z) represent the past and present external sequence from time step 1 to t, Φ represent the model's parameter set, i.e., the learnable parameters; t |z i，1：t-1 x 1：t ;Φ); indicates that in the known past observation sequence z i，1：t-1 and external sequence x 1：t Given the parameter set Φ, the observed value z at time step t. tThe conditional probability density function.

[0095] To make full use of the observations and covariates, we concatenate them to obtain the following augmented matrix.

[0096]

[0097]

[0098] y t Let z represent a vector. t-1 Let x represent the observation of vector z at time step t-1. t Let represent the observation of vector x at time step t, and ο represent element-wise operations. Y represents the (d+1)-dimensional real space; t This represents the vectors y1, ..., y2. t The matrix formed has dimensions t×(d+1), y1, ..., y2. t This represents the observations of vector y at different time steps, where T represents the transpose. Let represent the real number space of row t and column d+1.

[0099] In the above formula, y t The definition includes z t-1 (observed values) and x t The element-level concatenation or combination of (covariates) further generates augmented vectors containing information about the observations and covariates. In observation sequence analysis, these augmented vectors are stacked together according to time steps to obtain the augmented matrix Y. t Augmented matrix Y t Each row in the matrix corresponds to a time step, which contains the observations and covariate information at that time step, thus providing a more comprehensive data representation. The augmented matrix can help capture the relationships between observations and other relevant information, as well as their dynamic changes between different time steps.

[0100] Further, a suitable model z needs to be generated. t ~f(Y t To predict a given distribution, in one or more embodiments of this application, a prediction model is built based on the Transformer architecture. The Transformer architecture utilizes a multi-head attention mechanism to instantiate f(Y). t This is because sub-attention enables prediction models under the Transformer architecture to capture long-term or short-term dependencies, and different attention heads learn to focus on different aspects of the observation sequence.

[0101] In the sub-attention layer, a multi-head attention sub-layer simultaneously focuses Y t Transform into H different query matrices key matrix and value matrix Where h = 1, ..., H, after these linear projections, the attention computation vector output sequence is:

[0102]

[0103] The attention-based computational vector output sequence is used to compute a series of context vectors by applying an attention mechanism. These vectors reflect the relationships between different positions in the input sequence and different positions in the generated target sequence. These context vectors are used in the decoder to generate each element of the target sequence, ultimately producing the prediction result.

[0104] In one or more embodiments of this application, 15 features are selected from environmental parameters, operational parameters, and user behavior parameters, and multiple observation sequences arranged in a time series are generated based on timestamps. That is, the observation sequences are generated based on one or more parameters from historical samples and used as input sequences for the prediction model. An example of an observation sequence could be:

[0105] [′iu34Status01′, ′iu46TempCv′, ′iu55Wind′, ′I1′, ′12′, ′I3′, ′14′, ′I5′, ′season′, ′inDl′, ′outDl′, ′temp_range′, ′temp_rec′, wind_rec′, mode_rec′]

[0106] Among them, iu34Status01 represents the actual operating status, iu46TempCv represents the temperature setting correction value; for example, it can be 0-4℃, iu55Wind represents the indoor fan speed, I1 represents the outdoor ambient temperature, I2 represents the outdoor ambient humidity, I3 represents the outdoor wind speed, I4 represents the outdoor wind direction, I5 represents the weather type; inDl represents the indoor comfort level; outDl represents the outdoor comfort level; temp_range represents the indoor-outdoor temperature difference; temp_rec represents the user's temperature preference; wind_rec represents the user's fan speed preference; and mode_rec represents the user's mode preference.

[0107] Among them, such as Figure 13As shown, the set temperature correction value is generated by a correction model independent of the prediction model, which is configured in the control module. For example, the correction model can fine-tune or correct the set temperature based on factors that affect the actual room temperature, such as weather changes or direct sunlight; or based on the heat generated by people's activities in the room; or based on the temperature differences between different locations in the room; or based on different user perceptions of comfort temperature. The correction model can employ various models disclosed in the prior art (such as algorithms disclosed in the background art), which are not the focus of this application and will not be elaborated upon here. The control module 200 is configured to associate the set temperature correction value with the corresponding timestamp and store it in a predetermined database as part of the historical samples.

[0108] Since the set temperature correction value itself originates from the correction model, and is then input into the neural network as a feature of the sample parameters, the prediction results of the correction model can be combined with the actual correction value. This provides the neural network with more comprehensive information, helping it to better understand the performance and bias of the correction model. Setting the temperature correction value in conjunction with weather changes, direct sunlight, heat generated by people's activities in the room, temperature differences between different locations in the room, and users' different perceptions of comfortable temperature can enrich the input features of the neural network, improve the model's predictive ability, and calibrate the bias of the correction model. By combining the advantages of the correction model and the neural network, a more stable and accurate prediction can be obtained. The neural network can also adjust in real time based on the patterns and trends of historical correction values, showing periodic changes.

[0109] In one or more embodiments of this application, user temperature preferences, user wind speed preferences, and user mode preferences can be obtained by analyzing user behavior patterns based on historical data. For example, user set temperature data, set wind speed data, and set mode data within a past set time window are collected, and mathematical models are used to analyze the trend of the set temperature to obtain user temperature preferences, user wind speed preferences, and / or user mode preferences. For instance, a user may prefer a lower temperature during a specific time period, a lower wind speed during a specific time period, or a certain mode (e.g., silent mode or sleep mode) during a specific time period.

[0110] In one or more embodiments of this application, indoor comfort and outdoor comfort can be represented by a temperature-humidity index (also known as a discomfort index) or a temperature-humidity index, which is based on temperature and relative humidity and is used to describe the comfort of the human body in a thermal environment.

[0111] In one or more embodiments of this application, indoor comfort and outdoor comfort can be obtained by the following formula:

[0112] DI = T - (0.55 - 0.55f)(T - 58)

[0113] Where T represents the real-time temperature (in Fahrenheit) and f represents the real-time humidity.

[0114] In one or more embodiments of this application, indoor comfort and outdoor comfort can also be represented by the perceived temperature index (also known as the thermal index). Perceived temperature is also based on temperature and humidity and is used to measure the actual temperature felt by the human body. The perceived temperature index further describes the actual temperature felt by the human body in cold wind and low-temperature environments, and based on a combination of air temperature and wind speed, it further reflects the effect of wind speed on heat dissipation from the human body.

[0115] In one or more embodiments of this application, indoor comfort and outdoor comfort can also be represented by the PMV index or PPD index, which comprehensively considers factors such as temperature, humidity, airflow speed, clothing, and metabolic rate.

[0116] The 15 features mentioned above are encoded into a feature dimension of (15, 512) and used as the input sequence for the prediction model. This can be achieved using methods such as embedding layers and fully connected layers.

[0117] The prediction is then implemented by the encoder and decoder in the Transformer architecture. The prediction process is detailed in the introduction to the Transformer model above.

[0118] like Figure 11 As shown, in one or more embodiments of this application, the air conditioning device further includes a preprocessing module 300, which is used to screen and process historical samples.

[0119] In one or more embodiments of this application, the preprocessing module 300 may also be configured with a preprocessing algorithm. The preprocessing algorithm is used to remove interference from invalid operation data of users, remove unstable data of users, and ensure the validity of data, that is, to screen the collected indoor environmental parameters.

[0120] In one or more embodiments of this application, the preprocessing algorithm includes: filtering valid indoor unit setting parameters; and merging outdoor environmental data.

[0121] In one or more embodiments of this application, filtering valid indoor unit setting parameters includes: obtaining indoor unit setting parameters of the same indoor unit based on the same device code and machine number.

[0122] A device code is a unique identifier used to distinguish different devices. It can be a string of numbers, letters, or a combination of numbers and letters. Device codes can be used to manage licenses, authorize devices to access specific services or functions, and track and manage devices.

[0123] A machine serial number is a unique identifier for a machine, assigned by the manufacturer to track, identify, and manage different machines in the production process. Machine serial numbers can be used for maintenance records, warranty tracking, and identifying problems in production.

[0124] like Figure 12 As shown, in one or more embodiments of this application, the preprocessing module is configured to enter a waiting state when the data acquisition module 100 acquires any one or more user-defined parameters, and determine whether the acquired parameters remain unchanged within a set time period; if the acquired parameters remain unchanged within the set time period, the acquired parameters are presumed to be valid user-defined parameters, and the valid user-defined parameters are configured to be associated with the corresponding timestamp and stored in a predetermined database as historical samples.

[0125] In one or more embodiments of this application, the duration can be set to 10 minutes.

[0126] In one or more embodiments of this application, merging outdoor environmental data includes: acquiring outdoor environmental parameters; and merging the outdoor environmental data when the operation time and the acquisition time of the outdoor environmental parameters do not exceed a set duration.

[0127] In one or more embodiments of this application, the set duration may be 30 minutes.

[0128] In one or more embodiments of this application, the preprocessing algorithm for the filtered indoor environmental parameters includes: missing value processing, outlier processing, and / or normalization processing.

[0129] In one or more embodiments of this application, missing value handling employs interpolation. Interpolation estimates missing values ​​using known data points, such as linear interpolation, polynomial interpolation, spline interpolation, etc. For example, spline interpolation can fill in missing values ​​between data points by fitting a smooth curve or surface. It divides the entire data range into multiple small intervals and then fits a polynomial or other smooth function within each interval. The connection points between the small intervals are nodes, which can be selected using rules such as uniform distribution or quantiles. Spline interpolation uses piecewise polynomials, fitting a low-degree polynomial within each interval, optionally, for example, a quadratic or cubic polynomial, to ensure smoothness within a local range. These polynomials are continuous at the nodes, and their derivatives at the nodes are also continuous to maintain overall smoothness.

[0130] In one or more embodiments of this application, missing value processing may also employ the mean, median, mode, fixed value, or values ​​from similar samples for filling.

[0131] In one or more embodiments of this application, missing value processing can also be achieved using observation sequence interpolation, regression methods, machine learning models, or multiple interpolation.

[0132] In one or more embodiments of this application, outlier processing includes two steps: identifying outliers and deleting outliers; or two steps: identifying outliers and correcting outliers; or both. Identifying outliers can employ statistical methods, such as using visualization tools (e.g., box plots, scatter plots, etc.); or it can employ outlier detection algorithms, such as clustering-based methods, to automatically identify and process outliers. Correcting outliers can involve replacing them with other suitable values, such as using statistical indicators like the mean or median.

[0133] In one or more embodiments of this application, the normalization process can employ min-max normalization, standardization, fractional scaling standardization, or norm normalization. Min-max normalization refers to linearly mapping the feature values ​​to a specified minimum and maximum value. Standardization involves subtracting the mean and dividing by the standard deviation to ensure the feature values ​​have a distribution with a mean of 0 and a standard deviation of 1. Fractional scaling standardization involves dividing the feature values ​​by the cardinality to ensure the absolute values ​​of the feature values ​​are within the range [0.1, 1). Norm normalization scales the feature vector of each sample so that the norm of the vector equals 1. Norm normalization ensures that each sample's position in the multidimensional space is at the same distance from the origin. Normalized sample data can eliminate dimensional differences between features, resulting in a more balanced impact of each feature on the model.

[0134] In one or more embodiments of this application, for air conditioning devices, due to various events such as extreme weather (e.g., rapid temperature changes) or uncontrollable factors, the patterns in the observation sequence may change significantly over time. This results in several outliers and observation sequences corresponding to these change points that cannot be eliminated by the preprocessing module 300 and are input into the Transformer model. However, in the Transformer's sub-attention layer, the similarity between the query matrix and the key matrix is ​​calculated based on their point-wise relationships. When generating observation sequences corresponding to outliers and change points, the "point" refers to each vector, without fully utilizing the contextual information. This local context can confuse the attention sub-layer, making the observed values ​​appear as outliers. The sub-attention score only reflects the correlation between single time points, which differs from the original intention of predicting the observation sequence and introduces potential optimization problems.

[0135] In particular, if a convolution operation with a kernel of 1 and a stride of 1 is used when calculating Query, Key, and Value, the dot product operation after the convolution operation may make the relationship between features at different positions unclear, since the calculation of attention weights is done through the dot product operation between the query matrix and the key matrix.

[0136] In one or more embodiments of this application, a causal convolutional layer is configured in the prediction model; the prediction model uses the output sequence of the causal convolutional layer to compute a query matrix and a key matrix, and uses the query matrix and key matrix to compute the attention weights of the prediction model to handle the dependencies between features at different locations in the observation sequence.

[0137] Specifically, in causal convolutional layers, such as Figure 8 and Figure 9 As shown, the Query, Key, and Value are calculated through a convolution operation with kernel k and stride 1. This allows the Query and Key matrices at different locations to consider the relationships between them and their positional differences. This positional awareness helps the model better understand the correlations between different locations, thereby improving model performance, especially when processing observation sequences. Causal convolutional layers ensure that the current location does not access future information. By using causal convolutional layers, the generated Query and Key matrices can more clearly understand the remaining time points, or context, relative to the current location. Therefore, calculating their similarity through their local contextual information improves prediction accuracy, overcomes the interference of outliers, and, given the increasing frequency of extreme weather events, makes the model more resilient to extreme conditions, further improving the intelligent performance of air conditioning devices.

[0138] For air conditioning devices, the environmental parameters, operating parameters, and user behavior parameters collected by the data acquisition module 100 are fine-grained, long-term dependent time series. In order to meet the high accuracy requirements, the acquisition time intervals for parameters such as temperature and humidity are small (i.e., fine-grained), and there are long-term dependencies between data points (such as preference), which also means that the observation at the current time step is affected by multiple past time steps. When calculating the sub-attention score, it will lead to O(L2) memory usage, which will further increase the modeling difficulty from a hardware perspective.

[0139] To address this issue, a sparse mode is configured in the prediction model. The prediction model can be configured to selectively compute attention weights in sparse mode to enhance the attention weights of features at some locations in the observation sequence and reduce the attention weights of features at other locations. Specifically, sparse mode computes the similarity between the query matrix and the key matrix, for example, using dot product or other methods, and outputs a similarity matrix. Each element in the similarity matrix represents a similarity score between the query matrix and the key matrix. An attention weight matrix of the same size as the similarity matrix is ​​created, and all elements are initialized to zero. At specified locations in sparse mode, the similarity scores are applied to the attention weights. At these specified locations in sparse mode, the similarity scores are exponentialized. The exponentialized attention weights are normalized to ensure that their sum is 1 or close to 1. By exponentializing the similarity scores and applying them to the attention weights, the specified locations in sparse mode receive enhanced weights when calculating attention, thus better capturing information that the model deems important. This approach enhances the model's attention to specific locations while maintaining computational efficiency. For the observation sequence, this means that the attention weight of features at some locations in the observation sequence can be increased while the attention weight of features at other locations in the observation sequence can be decreased.

[0140] The specified position in the sparse pattern can be selected randomly or based on the importance of features. For example, user preferences can be used as the specific position in the sparse pattern.

[0141] like Figure 10 As shown, the sparse pattern can be formed based on the LogSparse Self Attention strategy. The prediction model includes multiple sub-attention layers, each configured with a sparse pattern. That is, the above steps are performed in each sub-attention layer, and finally multiple sub-attention layers are stacked to enhance the model's expressive power and significantly reduce memory costs.

[0142] Based on the Transformer prediction model provided by this invention, long-term dependencies can be captured. Compared with recent RNN-based methods, matrix factorization methods, and classical statistical methods, the proposed convolutional attention further improves the performance of Transformer and reaches the state-of-the-art level in different environments. In addition, under the same memory budget, the sparse attention model can achieve better results on data with long-term dependencies.

[0143] The control module 200 can also be configured to configure a first sparse mode strategy, a second sparse mode strategy, and a third sparse mode strategy in the prediction model based on the Transformer architecture built from historical samples. All three sparse mode strategies employ a LogSparse Self-Attention strategy. In the first sparse mode strategy, the specified position represents user preferences; in the second sparse mode strategy, the specified position represents environmental parameters at several closest time points; and in the third sparse mode strategy, the specified position represents energy price parameters. The control module 200 can select the first, second, or third sparse mode strategy based on user needs. For example, during normal use, the control module 200 executes a first sparse mode strategy to increase the influence of user preferences on the prediction; when predicting a set temperature in the near future, the control module 200 executes a second sparse mode strategy to increase the influence of environmental parameters from the most recent time points on the prediction; when focusing more on energy prices, the control module 200 executes a third sparse mode strategy to increase the influence of energy price parameters on the prediction, thereby enabling the prediction model to better capture past trends and patterns and achieve optimal configuration so that the model can perform well when predicting future set temperatures.

[0144] like Figure 14 As shown, in some embodiments of this application, the air conditioning device further includes a training module 400. The training module 400 is configured to input multiple observation sequences to train a prediction model and adjust the hyperparameters of the prediction model until the training model converges. More specifically, the training module 400 uses the observation sequences to fine-tune the parameters of the prediction model, selecting appropriate hyperparameters based on the loss function; for example, the number of layers in the encoder and decoder, the number of attention heads, etc., until the prediction model reaches a convergent state.

[0145] like Figure 15 As shown, in some embodiments of this application, the air conditioning device further includes a testing module 500. The testing module 500 tests the prediction model trained by the training module 400 to evaluate whether the prediction results of the prediction model meet the set accuracy conditions. Evaluation metrics include root mean square error (RMSE) or mean absolute error (MAE), etc.

[0146] like Figure 16 As shown, in some embodiments of this application, the air conditioning device further includes an evaluation module 600. The evaluation module 600 is configured to correct the user-set temperature to the predicted set temperature when the user-set temperature deviates from the predicted set temperature at a future target time; and to collect data on whether an intervention set temperature output by the user is received within a set evaluation period; and, upon receiving the intervention set temperature, to update the observation sequence and configure the training module 400 to retrain the prediction model. When an intervention set temperature output by the user is received, it indicates that the user's subjective experience was unsatisfactory during the evaluation phase, requiring modification of the observation sequence and re-conducting the experiment.

[0147] like Figure 17 As shown, in one or more embodiments of this application, an air conditioning device is also provided. The air conditioning device includes a data acquisition module 100 and a control module 200. The data acquisition module 100 is configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device. The control module 200 is configured to establish a prediction model based on a Transformer architecture according to historical samples; and, based on the established prediction model, predict the predicted set temperature at a target future time corresponding to the restart of the air conditioning device; and, when the air conditioning device restarts, use the predicted set temperature as the set temperature of the air conditioning device. Historical samples are generated and stored in a predetermined database based on environmental parameters, operating parameters, and / or user behavior parameters associated with corresponding timestamps.

[0148] In the use of air conditioning devices, there is a situation where the indoor unit or controller has a standby function, meaning that the air conditioning device is usually set to retain the previous set temperature when it is turned off. That is, the set temperature is stored in the air conditioning device's non-volatile memory. When users leave the room or go to sleep, they often choose to turn off the air conditioning equipment. However, when the air conditioning device is restarted, the standby stored set temperature is often significantly different from the actual temperature. In this application, the control module 200 corrects the standby stored set temperature to a predicted set temperature when the air conditioning device is restarted. This can meet the needs of the actual current situation and respond more quickly to temperature changes and user needs, rather than relying solely on previous static settings. For example, it can provide a comfortable environment for users as soon as they wake up in the morning.

[0149] In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0150] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An air conditioning device, comprising: A data acquisition module, configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device; The air conditioning device is characterized in that it further includes: The control module is configured to establish a prediction model based on the Transformer architecture based on historical samples; and to predict the predicted set temperature for several future target times based on the established prediction model; and to correct the user set temperature to the predicted set temperature when the user set temperature deviates from the predicted set temperature at a future target time. The historical samples are generated and stored in a predetermined database based on the environmental parameters, operating parameters and / or user behavior parameters associated with the corresponding timestamps. The prediction model is configured with causal convolutional layers; the prediction model uses the output sequence of the causal convolutional layers to calculate a query matrix and a key matrix, and uses the query matrix and the key matrix to calculate the attention weights of the prediction model to handle the dependencies between features at different locations in the observation sequence; The observation sequence is generated based on one or more parameters from the historical samples and serves as the input sequence for the prediction model.

2. The air conditioning device according to claim 1, characterized in that: The prediction model is also configured with a sparse mode; the prediction model can be configured to selectively compute attention weights in the sparse mode to enhance the attention weights of features at some locations in the observation sequence and reduce the attention weights of features at other locations in the observation sequence. The observation sequence is generated based on one or more parameters from the historical samples and serves as the input sequence for the prediction model.

3. The air conditioning device according to claim 2, characterized in that: The sparse pattern is formed based on the LogSparse Self Attention strategy; the prediction model includes multiple sub-attention layers, and each sub-attention layer is configured with the sparse pattern. The prediction model is stacked with multiple sub-attention layers of the sparse pattern.

4. The air conditioning device according to claim 1 or 2, characterized in that: The operating parameters include the actual operating status of the air conditioning unit; The actual operating status includes user-defined parameters; The air conditioning device also includes: The preprocessing module is configured to enter a waiting state when the data acquisition module acquires any one or more user-defined parameters, and determine whether the acquired parameters remain unchanged within a set time period; if the acquired parameters remain unchanged within the set time period, the acquired parameters are presumed to be valid user-defined parameters, and the valid user-defined parameters are configured to be associated with the corresponding timestamps and stored in a predetermined database as part of the historical samples.

5. The air conditioning device according to claim 1 or 2, characterized in that: The control module is also equipped with a correction model independent of the prediction model, which is configured to generate a set temperature correction value. The control module is configured to associate the set temperature correction value with the corresponding timestamp and store it in a predetermined database as part of the historical sample.

6. The air conditioning device according to claim 1 or 2, characterized in that: The air conditioning device also includes: The training module is configured to input multiple observation sequences to train the prediction model and adjust the hyperparameters of the prediction model until the prediction model converges.

7. The air conditioning device according to claim 6, characterized in that: The air conditioning device also includes: The testing module tests the prediction model trained by the training module to evaluate whether the prediction results of the prediction model meet the set accuracy conditions.

8. The air conditioning device according to claim 7, characterized in that: The air conditioning device also includes: The evaluation module is configured to correct the user-set temperature to the predicted set temperature when the user-set temperature deviates from the predicted set temperature at a future target time; collect data on whether the user-output intervention set temperature is received within a set evaluation period; and update the observation sequence and configure the training module to retrain the prediction model when the intervention set temperature is received.

9. An air conditioning device, comprising: A data acquisition module, configured to collect environmental parameters, operating parameters, and / or user behavior parameters of the air conditioning device; The air conditioning device is characterized in that it further includes: The control module is configured to establish a prediction model based on the Transformer architecture based on historical samples; and to predict the predicted set temperature at the target future time corresponding to the restart of the air conditioning device based on the established prediction model; and to correct the set temperature stored in standby mode to the predicted set temperature when the air conditioning device restarts. The historical samples are generated and stored in a predetermined database based on the environmental parameters, operating parameters and / or user behavior parameters associated with the corresponding timestamps. The prediction model is configured with causal convolutional layers; the prediction model uses the output sequence of the causal convolutional layers to calculate a query matrix and a key matrix, and uses the query matrix and the key matrix to calculate the attention weights of the prediction model to handle the dependencies between features at different locations in the observation sequence; The observation sequence is generated based on one or more parameters from the historical samples and serves as the input sequence for the prediction model.

Images