A dehumidifier water level prediction method and system based on CT-SVM

Abstract

The present invention relates to the technical field of dehumidifiers, and particularly to a dehumidifier water level prediction method and system based on CT-SVM. The present invention collects relevant data of historical dehumidifiers, adopts a hybrid feature selection method combining filter and wrapper methods, takes detection rate and false detection rate as evaluation basis for feature selection, uses a relief method to sort features by relevance, adopts an SBS search strategy from back to front according to the sorted features, uses SVM as a classifier, performs feature screening on the collected dehumidifier-related data, selects an optimal feature subset, and obtains an adaptive coefficient through an adaptive feature operator extraction algorithm. Based on the optimal feature subset and the adaptive coefficient, a CT-SVM algorithm is used to predict the water level of the dehumidifier. The present invention adopts the CT-SVM algorithm to perform water level prediction, which is more accurate and adaptable and reduces model training.

Description

Technical Field

[0001] The present invention relates to the technical field of dehumidifiers, and in particular to a dehumidifier water level prediction method and system based on CT-SVM. Background Art

[0002] Dehumidifiers, essential household appliances for regulating indoor humidity, are widely used in daily life and industrial production. Traditional dehumidifiers typically feature a water tank to collect moisture condensed from the air during the dehumidification process. However, in practice, users cannot determine when the water tank is full. This problem causes significant inconvenience. For example, if users fail to detect a full tank, the dehumidifier may overflow, soaking the surrounding area and damaging furniture, floors, and other items. Frequent checks on the water tank waste unnecessary time and effort, reducing the user experience.

[0003] To solve this problem, some dehumidifiers in the prior art use a water-full indicator light or an alarm device. However, the water-full indicator light requires the user to actively observe the dehumidifier panel to know the status of the water tank. When the user is not paying attention to the dehumidifier, the user may still miss the reminder that the water tank is full. Although the alarm device can sound a reminder when the water tank is full, it cannot ensure that the user receives the notification in time in some noisy environments or when the user is not present. In addition, these prompting methods cannot let the user know in advance that the water tank is about to be full, making it difficult to prepare in advance and there is a certain lag. Therefore, the development of a technology that can accurately and promptly let the user know the status of the water tank, especially a technology that can predict when the water tank is full, is of great significance to improving the performance and user experience of the dehumidifier. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention aims to provide a dehumidifier water level prediction method and system based on CT-SVM, which can predict the water level in the dehumidifier water tank and facilitate users to understand the water level of the dehumidifier water tank.

[0005] In order to solve the above problems, the present invention adopts the following technical solutions:

[0006] In one aspect, the present invention provides a dehumidifier water level prediction method based on CT-SVM, comprising:

[0007] Collect historical dehumidifier related data, including indoor environment data, outdoor environment data, operating status data, time and region data and water tank water level change rate.

[0008] A hybrid feature selection method combining the Filter and Wrapper methods was adopted. The detection rate and false detection rate were used as the evaluation basis for feature selection. The Relief method was used to sort the features by relevance. The SBS search strategy was adopted from back to front according to the sorted features. Support Vector Machine (SVM) was used as the classifier to screen the features of the collected dehumidifier-related data. The optimal feature subset was selected, and the adaptive coefficient was obtained through the adaptive feature operator extraction algorithm.

[0009] Based on the optimal feature subset and adaptive coefficient, the CT-SVM algorithm is used to predict the water level of the dehumidifier.

[0010] As an implementable method, selecting the optimal feature subset includes:

[0011] S201. In the Filter stage, the relief weight of each feature is calculated for the original features in the dehumidifier-related data.

[0012] S202 : sorting the original features in the dehumidifier-related data in descending order according to the calculated relief weight of each feature, and constructing a feature sequence.

[0013] S203, in the Wrapper stage, the SBS search strategy is used to remove one feature X from the feature sequence from the back to the front, and SVM is used as the classifier to calculate the detection rate and false detection rate of the remaining feature subsets and the feature evaluation value J. If J is satisfied after removing feature X, 剔除Xk Greater than J 剔除Xk+1 , then repeat S203 until J 剔除Xk Less than or equal to J 剔除Xk+1 The remaining features in the current feature sequence are determined as the optimal feature subset, where k is the kth feature removed from the feature sequence.

[0014] As an implementable method, obtaining the adaptive coefficient includes:

[0015] Based on the historical data of the optimal feature subset, the random forest algorithm is used to calculate the prediction error α of each sample i , where α i =|Y i -yβ i |, where i is the sample number, Y i is the actual water level change rate of the i-th sample, y is the predicted water level change rate of the water tank, β i is the adaptive coefficient of the i-th sample.

[0016] If the prediction error is greater than the set threshold, the sample weight is adjusted and the prediction adaptive coefficient is retrained; if the prediction error is less than or equal to the set threshold, the training is completed and the final adaptive coefficient is obtained.

[0017] As an implementable method, the water level prediction of the dehumidifier using the CT-SVM algorithm includes:

[0018] The compressed tracking CT technology is used to dynamically and real-time sort the importance of the features in the optimal feature subset based on a tree model, and the features whose importance is lower than the importance threshold are screened out to obtain the optimal low-dimensional feature subset.

[0019] The adaptive coefficient is combined with the optimal low-dimensional feature subset to construct an enhanced input vector, which is input into a nonlinear support vector machine classification model. The water level change rate of the water tank is divided into multiple types as output, the volume of the water tank is determined according to the product model, and the current water volume of the water tank and the expected time to be full are calculated.

[0020] As an implementable method, the classification decision function of the nonlinear support vector machine classification model is:

[0021]

[0022]

[0023] .

[0024] Where N is the number of augmented input vectors, is the optimal solution of the Lagrange multiplier, is the category label of the i-th training sample, is the kernel function, representing the vector in the original input space and training sample vectors Mapping to high-dimensional feature space, is the intercept term of the classification hyperplane, is the category label of the jth training sample, is the kernel function, representing the i-th training sample and the jth training sample Inner product in high-dimensional feature space, is 1 / 5.

[0025] On the other hand, the present invention provides a dehumidifier water level prediction system based on CT-SVM, including an acquisition module, a screening module and a water level prediction module.

[0026] The acquisition module is used to collect historical dehumidifier related data, including indoor environment data, outdoor environment data, operating status data, time and region data and water tank water level change rate.

[0027] The screening module is used to adopt a hybrid feature selection method that combines the Filter and Wrapper methods, takes the detection rate and false detection rate as the evaluation basis for feature selection, uses the Relief method to sort the features by relevance, adopts the SBS search strategy from back to front according to the sorted features, uses SVM as the classifier, performs feature screening on the collected dehumidifier-related data, selects the optimal feature subset, and obtains the adaptive coefficient through the adaptive feature operator extraction algorithm.

[0028] The water level prediction module is used to predict the water level of the dehumidifier using the CT-SVM algorithm based on the optimal feature subset and the adaptive coefficient.

[0029] As an implementable method, selecting the optimal feature subset includes:

[0030] S201. In the Filter stage, the relief weight of each feature is calculated for the original features in the dehumidifier-related data.

[0031] S202 : sorting the original features in the dehumidifier-related data in descending order according to the calculated relief weight of each feature, and constructing a feature sequence.

[0032] S203, in the Wrapper stage, the SBS search strategy is used to remove one feature X from the feature sequence from the back to the front, and SVM is used as the classifier to calculate the detection rate and false detection rate of the remaining feature subsets and the feature evaluation value J. If J is satisfied after removing feature X, 剔除Xk Greater than J 剔除Xk+1 , then repeat S203 until J 剔除Xk Less than or equal to J 剔除Xk+1 The remaining features in the current feature sequence are determined as the optimal feature subset, where k is the kth feature removed from the feature sequence.

[0033] As an implementable method, obtaining the adaptive coefficient includes:

[0034] Based on the historical data of the optimal feature subset, the random forest algorithm is used to calculate the prediction error α of each sample i , where α i =|Y i -yβ i |, where i is the sample number, Y i is the actual water level change rate of the i-th sample, y is the predicted water level change rate of the water tank, β i is the adaptive coefficient of the i-th sample.

[0035] If the prediction error is greater than the set threshold, the sample weight is adjusted and the prediction adaptive coefficient is retrained; if the prediction error is less than or equal to the set threshold, the training is completed and the final adaptive coefficient is obtained.

[0036] As an implementable method, the water level prediction of the dehumidifier using the CT-SVM algorithm includes:

[0037] The compressed tracking CT technology is used to dynamically and real-time sort the importance of the features in the optimal feature subset based on a tree model, and the features whose importance is lower than the importance threshold are screened out to obtain the optimal low-dimensional feature subset.

[0038] The adaptive coefficient is combined with the optimal low-dimensional feature subset to construct an enhanced input vector, which is input into a nonlinear support vector machine classification model. The water level change rate of the water tank is divided into multiple types as output, the volume of the water tank is determined according to the product model, and the current water volume of the water tank and the expected time to be full are calculated.

[0039] As an implementable method, the classification decision function of the nonlinear support vector machine classification model is:

[0040]

[0041]

[0042] .

[0043] Where N is the number of augmented input vectors, is the optimal solution of the Lagrange multiplier, is the category label of the i-th training sample, is the kernel function, representing the vector in the original input space and training sample vectors Mapping to high-dimensional feature space, is the intercept term of the classification hyperplane, is the category label of the jth training sample, is the kernel function, representing the i-th training sample and the jth training sample Inner product in high-dimensional feature space, is 1 / 5.

[0044] The beneficial effects of the present invention are: the present invention adopts a hybrid feature selection method combining the Filter and Wrapper methods to screen the optimal feature subset and calculate the adaptive coefficient, and successfully constructs a water level prediction model through the CT-SVM algorithm model to realize the prediction of the water level in the dehumidifier. It is more accurate than the general prediction model, and is suitable for various types of dehumidifiers, reducing model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a dehumidifier water level prediction method based on CT-SVM in the present invention.

[0046] Figure 2 Schematic diagram of the process of selecting the optimal feature subset in the present invention.

[0047] Figure 3 This is a schematic diagram of a dehumidifier water level prediction system based on CT-SVM in the present invention. DETAILED DESCRIPTION

[0048] The present invention will be further described in detail below with reference to specific embodiments.

[0049] It should be noted that these embodiments are only used to illustrate the present invention, rather than to limit the present invention. Simple improvements to the method based on the concept of the present invention fall within the scope of protection claimed by the present invention.

[0050] See also Figure 1 , is a dehumidifier water level prediction method based on CT-SVM, including:

[0051] S100 , collecting historical dehumidifier-related data, including indoor environment data, outdoor environment data, operating status data, time and region data, and water tank water level change rate.

[0052] Raw data capture

[0053] Based on the IoT platform, historical dehumidifier-related data is uploaded to the big data cloud, forming a big data source. The data source includes: indoor environment data, outdoor environment data, operating status data, time and region data, and water tank water level change rate.

[0054] Indoor environment data includes sensor data such as indoor temperature, indoor humidity, and air inlet volume. This data records the indoor environment in which the dehumidifier is running.

[0055] The outdoor environment data includes sensor and meteorological data such as outdoor temperature, outdoor humidity, and outdoor weather. This data records the outdoor environment in which the dehumidifier is running.

[0056] The operating status data includes data such as the water level volume, operating time, operating power, and operating wind speed during the operation of the dehumidifier. This type of data records the operating status of the dehumidifier.

[0057] Time and region data: Region information mainly refers to the geographical location of the dehumidifier; time information mainly includes data such as month and time.

[0058] S200 adopts a hybrid feature selection method that combines the Filter and Wrapper methods, takes the detection rate and false detection rate as the evaluation basis for feature selection, uses the Relief method to sort the features by relevance, and adopts the SBS search strategy from back to front according to the sorted features. It uses SVM as the classifier to screen the features of the collected dehumidifier-related data, selects the optimal feature subset, and obtains the adaptive coefficient through the adaptive feature operator extraction algorithm.

[0059] The hybrid feature selection method based on the combination of filter and wrapper methods evaluates the quality of a feature based on the inherent relationship of the data. It usually uses evaluation criteria to calculate the corresponding value of each feature, compares it with a preset threshold, and selects features that are higher than the threshold. For the specific process, see Figure 2 :

[0060] S201. In the Filter stage, the relief weight of each feature is calculated for the original features in the dehumidifier-related data.

[0061] S202 : sorting the original features in the dehumidifier-related data in descending order according to the calculated relief weight of each feature, and constructing a feature sequence.

[0062] S203, in the Wrapper stage, the SBS search strategy is used to remove one feature X from the feature sequence from the back to the front, and SVM is used as the classifier to calculate the detection rate and false detection rate of the remaining feature subsets and the feature evaluation value J. If J is satisfied after removing feature X, 剔除Xk Greater than J 剔除Xk+1 (This means that a subset of size k has been found, but its performance is better than the original subset. Similarly, subsets of size k-1, k-2, etc. with better performance can be found.) Repeat S203 until J 剔除Xk Less than or equal to J 剔除Xk+1 The remaining features in the current feature sequence are determined as the optimal feature subset, where k is the kth feature removed from the feature sequence.

[0063] Considering the impact of dehumidifier usage time, motor performance fluctuations, and refrigerant leakage on dehumidification performance during operation, dehumidification performance may decline. Therefore, adaptive feature extraction operator technology is introduced. The adaptive feature operator extraction technology uses the historical data of the dehumidifier's optimal feature subset as input. By training the machine learning algorithm model, it obtains the adaptive feature β, which is introduced into the subsequent model input to reduce errors. The specific process includes:

[0064] Based on the historical data of the optimal feature subset, the random forest algorithm is used to calculate the prediction error α of each sample i , where α i =|Yi -yβ i |, where i is the sample number, Y i is the actual water level change rate of the i-th sample, y is the predicted water level change rate of the water tank, β i is the adaptive coefficient of the i-th sample.

[0065] If the prediction error is greater than the set threshold, the sample weight is adjusted and the prediction adaptive coefficient is retrained; if the prediction error is less than or equal to the set threshold, the training is completed and the final adaptive coefficient is obtained.

[0066] Based on the above process, the optimal feature subset and adaptive coefficient examples are screened and shown in Table 1.

[0067] Table 1 Optimal feature subsets

[0068]

[0069] S300 uses the CT-SVM algorithm to predict the water level of the dehumidifier based on the optimal feature subset and the adaptive coefficient.

[0070] This paper proposes a CT (Compressive Tracking)-SVM algorithm for water level prediction. Its core concept is to first use a tree model to select features by importance, eliminating the impact of invalid features on model accuracy. Then, the SVM model is used to predict water levels. Specifically, this algorithm includes:

[0071] The compressed tracking CT technology is used to dynamically and real-time sort the importance of the features in the optimal feature subset based on the tree model, and the features whose importance is lower than the importance threshold are screened out to obtain the optimal low-dimensional feature subset. The example is shown in Table 2.

[0072] Table 2 Optimal low-dimensional feature subsets

[0073]

[0074] The adaptive coefficient is combined with the optimal low-dimensional feature subset to construct an enhanced input vector, which is input into a nonlinear support vector machine classification model. The water level change rate of the water tank is divided into multiple types as output, the volume of the water tank is determined according to the product model, and the current water volume of the water tank and the estimated time until it is full are calculated.

[0075] For example, the enhanced input vector is: indoor humidity h1Humidy is recorded as x1, operating power P is recorded as x2, operating wind speed F is recorded as x3, indoor temperature t1Temp is recorded as x4, and adaptive coefficient β is recorded as x5. The enhanced input vector X=[x1, x2, x3, x4, x5].

[0076] The support vector machine (SVM) is a binary classification model. Its basic model is a linear classifier defined by the maximum margin in feature space, which distinguishes it from a perceptron. The SVM also incorporates kernel techniques, making it a nonlinear classifier. The fundamental idea behind SVM learning is to find a separating hyperplane that correctly partitions the training dataset and maximizes the geometric margin. w·x+b = 0 is the separating hyperplane. For a linearly separable dataset, there are infinitely many such hyperplanes (i.e., perceptrons), but the separating hyperplane with the maximum geometric margin is unique.

[0077] This paper uses a nonlinear support vector machine classification model to transform a nonlinear classification problem in the input space into a linear classification problem in a certain dimensional feature space through a nonlinear transformation, and then learns a linear support vector machine in the high-dimensional feature space. Since in the dual problem of linear support vector machine learning, both the objective function and the classification decision function only involve the inner product between instances, a kernel function is used to represent the inner product between two instances after a nonlinear transformation.

[0078] Input: training dataset constructed from augmented input vectors in, , , The set of enhanced input vectors, n is the number of enhanced input vector features, and N is the number of enhanced input vectors.

[0079] The objective function is:

[0080] .

[0081] w is the weight vector, β i is the adaptive coefficient of the i-th enhanced input vector, ξ i is the slack variable of the i-th augmented input vector.

[0082] Output: Tank water level change rate category.

[0083] Select kernel function and penalty parameter C > 0, construct and solve the convex quadratic programming problem:

[0084]

[0085] .

[0086] Get the optimal solution:

[0087] .

[0088] choose A component of , meeting the conditions ,calculate .

[0089] The classification decision function for:

[0090] .

[0091] Kernel Function for:

[0092] .

[0093] Where N is the number of augmented input vectors, is the optimal solution of the Lagrange multiplier, is the category label of the i-th training sample, is the kernel function, representing the vector in the original input space and training sample vectors Mapping to high-dimensional feature space, is the intercept term of the classification hyperplane, is the category label of the jth training sample, is the kernel function, representing the i-th training sample and the jth training sample Inner product in high-dimensional feature space, is 1 / 5.

[0094] For example, a preliminary test is conducted on four product capability segments sold in the market. The specific product capability segments and parameter details are shown in Table 3. The data from the laboratory test is used as the initial value of the prediction data.

[0095] Table 3 Parameter details of laboratory test models

[0096]

[0097] Indoor humidity environment: 30% - 90%, with a minimum interval threshold of 5%.

[0098] Indoor temperature environment: 5℃-35℃, with the minimum interval threshold being 5℃.

[0099] Operating wind speed: low, medium, high, with the minimum interval threshold being one wind stop.

[0100] This prediction algorithm combines feature selection with a nonlinear SVM algorithm to perform CT-SVM predictions based on individual differences in dehumidifier usage across different regions, fluctuations in compressor motor performance, and refrigerant leaks. The four features selected based on feature importance serve as SVM inputs, and the water tank level change rate is used as the output, which is divided into multiple categories for SVM multi-classification prediction. The app uses the water tank level change rate predicted by cloud-based big data and the volume of the water tank of the product model to calculate the tank's full time and current water level.

[0101] Theoretical calculations show that: 1. The MAE (mean absolute error) of the CT-SVM water-fill prediction model (after introducing β) employed in this invention is 19.6% lower than that of the linear regression model and 8.4% lower than that of the SVM model; 2. The RMAE (root mean square error) of the CT-SVM water-fill prediction model is 72% lower than that of the linear regression model and 18% lower than that of the SVM model. Furthermore, β dynamically responds to device aging and environmental changes, achieving a "one device, one model" model. β values ​​can be updated in the cloud, eliminating the need to retrain the SVM, making it suitable for IoT deployments.

[0102] See also Figure 3 , which is a dehumidifier water level prediction system based on CT-SVM, including an acquisition module 100, a screening module 200 and a water level prediction module 300.

[0103] The acquisition module 100 is used to collect historical dehumidifier-related data, including indoor environment data, outdoor environment data, operating status data, time and region data, and water tank water level change rate.

[0104] The screening module 200 is used to adopt a hybrid feature selection method that combines the Filter and Wrapper methods, uses the detection rate and false detection rate as the evaluation basis for feature selection, uses the Relief method to sort the features by relevance, and adopts the SBS search strategy from back to front according to the sorted features with SVM as the classifier to perform feature screening on the collected dehumidifier-related data, select the optimal feature subset, and obtain the adaptive coefficient through the adaptive feature operator extraction algorithm.

[0105] The water level prediction module 300 is used to predict the water level of the dehumidifier using the CT-SVM algorithm based on the optimal feature subset and the adaptive coefficient.

[0106] Among them, the selection of the optimal feature subset includes:

[0107] S201. In the Filter stage, the relief weight of each feature is calculated for the original features in the dehumidifier-related data.

[0108] S202 : sorting the original features in the dehumidifier-related data in descending order according to the calculated relief weight of each feature, and constructing a feature sequence.

[0109] S203, in the Wrapper stage, the SBS search strategy is used to remove one feature X from the feature sequence from the back to the front, and SVM is used as the classifier to calculate the detection rate and false detection rate of the remaining feature subsets and the feature evaluation value J. If J is satisfied after removing feature X, 剔除Xk Greater than J 剔除Xk+1 , then repeat S203 until J 剔除Xk Less than or equal to J 剔除Xk+1 The remaining features in the current feature sequence are determined as the optimal feature subset, where k is the kth feature removed from the feature sequence.

[0110] The acquisition of the adaptive coefficient includes:

[0111] Based on the historical data of the optimal feature subset, the random forest algorithm is used to calculate the prediction error α of each sample i , where α i =|Y i -yβ i |, where i is the sample number, Y i is the actual water level change rate of the i-th sample, y is the predicted water level change rate of the water tank, β i is the adaptive coefficient of the i-th sample.

[0112] If the prediction error is greater than the set threshold, the sample weight is adjusted and the prediction adaptive coefficient is retrained; if the prediction error is less than or equal to the set threshold, the training is completed and the final adaptive coefficient is obtained.

[0113] Among them, the water level prediction of the dehumidifier using the CT-SVM algorithm includes:

[0114] The compressed tracking CT technology is used to dynamically and real-time sort the importance of the features in the optimal feature subset based on a tree model, and the features whose importance is lower than the importance threshold are screened out to obtain the optimal low-dimensional feature subset.

[0115] The adaptive coefficient is combined with the optimal low-dimensional feature subset to construct an enhanced input vector, which is input into a nonlinear support vector machine classification model. The water level change rate of the water tank is divided into multiple types as output, the volume of the water tank is determined according to the product model, and the current water volume of the water tank and the expected time to be full are calculated.

[0116] Among them, the classification decision function of the nonlinear support vector machine classification model is:

[0117]

[0118]

[0119] .

[0120] Where N is the number of augmented input vectors, is the optimal solution of the Lagrange multiplier, is the category label of the i-th training sample, is the kernel function, representing the vector in the original input space and training sample vectors Mapping to high-dimensional feature space, is the intercept term of the classification hyperplane, is the category label of the jth training sample, is the kernel function, representing the i-th training sample and the jth training sample Inner product in high-dimensional feature space, is 1 / 5.

[0121] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described with reference to the preferred embodiments of the present invention, it should be understood by those skilled in the art that various changes can be made in form and details without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. A dehumidifier water level prediction method based on CT-SVM, characterized in that: include: Collect historical dehumidifier-related data, including indoor environment data, outdoor environment data, operating status data, time and region data, and water tank water level change rate; A hybrid feature selection method combining the filter and wrapper methods was adopted. The detection rate and false positive rate were used as the evaluation criteria for feature selection. The features were sorted by relevance using the relief method. The SBS search strategy was used from back to front based on the sorted features. The SVM was used as the classifier to screen the features of the collected dehumidifier-related data. The optimal feature subset was selected, and the adaptive coefficient was obtained through the adaptive feature operator extraction algorithm. Based on the optimal feature subset and adaptive coefficient, the CT-SVM algorithm is used to predict the water level of the dehumidifier; The obtaining of the adaptive coefficient comprises: Based on the historical data of the optimal feature subset, the random forest algorithm is used to calculate the prediction error α of each sample i , where α i =|Y i -yβ i |, where i is the sample number, Y i is the actual water level change rate of the i-th sample, y is the predicted water level change rate of the water tank, β i is the adaptive coefficient of the i-th sample; If the prediction error is greater than the set threshold, the sample weight is adjusted and the prediction adaptive coefficient is retrained; if the prediction error is less than or equal to the set threshold, the training is completed and the final adaptive coefficient is obtained; The water level prediction of the dehumidifier using the CT-SVM algorithm includes: Using compressed tracking CT technology to dynamically and in real time sort the features in the optimal feature subset based on a tree model, screen out features whose importance is lower than an importance threshold, and obtain an optimal low-dimensional feature subset; The adaptive coefficient is combined with the optimal low-dimensional feature subset to construct an enhanced input vector, which is input into a nonlinear support vector machine classification model. The water level change rate of the water tank is divided into multiple types as output, the volume of the water tank is determined according to the product model, and the current water volume of the water tank and the expected time to be full are calculated.

2. The dehumidifier water level prediction method based on CT-SVM according to claim 1 is characterized in that: The selecting of the optimal feature subset comprises: S201, in the Filter stage, calculating the relief weight of each feature for the original features in the dehumidifier-related data; S202, sorting the original features in the dehumidifier-related data in descending order according to the calculated relief weight of each feature to construct a feature sequence; S203, in the Wrapper stage, the SBS search strategy is used to remove one feature X from the feature sequence from the back to the front, and SVM is used as the classifier to calculate the detection rate and false detection rate of the remaining feature subsets and the feature evaluation value J. If J is satisfied after removing feature X, 剔除Xk Greater than J 剔除Xk+1 , then repeat S203 until J 剔除Xk Less than or equal to J 剔除Xk+1 The remaining features in the current feature sequence are determined as the optimal feature subset, where k is the kth feature removed from the feature sequence.

3. The dehumidifier water level prediction method based on CT-SVM according to claim 2 is characterized in that: The classification decision function of the nonlinear support vector machine classification model is: ； Where N is the number of augmented input vectors, is the optimal solution of the Lagrange multiplier, is the category label of the i-th training sample, is the kernel function, representing the vector in the original input space and training sample vectors Mapping to high-dimensional feature space, is the intercept term of the classification hyperplane, is the category label of the jth training sample, is the kernel function, representing the i-th training sample and the jth training sample Inner product in high-dimensional feature space, is 1 / 5.

4. A dehumidifier water level prediction system based on CT-SVM, characterized in that: Including acquisition module, screening module and water level prediction module; The acquisition module is used to collect historical dehumidifier related data, including indoor environment data, outdoor environment data, operating status data, time and region data and water tank water level change rate; The screening module is used to adopt a hybrid feature selection method that combines the filter and wrapper methods, uses the detection rate and false detection rate as the evaluation basis for feature selection, uses the relief method to sort the features by relevance, adopts the SBS search strategy from back to front according to the sorted features, uses the support vector machine (SVM) as the classifier, performs feature screening on the collected dehumidifier-related data, selects the optimal feature subset, and obtains the adaptive coefficient through the adaptive feature operator extraction algorithm; The water level prediction module is used to predict the water level of the dehumidifier using the CT-SVM algorithm based on the optimal feature subset and the adaptive coefficient; The obtaining of the adaptive coefficient comprises: Based on the historical data of the optimal feature subset, the random forest algorithm is used to calculate the prediction error α of each sample i , where α i =|Y i -yβ i |, where i is the sample number, Y i is the actual water level change rate of the i-th sample, y is the predicted water level change rate of the water tank, β i is the adaptive coefficient of the i-th sample; If the prediction error is greater than the set threshold, the sample weight is adjusted and the prediction adaptive coefficient is retrained; if the prediction error is less than or equal to the set threshold, the training is completed and the final adaptive coefficient is obtained; The water level prediction of the dehumidifier using the CT-SVM algorithm includes: Using compressed tracking CT technology to dynamically and in real time sort the features in the optimal feature subset based on a tree model, screen out features whose importance is lower than an importance threshold, and obtain an optimal low-dimensional feature subset; The adaptive coefficient is combined with the optimal low-dimensional feature subset to construct an enhanced input vector, which is input into a nonlinear support vector machine classification model. The water level change rate of the water tank is divided into multiple types as output, the volume of the water tank is determined according to the product model, and the current water volume of the water tank and the expected time to be full are calculated.

5. The dehumidifier water level prediction system based on CT-SVM according to claim 4 is characterized in that: The selecting of the optimal feature subset includes: S201, in the Filter stage, calculating the relief weight of each feature for the original features in the dehumidifier-related data; S202. Sort the original features in the dehumidifier-related data in descending order according to the calculated relief weight of each feature to construct a feature sequence; S203, in the Wrapper stage, the SBS search strategy is used to remove one feature X from the feature sequence from the back to the front, and SVM is used as the classifier to calculate the detection rate and false detection rate of the remaining feature subsets and the feature evaluation value J. If J is satisfied after removing feature X, 剔除Xk Greater than J 剔除Xk+1 , then repeat S203 until J 剔除Xk Less than or equal to J 剔除Xk+1 The remaining features in the current feature sequence are determined as the optimal feature subset, where k is the kth feature removed from the feature sequence.

6. The dehumidifier water level prediction system based on CT-SVM according to claim 5 is characterized in that: The classification decision function of the nonlinear support vector machine classification model is: ； Where N is the number of augmented input vectors, is the optimal solution of the Lagrange multiplier, is the category label of the i-th training sample, is the kernel function, representing the vector in the original input space and training sample vectors Mapping to high-dimensional feature space, is the intercept term of the classification hyperplane, is the category label of the jth training sample, is the kernel function, representing the i-th training sample and the jth training sample Inner product in high-dimensional feature space, is 1 / 5.

Images