An automotive air pump with a prediction function and its prediction method

Abstract

The present invention discloses an automobile inflator with a prediction function and its prediction method, which relates to the field of automobile inflators. The method includes: collecting relevant inflation data and historical inflation data information during the inflation process through the inherent information and usage status of automobile tires, performing tire performance state recognition, and predicting the change in automobile tire pressure based on a double-layer prediction model according to the recognition result of the automobile tire performance state and environmental conditions, and making an inflation forecast in advance. It realizes convenient tire pressure prediction and forecast, and notifies the vehicle owner to inflate in advance. Moreover, by introducing the double-layer prediction method, accurate prediction is achieved; at the same time, a data augmentation method is introduced, and the augmented data set is corrected by calculating the similarity between data, so as to increase the amount of data for algorithm learning, and further realize the accuracy of prediction.

Description

Technical Field

[0001] The invention relates to the field of automobile air pumps, in particular to an automobile air pump with a prediction function and a prediction method thereof. Background Art

[0002] With the development of modern society, the number of cars is increasing. Tire pressure is closely related to driving safety. The portable car air pump has become an essential tool for car owners, especially on highways, where emergencies are difficult to predict. The car air pump can not only inflate the tires at any time, but also inflate the vehicle in emergency situations.

[0003] In the prior art, the automobile air pump can only realize the inflation function simply, and cannot predict the automobile tire pressure, thereby reminding the user to inflate in advance. In addition, the automobile tire pressure data is closed data. If the sensor is not installed, the tire pressure data cannot be obtained. The present invention aims to realize a convenient automobile tire pressure prediction method to remind the user to inflate in advance. Summary of the invention

[0004] In order to solve the technical problem that a portable automobile air pump in the prior art cannot predict tire pressure, the present invention provides an automobile air pump with a prediction function and a prediction method thereof.

[0005] The present invention is achieved through the following technical solutions:

[0006] A prediction method for an automobile air pump with a prediction function, comprising:

[0007] S1: determining the inherent information and usage status of the automobile tire through the image and the user input information; specifically, the inherent information includes the brand, model and service life of the automobile tire, which is input through the automobile air pump input interface; and photographing the vehicle tire usage status through the camera set on the air pump;

[0008] S2: During the inflation process, relevant inflation data is collected and saved, and historical inflation data information is obtained;

[0009] The relevant inflation data collected during the inflation process include the flow rate during the inflation process and the time series of tire pressure changes;

[0010] The historical inflation data information includes the flow rate of the inflation process of the historical inflation, the tire pressure change time series, and the tire pressure reduction rate between two inflations;

[0011] S3: Identify tire performance status by combining tire inherent information, usage status, and historical inflation data information;

[0012] S4: Based on the double-layer prediction model, the tire pressure change is predicted according to the tire performance status identification results and environmental conditions, and inflation forecast is made in advance.

[0013] Furthermore, the step S1 also includes processing the collected vehicle tire image information and identifying the tire usage status, including obtaining tire image information, image preprocessing, feature extraction and recognition, and usage status calculation; determining the tire usage status based on the identified features, and presenting the usage status output result in a numerical form.

[0014] Furthermore, the step S2 further includes: if the amount of historical inflation data information of the inflation pump device is less than a set threshold, data expansion is performed, and the data expansion method includes:

[0015] The inflation history data of other vehicles is obtained from the cloud platform, and similarity matching is performed on the tire inherent information, usage status, tire pressure reduction rate between two inflations, flow rate during the inflation process, and tire pressure change time series. The inflation history data of other vehicles obtained from the cloud platform is corrected based on the similarity matching results, and the corrected data is expanded to the historical inflation data information.

[0016] Furthermore, the correction method includes calculating an adjustment coefficient based on the similarity, and correcting the flow rate of the inflation process, the tire pressure change time series, and the tire pressure reduction rate between two inflations based on the adjustment coefficient; obtaining the adjustment coefficient includes obtaining it from a set database, or generating it by a neural network.

[0017] Furthermore, step S3 performs tire performance status identification based on a random forest model, wherein the model input includes tire inherent information, usage status, and expanded historical inflation data information, and the output is the tire performance status.

[0018] Furthermore, the two-layer prediction model of step S4 is a Prophet-LSTM prediction model, the first layer prediction model performs time series prediction based on the Prophet prediction model, and the second layer prediction model is based on the LSTM neural network model.

[0019] Furthermore, the first-layer prediction model inputs the tire pressure historical data set to obtain the prediction results output by the first-layer prediction model, that is, the first predicted tire pressure change time series. The second-layer prediction model inputs the vehicle tire performance status identification results, environmental conditions and the results of the first-layer prediction into the model, adjusts the model parameters to minimize the error of the error term, and outputs the second predicted tire pressure change time series.

[0020] Furthermore, the step S4 also includes: predicting the next inflation time according to the predicted tire pressure change time series obtained by the prediction model, and making a forecast when the time to the next inflation is less than a set threshold.

[0021] The present invention also provides an automobile air pump with a prediction function, based on the prediction method of the automobile air pump as described above, which comprises:

[0022] A camera and a display screen, wherein the camera is used to photograph the use of vehicle tires, and the display screen input interface is used to input the brand, model and age of the vehicle tires;

[0023] A data storage module, which is used to obtain relevant inflation data collected during the inflation process; the relevant inflation data collected during the inflation process includes the flow rate during the inflation process and the tire pressure change time series;

[0024] A historical inflation data information acquisition module, which is used to acquire historical inflation data information, wherein the historical inflation data information includes the flow rate of the inflation process of historical inflation, the tire pressure change time series, and the tire pressure reduction rate between two inflations, wherein the tire pressure reduction rate is the quotient of the tire pressure drop value between two inflations and the interval time between the two inflations;

[0025] A data expansion module is used to obtain inflation history data of other vehicles from the cloud platform, perform similarity matching on tire inherent information, usage status, tire pressure reduction rate between two inflations, flow rate during the inflation process, and tire pressure change time series, and correct inflation history data of other vehicles obtained from the cloud platform based on the similarity matching results and expand the corrected data into the historical inflation data information;

[0026] A tire performance status recognition module, which is used to recognize tire performance status based on a random forest model combined with tire inherent information, usage status, and historical inflation data information;

[0027] The tire pressure prediction module is used to predict the changes in tire pressure based on the double-layer prediction model, the tire performance status recognition results and environmental conditions, and to make inflation forecasts in advance.

[0028] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which program instructions of a prediction method for a vehicle air pump are stored. The program instructions based on the prediction method for a vehicle air pump can be executed by one or more processors to implement the steps of the prediction method for a vehicle air pump as described above.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The present invention realizes convenient tire pressure prediction and forecasting without the need for additional sensors, and can notify the owner of inflation in advance. In addition, a double-layer prediction method is introduced in the implementation method, and a prediction method based on Prophet-LSTM is proposed to achieve accurate prediction; at the same time, in view of the problem that artificial intelligence learning requires a large amount of data and the amount of data in reality is small, a data expansion method is introduced, and the expanded data set is corrected by calculating the similarity between the data, thereby increasing the amount of data for algorithm learning and further achieving the accuracy of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0032] Figure 1 It is a flow chart of a prediction method of a vehicle air pump with a prediction function according to an embodiment of the present application;

[0033] Figure 2 It is a schematic diagram of a double-layer prediction model according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0035] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0036] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention. The illustrations only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the form, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0037] See also Figure 1 , a prediction method for an automobile air pump with a prediction function, comprising the following steps:

[0038] S1: Determine the inherent information and usage status of the automobile tire through the image and user input information;

[0039] Specifically, the inherent information includes the brand, model and service life of the automobile tire, which is input through the automobile air pump input interface; the vehicle tire usage is photographed by a camera set on the air pump;

[0040] Furthermore, the collected vehicle tire image information is processed to identify the tire usage status, including the following steps:

[0041] a. Obtain tire image information; optionally, use a camera provided on the air pump to photograph the tire usage of the vehicle;

[0042] b. Image preprocessing: The original images collected often have problems such as noise, blur, and occlusion, and need to be preprocessed to improve the recognition accuracy. Preprocessing methods include denoising, deblurring, lane changing, and segmenting the region of interest;

[0043] c. Feature extraction and identification, including the location, size and number of cracks and bulges, pattern depth, and degree of grain deformation; optionally, feature extraction based on a CNN network.

[0044] d. Usage status calculation: determine the tire usage status based on the identified features;

[0045] Get the usage status output results, presented in numerical form. Optionally, set the maximum value of the usage status to 100 and the minimum value to 0. The smaller the value, the better the usage status.

[0046] S2: During the inflation process, relevant inflation data is collected and saved, and historical inflation data information is obtained;

[0047] The relevant inflation data collected during the inflation process include the flow rate during the inflation process and the time series of tire pressure changes;

[0048] The historical inflation data information includes the inflation flow rate of the historical inflation process, the tire pressure change time series, and the tire pressure reduction rate between two inflations. The tire pressure reduction rate is the quotient of the tire pressure drop value between two inflations and the interval time between the two inflations.

[0049] If the amount of historical inflation data information of the inflation pump device is less than a set threshold, in order to achieve the accuracy of the subsequent algorithm, the present invention performs data expansion, and the data expansion method includes:

[0050] Obtain inflation history data of other vehicles from the cloud platform to perform similarity matching on tire inherent information, usage status, tire pressure reduction rate between two inflations, and flow rate and tire pressure change time series during the inflation process;

[0051] The similarity calculation method is as follows:

[0052]

[0053] Among them, k 1 , k 2 , k 3 , k 4 , k 5 They are four similarity adjustment coefficients, and k 1 +k 2 +k 3 +k 4 +k 5 =1;Sim 1 is the tire inherent information similarity, Sim 2 is the usage status similarity, Sim 3 is the similarity of tire pressure reduction rate between two inflations, Sim 4 is the similarity of the flow change time series during the inflation process, Sim 5 is the similarity of the tire pressure change time series during the inflation process.

[0054] Among them, Sim 1 The calculation method includes obtaining the similarity coefficient s between different brands and models from the database p ; Similarity of service life , set to take only the data with a difference of m years in service life, n 1 、n 2 are the service life of the two tires of the vehicle and other vehicles being compared;

[0055] but, , where k p , k y are their respective weight coefficients.

[0056] Sim 2 The calculation is as follows:

[0057] , where d is the setting to only take data with a usage status difference of d. 1 d 2 They are the usage status of the two tires being compared. The smaller the value, the better the usage status.

[0058] Sim 3 and Sim 4 The calculation is as follows:

[0059]

[0060] Among them, q is the length of each time series, x i ,y i are the i-th values ​​of the two sequences respectively, and Softmax is the normalization function.

[0061] Correct the inflation history data of other vehicles obtained from the cloud platform based on the similarity matching results;

[0062] Optionally, the correction method includes calculating an adjustment coefficient based on the similarity, and correcting the flow rate of the inflation process, the tire pressure change time series, and the tire pressure reduction rate between two inflations based on the adjustment coefficient.

[0063] The obtaining of the adjustment coefficient includes obtaining it from a set database, or generating it by a neural network.

[0064] The corrected data is added to the historical inflation data information.

[0065] S3: Identify tire performance status by combining tire inherent information, usage status, and historical inflation data information;

[0066] Specifically, the tire performance status is identified based on a random forest model, wherein the model input includes tire inherent information, usage status, and expanded historical inflation data information, and the output is the tire performance status. Optionally, the performance status is represented by a quantitative value.

[0067] The random forest model uses Bootstrap resampling technology to generate multiple decision tree identifiers. The decision tree growth steps include:

[0068] a. Randomly sample from the original training sample data of size M using the replacement sampling method, repeat K times, and form a new training set M 1 , generate a recognition tree.

[0069] b. At each node of the recognition tree, randomly select N' features from the N input features. These N' features meet the principle of minimum node impurity, and N' will remain constant during the growth process. Select a feature for branch growth, and repeat the above process for each branch until the training set can be accurately identified or all branches meet the branch stopping rule.

[0070] The recognition tree follows the top-down splitting principle of the binary tree. The relevant formula for impurity is: Let R(n) be the impurity of node n. When all the recognition data belong to the same category, the impurity of node n is 0. If the recognition data is evenly distributed, the impurity is large. The calculation formula is:

[0071]

[0072] Among them, K is the total number of categories, p mk is the proportion of class k in node m.

[0073] After the recognition trees are generated, a random forest is formed and the recognition result of the algorithm is determined by voting.

[0074] S4: Based on the double-layer prediction model, the tire pressure change is predicted according to the tire performance status recognition results and environmental conditions, and inflation forecast is performed in advance, which specifically includes the following steps:

[0075] S41: Obtaining a vehicle tire performance status recognition result, environmental conditions, and tire pressure history data set;

[0076] Optionally, the environmental conditions include temperature and humidity information. The tire pressure history data set is obtained from a database.

[0077] S42: Establish a prediction model, such as Figure 2 As shown, prediction is made based on a double-layer Prophet-LSTM prediction model;

[0078] The first-level prediction model performs time series prediction based on the Prophet prediction model. The model is as follows:

[0079]

[0080] Among them, y(t) is the time series, g(t) is the trend term, s(t) is the periodic term, h(t) is the burst term, which is represented by normal distribution, and ε(t) is the error term, which obeys Gaussian distribution.

[0081] The trend item reflects the tendency of the tire pressure to gradually decrease over time. Optionally, a nonlinear logistic regression function is selected to predict the time trend of the tire pressure change:

[0082]

[0083] in, ， C(t) represents the predicted capacity, represents the upper limit of tire pressure change, s represents the sth trend change point, k represents the basic change rate, m represents the bias, δ represents the change in the change rate, a(t) represents the incremental change rate at timestamp t, and γ is the bias adjustment introduced to handle the boundary of the line segment.

[0084] The periodic term reflects the tire pressure change caused by the temperature change according to the season change. Optionally, the periodic change in the interval can be represented by the sine function and the cosine function. The expression is as follows:

[0085]

[0086] Among them, P is the period value of the time series, N is the total number of periods, and a n 、b n are the corresponding amplitudes of the cosine and sine functions.

[0087] The tire pressure history data set is input to obtain the prediction result output by the first-layer prediction model, that is, the first predicted tire pressure change time series.

[0088] The second-layer prediction model is based on the LSTM neural network model. The vehicle tire performance status recognition results, environmental conditions, and the first-layer prediction results are input into the model, and the model parameters are adjusted to minimize the error of the error term, and the second predicted tire pressure change time series is output.

[0089] The present invention introduces a two-layer prediction method and proposes a prediction method based on Prophet-LSTM. The first layer prediction uses the Prophet model to capture the characteristic items with relatively regular changes; on this basis, the second layer prediction uses the LSTM model to correct the first layer prediction results, thereby improving the accuracy of the overall prediction.

[0090] S43: Predict the next inflation time based on the prediction model and make a forecast.

[0091] The next inflation time is predicted based on the predicted tire pressure change time series obtained by the prediction model, and a forecast is made when the next inflation time is less than a set threshold. Optionally, the forecast information can be displayed on the inflation pump screen or sent to a mobile phone app.

[0092] In this embodiment, convenient tire pressure prediction and forecasting can be achieved without installing additional sensors, and the owner can be notified in advance to inflate the tire. In addition, a double-layer prediction method is introduced in the implementation method, and a prediction method based on Prophet-LSTM is proposed to achieve accurate prediction; at the same time, in view of the problem that artificial intelligence learning requires a large amount of data and the amount of data in reality is small, a data expansion method is introduced, and the expanded data set is corrected by calculating the similarity between the data, thereby increasing the amount of data for algorithm learning and further achieving the accuracy of the prediction.

[0093] The embodiment of the present invention further provides an automobile air pump with a prediction function, comprising:

[0094] A camera and a display screen, wherein the camera is used to photograph the use of vehicle tires, and the display screen input interface is used to input the brand, model and age of the vehicle tires;

[0095] A data storage module, which is used to obtain relevant inflation data collected during the inflation process; the relevant inflation data collected during the inflation process includes the flow rate during the inflation process and the tire pressure change time series;

[0096] A historical inflation data information acquisition module, which is used to acquire historical inflation data information, wherein the historical inflation data information includes the flow rate of the inflation process of historical inflation, the tire pressure change time series, and the tire pressure reduction rate between two inflations, wherein the tire pressure reduction rate is the quotient of the tire pressure drop value between two inflations and the interval time between the two inflations;

[0097] A data expansion module is used to obtain inflation history data of other vehicles from the cloud platform, perform similarity matching on tire inherent information, usage status, tire pressure reduction rate between two inflations, flow rate during the inflation process, and tire pressure change time series, and correct inflation history data of other vehicles obtained from the cloud platform based on the similarity matching results and expand the corrected data into the historical inflation data information;

[0098] A tire performance status recognition module, which is used to recognize tire performance status based on a random forest model combined with tire inherent information, usage status, and historical inflation data information;

[0099] The tire pressure prediction module is used to predict the changes in tire pressure based on the double-layer prediction model, the tire performance status recognition results and environmental conditions, and to make inflation forecasts in advance.

[0100] Optionally, the two-layer prediction model is a two-layer Prophet-LSTM prediction model.

[0101] In addition, an embodiment of the present invention also proposes a computer-readable storage medium, on which program instructions of a prediction method for a vehicle air pump with a prediction function are stored. The program instructions of the prediction method for a vehicle air pump with a prediction function can be executed by one or more processors to implement the steps of the prediction method for a vehicle air pump with a prediction function as described above.

[0102] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A prediction method for an automobile air pump with a prediction function, characterized in that: include: Step S1: determining the inherent information and usage status of the vehicle tire through the image and the user input information; specifically, the inherent information includes the brand, model and service life of the vehicle tire; photographing the usage status of the vehicle tire through a camera; Step S2: During the inflation process, relevant inflation data is collected and saved, and historical inflation data information is obtained; The relevant inflation data collected during the inflation process include the flow rate during the inflation process and the time series of tire pressure changes; The historical inflation data information includes the flow rate of the inflation process of the historical inflation, the tire pressure change time series, and the tire pressure reduction rate between two inflations; Step S3: Identify tire performance status by combining tire inherent information, usage status, and historical inflation data information; Step S4: Based on the double-layer prediction model, the change in the tire pressure of the vehicle is predicted according to the identification result of the tire performance status and the environmental conditions, and inflation forecast is performed in advance.

2. The prediction method for an automobile air pump with prediction function according to claim 1, characterized in that: The step S1 also includes processing the collected vehicle tire image information and identifying the tire usage status, including obtaining tire image information, image preprocessing, feature extraction and recognition, and usage status calculation; determining the tire usage status based on the identified features, and presenting the usage status output result in a numerical form.

3. The prediction method for an automobile air pump with prediction function according to claim 1, characterized in that: The step S2 further includes: if the amount of historical inflation data information of the inflation pump device is less than a set threshold, data expansion is performed, and the data expansion method includes: The inflation history data of other vehicles is obtained from the cloud platform, and similarity matching is performed on the tire inherent information, usage status, tire pressure reduction rate between two inflations, flow rate during the inflation process, and tire pressure change time series. The inflation history data of other vehicles obtained from the cloud platform is corrected based on the similarity matching results, and the corrected data is expanded to the historical inflation data information.

4. The prediction method for an automobile air pump with prediction function according to claim 3, characterized in that: The correction method includes calculating an adjustment coefficient based on similarity, and correcting the flow rate during the inflation process, the tire pressure change time series, and the tire pressure reduction rate between two inflations based on the adjustment coefficient; the adjustment coefficient is obtained from a set database or generated by a neural network.

5. The prediction method for an automobile air pump with prediction function according to claim 1, characterized in that: Step S3 identifies the tire performance status based on a random forest model, wherein the input of the random forest model includes tire inherent information, usage status, and expanded historical inflation data information, and the output is the tire performance status.

6. The prediction method for an automobile air pump with prediction function according to claim 1, characterized in that: The double-layer prediction model of step S4 is a Prophet-LSTM prediction model, the first-layer prediction model performs time series prediction based on the Prophet prediction model, and the second-layer prediction model is based on the LSTM neural network model.

7. The prediction method for an automobile air pump with prediction function according to claim 6, characterized in that: The first-layer prediction model inputs the tire pressure historical data set to obtain the prediction results output by the first-layer prediction model, that is, the first predicted tire pressure change time series. The second-layer prediction model inputs the vehicle tire performance status identification results, environmental conditions and the results of the first-layer prediction into the model, adjusts the model parameters to minimize the error of the error term, and outputs the second predicted tire pressure change time series.

8. The prediction method for an automobile air pump with prediction function according to claim 7, characterized in that: The step S4 also includes: predicting the next inflation time according to the predicted tire pressure change time series obtained by the prediction model, and making a forecast when the time to the next inflation is less than a set threshold.

9. An automobile air pump with a prediction function, characterized in that: The prediction method based on the automobile air pump with prediction function according to any one of claims 1 to 8 comprises: A camera and a display screen, wherein the camera is used to photograph the use status of vehicle tires, and the display screen input interface is used to input the brand, model and service life of the vehicle tires; A data storage module, which is used to obtain relevant inflation data collected during the inflation process; the relevant inflation data collected during the inflation process includes the flow rate during the inflation process and the tire pressure change time series; A historical inflation data information acquisition module, which is used to acquire historical inflation data information, wherein the historical inflation data information includes the flow rate of the inflation process of historical inflation, the tire pressure change time series, and the tire pressure reduction rate between two inflations, wherein the tire pressure reduction rate is the quotient of the tire pressure drop value between two inflations and the interval time between the two inflations; A data expansion module is used to obtain inflation history data of other vehicles from the cloud platform, perform similarity matching on tire inherent information, usage status, tire pressure reduction rate between two inflations, flow rate during the inflation process, and tire pressure change time series, and correct inflation history data of other vehicles obtained from the cloud platform based on the similarity matching results and expand the corrected data into the historical inflation data information; A tire performance status recognition module, which is used to recognize tire performance status based on a random forest model combined with tire inherent information, usage status, and historical inflation data information; The tire pressure prediction module is used to predict the changes in tire pressure based on the double-layer prediction model, the tire performance status recognition results and environmental conditions, and to make inflation forecasts in advance.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions of a prediction method for a vehicle air pump with a prediction function, and the program instructions of the prediction method for a vehicle air pump with a prediction function can be executed by one or more processors to implement the prediction method for a vehicle air pump with a prediction function as described in any one of claims 1 to 8.

Images