A method for predicting the preheating time of vehicle drying equipment

Abstract

The present invention provides a method for predicting the preheating time of vehicle drying equipment, and relates to the field of vehicle drying equipment. The technical problem solved by the present invention is how to improve the accuracy of the preheating time prediction of the drying equipment. The present invention comprises the following steps: establishing a preheating time prediction model by means of a random forest regression algorithm, obtaining current predicted preheating time parameters of the drying equipment including external temperature data, residual temperature data of the drying equipment, and startup data of the drying equipment, screening the current predicted preheating time parameters of the drying equipment by means of association merging, substituting the screened parameters into the preheating time prediction model, and outputting the current predicted preheating time value of the drying equipment, and establishing a preheating time prediction model to improve the accuracy of the predicted preheating time.

Description

Technical Field

[0001] The present invention relates to the field of vehicle drying equipment, and in particular to a method for predicting the preheating time of vehicle drying equipment. Background Art

[0002] Before the start of factory production, the vehicle drying equipment needs to be preheated to the manufacturing temperature. Currently, it is highly dependent on experience to estimate the heating time, which often has large errors. As a result, production personnel have to wait for the equipment to be preheated before going to work, or the equipment has to be preheated before waiting for production personnel to go to work, which will cause waste of manpower and energy in the factory.

[0003] In order to solve the above problems, some people have proposed to solve them by establishing a model through big data, such as the intelligent drying remote control method and system based on cloud platform big data recommendation disclosed in Chinese patent application No. 201610885639.1. This method forms an influencing factor matrix based on the collected humidity of the raw materials to be dried, the indoor and outdoor temperature and humidity curves of the drying room, the drying time, and the images of the raw materials to be dried, and uploads it to the cloud server, wherein the indoor and outdoor temperature and humidity curves of the drying room and the drying time are decision variables, and an indicator matrix is ​​formed according to the product drying degree, uniformity, total time consumption, and energy consumption samples collected under the influence of different decision variables, and the Elman neural network is used to train and test the indicator matrix, and a dynamic drying model is established. Finally, the dynamic drying model is used to predict the real-time data to obtain the current optimal indoor temperature curve, humidity curve, and drying time.

[0004] The above method establishes an indicator matrix as the data source for building the model. The indicator matrix is ​​a quantitative indicator to measure the effect of controlling the decision-making process, that is, the proportion of one of the influencing factors such as the product drying degree in the above method on the indoor and outdoor temperature curves of the drying room in all influencing factors. Because the above method does not screen the data when constructing the indicator matrix according to the influencing factor matrix, there will be errors between the proportion of each influencing factor in the constructed indicator matrix and the actual proportion, which will affect the accuracy of model establishment, resulting in inaccurate prediction results, which may lead to a decrease in production efficiency. Summary of the invention

[0005] In view of the above problems existing in the prior art, the present invention provides a method for predicting the preheating time of a vehicle drying device, and the technical problem solved by the present invention is how to improve the accuracy of the prediction of the preheating time of the drying device.

[0006] The present invention is implemented by the following technical scheme: a method for predicting the preheating time of a vehicle drying device, the method comprising the following steps:

[0007] A warm-up time prediction model was established using the random forest regression algorithm;

[0008] Obtain current predicted drying equipment preheating time parameters including outside air temperature data, drying equipment residual temperature data, and drying equipment startup data;

[0009] The current predicted drying equipment preheating time parameters are screened by including associated merging;

[0010] Substitute the filtered parameters into the preheating time prediction model and output the current drying equipment preheating time prediction value.

[0011] A preheating time prediction model is established through the random forest regression algorithm. After the model is established, the current predicted drying equipment preheating time parameters including the outside temperature data, the residual temperature data of the drying equipment and the drying equipment startup data are obtained. After obtaining the current predicted drying equipment preheating time parameters, the current predicted drying equipment preheating time parameters are filtered by means of association merging, and the filtered parameters are substituted into the preheating time prediction model to output the current drying equipment preheating time prediction value.

[0012] The present invention improves the method for predicting the preheating time of vehicle drying equipment by screening the obtained predicted preheating time parameters of drying equipment in a manner including association merging. After the model is established, the obtained current preheating time parameters of drying equipment are screened in a manner including association merging to ensure that the data is representative and available data, so that the drying equipment preheating time prediction value obtained after being substituted into the model has higher authenticity and accuracy, thereby improving the accuracy of the drying equipment preheating time prediction. The present invention also improves the method for predicting the preheating time of vehicle drying equipment by establishing a preheating time prediction model through a random forest regression algorithm. The preheating time prediction model established by the random forest regression algorithm has greatly improved accuracy than the preheating time prediction value obtained through experience, thereby improving the accuracy of the drying equipment preheating time prediction.

[0013] In the above-mentioned method for predicting the preheating time of vehicle drying equipment, the above-mentioned step of establishing the preheating time prediction model includes obtaining the historical prediction drying equipment preheating time parameters of the outside air temperature data, the residual temperature data of the drying equipment, and the startup data of the drying equipment, screening the historical prediction drying equipment preheating time parameters by means of association merging, and establishing the preheating time prediction model through the random forest regression algorithm for the screened parameters. The random forest regression algorithm has high accuracy, simple implementation, and fast speed, which can improve the accuracy of the establishment of the preheating time prediction model, thereby improving the accuracy of the drying equipment preheating time prediction.

[0014] In the above-mentioned method for predicting the preheating time of vehicle drying equipment, the above-mentioned operation steps of screening the current predicted drying equipment preheating time parameters by means of association merging and screening the historical predicted drying equipment preheating time parameters by means of association merging both include screening the current or historical predicted drying equipment preheating time parameters to obtain useful data, performing data cleaning on the useful data to obtain valid data, and associating and merging the valid data to obtain a prediction associated data table. Multiple screening of data can improve the accuracy of model establishment and the accuracy of input data, thereby improving the accuracy of drying equipment preheating time prediction.

[0015] In the above method for predicting the preheating time of vehicle drying equipment, the above screening of the current or historical predicted preheating time parameters of the drying equipment to obtain useful data includes screening the predicted preheating time parameters of the drying equipment according to correlation analysis and selecting useful data with high correlation with the heating time, including outside air temperature data, drying equipment residual temperature data, and drying equipment startup data. The screening of useful data can remove data that is irrelevant to the preheating time of the drying room, improve the accuracy of model establishment and the accuracy of input data, and thus improve the accuracy of the prediction of the preheating time of the drying equipment.

[0016] In the above method for predicting the preheating time of vehicle drying equipment, the above data cleaning of useful data to obtain valid data includes comparing the useful data with the standard data after obtaining the useful data, and obtaining valid data including outside temperature data, drying equipment residual temperature data and drying equipment startup data after removing the data different from the standard data. Data cleaning can detect and correct unreasonable parts in the data, improve the accuracy of model establishment and the accuracy of input data, and thus improve the accuracy of drying equipment preheating time prediction.

[0017] In the above method for predicting the preheating time of the vehicle drying equipment, the above-mentioned associating and merging the valid data to obtain the prediction associated data table includes associating the valid data including the outside temperature data, the residual temperature data of the drying equipment and the start data of the drying equipment at the same time to form a prediction associated data group, and then sorting all the prediction associated data groups in chronological order to form a prediction associated data table. Data association merging facilitates data integration and can improve calculation efficiency, thereby improving the efficiency of model establishment.

[0018] In the above method for predicting the preheating time of vehicle drying equipment, the outside air temperature data includes outside temperature, outside humidity, outside enthalpy value and outside moisture content data, the residual temperature data of the drying equipment includes the return air temperature data of the first to sixth stages, and the startup data of the drying equipment includes whether production was carried out the previous day and the historical heating time data. The outside air temperature data includes the outside temperature, outside humidity, outside enthalpy value, outside moisture content, the return air temperature of the first to sixth stages, whether production was carried out the previous day and the historical heating time, which are all data with high correlation with the preheating time, and these data are helpful to establish a more accurate preheating time prediction model.

[0019] In the above method for predicting the preheating time of vehicle drying equipment, the preheating time prediction model is

[0020]

[0021] where f t (x) is the current prediction-related data set, T is the historical heating time, and f(x) is the preheating time prediction value. When the current prediction-related data set is obtained and substituted into the preheating time prediction model, the current drying equipment preheating time prediction value is obtained according to the average of the historical heating time at the same time. By establishing a preheating time prediction model, it is convenient to predict the current heating time, and the accuracy is higher than that of empirical prediction.

[0022] Compared with the prior art, the method for predicting the preheating time of vehicle drying equipment has the following advantages:

[0023] 1. Screening the obtained predicted drying equipment preheating time parameters reduces data errors, improves the accuracy of model establishment and thus improves the accuracy of preheating time prediction.

[0024] 2. The present invention establishes a preheating time prediction model, which can accurately and efficiently predict the preheating time, thereby improving the accuracy of the preheating time prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0026] In order to make the technical problems, technical solutions and advantages solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0027] like Figure 1 As shown, the method for predicting the preheating time of vehicle drying equipment includes the following steps:

[0028] A preheating time prediction model is established through the random forest regression algorithm. The steps of establishing the preheating time prediction model include obtaining historical predicted drying equipment preheating time parameters including external temperature data, drying equipment residual temperature data and drying equipment startup data, screening the historical predicted drying equipment preheating time parameters by means of association merging, and using the screened parameters to establish the preheating time prediction model through the random forest regression algorithm.

[0029] The operation steps of screening the historical predicted drying equipment preheating time parameters by means of association merging include screening the historical predicted drying equipment preheating time parameters to obtain useful data, performing data cleaning on the useful data to obtain valid data, and associating and merging the valid data to obtain a prediction associated data table.

[0030] Screening the historical predicted drying equipment preheating time parameters to obtain useful data includes screening the predicted drying equipment preheating time parameters according to correlation analysis and then selecting useful data with a high correlation with the heating time. In this embodiment, the data with a high correlation with the heating time can be selected based on experience. The highly correlated data include whether there was production the day before, the return air temperature of stages one to six, the outside temperature, the outside humidity, the outside enthalpy value, and the outside moisture content. Screening useful data can reduce the processing of unnecessary data and improve the efficiency of model building.

[0031] Data cleaning of useful data to obtain valid data includes comparing the useful data with standard data after obtaining the useful data, and obtaining valid data after removing the data different from the standard data. The standard data is preset and is set for the convenience of data processing for all the data. Cleaning data includes cleaning missing values, format content and logic errors of the obtained useful data in sequence to obtain valid data. The existing cleaning of missing values ​​is to compare the format of the obtained data with the format of the standard data, complete the important data, remove unnecessary fields and remove the erroneous important data, and then re-obtain the important data. The existing format content cleaning includes unifying the display format, removing non-existent characters and adjusting the content that does not conform to the field format. The existing logical error cleaning includes deduplication, removal of unreasonable values ​​and correction of inconsistent content. Data cleaning is to remove non-conforming data to avoid affecting data processing and improve the accuracy of model establishment. Important data refers to data necessary for model establishment, which in this embodiment includes whether production was produced the day before, historical heating time, return air temperature from one to six stages, external temperature, external humidity, external enthalpy value and external moisture content and historical heating time data.

[0032] The effective data are associated and merged to obtain a prediction associated data table, including associating the effective data of the same time, including whether production occurred the previous day, the historical heating time, the return air temperature of stages one to six, the outside temperature, the outside humidity, the outside enthalpy value and the outside moisture content, to form a prediction associated data group, and then sorting each prediction associated data group in chronological order to form a prediction associated data table.

[0033] In this implementation, the associated data table is as follows:

[0034]

[0035] The above historical heating time indicates the time required for the drying equipment to start up to the manufacturing temperature, for example, from 20 degrees Celsius to 100 degrees Celsius, in minutes. Whether there was production the day before indicates whether there was production the day before. There are only two cases, 1 represents production the day before, and 0 represents no production the day before. Whether there was production the day before will be automatically recorded and uploaded. The drying equipment is divided into multiple sections for assembly line operation. The return air temperature in the prior art refers to the outlet temperature of the air flow after heat exchange. The first section return air temperature indicates the outlet temperature of the air flow of the drying equipment in the first section after heat exchange, and so on. The second to sixth sections return air temperatures indicate the second to sixth sections return air temperatures. The outlet temperature of the air flow of the six-stage area drying equipment after heat exchange is in degrees Celsius. The outside temperature indicates the temperature of the environment around the drying equipment in degrees Celsius. The outside humidity indicates the humidity of the environment around the drying equipment in percentage. The outside enthalpy value indicates the total heat contained in the air in kilojoules or kilojoules per kilogram. The outside moisture content indicates the mass of water vapor mixed in every kilogram of dry air in grams. From the above, it can be seen that when the return air temperature of sections 1 to 6 is 20-30 degrees Celsius, it means that no production was carried out the day before. When the return air temperature of sections 1 to 6 is generally 40-60 degrees Celsius, it means that production was carried out the day before.

[0036] After establishing the preheating time prediction model, the current predicted drying equipment preheating time parameters including the outside air temperature data, the drying equipment residual temperature data and the drying equipment startup data are obtained. The predicted drying equipment preheating time parameters include the drying equipment humidity, the drying equipment energy consumption, whether there was production the day before, the return air temperature of the first to sixth stages, the outside temperature, the outside humidity, the outside enthalpy value, the outside moisture content and the drying equipment heating time.

[0037] Screen the current predicted preheating duration parameter of the drying equipment. The operation steps of screening the current predicted preheating duration parameter of the drying equipment by means of including correlation and merging include screening the predicted preheating duration parameter of the drying equipment to obtain useful data, cleaning the useful data to obtain valid data, and performing correlation and merging on the valid data to obtain a predicted correlation data table. The specific steps of screening the current predicted preheating duration parameter of the drying equipment are the same as those of screening the above-mentioned historical predicted preheating duration parameter of the drying equipment. The only difference is that the column of historical heating duration in the predicted correlation data table established this time is converted into the current heating duration and is blank. The current heating duration is the predicted value of the current preheating duration of the drying equipment to be obtained.

[0038] The steps of establishing a preheating duration prediction model by the random forest regression algorithm include:

[0039] Obtain valid data, use the valid data as the training data set D required for model training, obtain the number M of data including the valid data, and specify the number T of trees in the random forest and the maximum number Q of nodes of each tree;

[0040] Randomly draw N samples from the training data set D with replacement, and then randomly select m (m << M) variables from the M input variables of these N samples to form a new training data set D k , Randomly drawing with replacement improves the randomness of sample extraction, making it possible for the sample to draw the same data, and improving the accuracy of model establishment;

[0041] Traverse the training set D k Among the m variables, select the optimal splitting variable j and splitting point s, that is, the pair (j, s) that makes the value obtained by the following formula reach the minimum. Traversing means following a certain search route and visiting each node of the tree once in turn. Without missing data, it is guaranteed that the accurate minimum pair found improves the accuracy of model establishment.

[0042]

[0043] where x i is the input variable, y i is the output variable, R1(j, s) and R2(j, s) are two regions divided by the pair (j, s) of the training data set D k , c1 is the mean value of the output variable y i in the region R1(j, s), and c2 is the mean value of the output variable y i in the region R2(j, s);

[0044] Use the pair (j, s) obtained above to divide the region and determine the corresponding output value:

[0045] R1(j,s)={x|x (j) ≤s}, R2(j,s)={x|x (j) >s}

[0046]

[0047] Determine whether the stopping condition for generating a decision tree is met. If the total number of nodes in the generated tree is equal to the maximum number of nodes Q, the stopping condition for generating a decision tree is met. If the stopping condition for generating a decision tree is not met, re-traverse the training data set D. k m variables are added and the steps of establishing the warm-up time prediction model are repeated until the stop condition of generating the decision tree is met. If there are too many nodes, more resources are required but the same conclusion is drawn, resulting in a waste of resources. If there are too few nodes, the required conclusion cannot be obtained or there is a deviation from the required conclusion, resulting in insufficient prediction accuracy. Therefore, judging whether the stop condition of generating the decision tree is met can improve the accuracy of establishing the warm-up time prediction model.

[0048] Output value Divided into L t Regions Generate the tth decision tree:

[0049]

[0050] Where I is the polling period for selecting sample N, and the polling period is the access period. For example, if there are 10 data, I is 3, and starting from the third data, 6, 9 and 2 are selected.

[0051] Determine whether the stopping condition for generating the warm-up time prediction model is met. If the number of decision trees t generated is equal to the number of random forest maximum decision trees T, the stopping condition for generating the warm-up time prediction model is met. If the stopping condition for generating the warm-up time prediction model is not met, a new data set D is reconstructed. k Repeat the steps of establishing the preheating time prediction model until the stop condition of generating the preheating time prediction model is met. If there are too many decision trees, more resources are required but the same conclusion is drawn, resulting in a waste of resources. If there are too few decision trees, the required conclusion cannot be obtained or there is a deviation from the required conclusion, resulting in insufficient prediction accuracy. Therefore, judging whether the stop condition of generating the preheating time prediction model is met can improve the accuracy of establishing the preheating time prediction model.

[0052] The mean of the outputs of all generated regression decision trees is taken as the final output result, and the warm-up time prediction model is established as follows:

[0053]

[0054] where f t(x) is the current prediction-related data set, T is the historical heating time, and f(x) is the predicted value of the preheating time.

[0055] Substitute the filtered parameters into the preheating time prediction model and output the current drying equipment preheating time prediction value. When the current prediction associated data group is obtained and substituted into the preheating time prediction model, the current drying equipment preheating time prediction value is obtained according to the average of the historical heating time at the same time.

[0056] The present invention improves the method for predicting the preheating time of vehicle drying equipment by screening the obtained predicted preheating time parameters of the drying equipment in a manner including association merging. After the model is established, the obtained current preheating time parameters of the drying equipment are screened in a manner including association merging to ensure that the data is representative and available data, so that the drying equipment preheating time prediction value obtained after being substituted into the model has higher authenticity and accuracy, thereby improving the accuracy of the drying equipment preheating time prediction. The present invention also improves the method for predicting the preheating time of vehicle drying equipment by establishing a preheating time prediction model through a random forest regression algorithm. The preheating time prediction model established through the random forest regression algorithm greatly improves the accuracy of the preheating time prediction value obtained through experience, thereby improving the accuracy of the drying equipment preheating time prediction.

[0057] The specific embodiments described herein are merely examples of the present invention. A person skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner without departing from the spirit of the present invention or exceeding the scope defined by the appended claims.

Claims

1. A method for predicting the preheating time of a vehicle drying equipment, characterized in that: The method comprises the following steps: A preheating time prediction model is established by using a random forest regression algorithm. The operation includes obtaining the historical prediction drying equipment preheating time parameters of the external temperature data, the residual temperature data of the drying equipment, and the startup data of the drying equipment. The historical prediction drying equipment preheating time parameters are screened by a method including association merging. The screened parameters are used to establish a preheating time prediction model by using a random forest regression algorithm. Obtain current predicted drying equipment preheating time parameters including outside air temperature data, drying equipment residual temperature data, and drying equipment startup data; The current predicted drying equipment preheating time parameters are screened by including associated merging; Substitute the filtered parameters into the preheating time prediction model and output the current drying equipment preheating time prediction value; The above-mentioned operation steps of screening the current predicted drying equipment preheating time parameters by means of association merging and screening the historical predicted drying equipment preheating time parameters by means of association merging both include: screening the current or historical predicted drying equipment preheating time parameters to obtain useful data, performing data cleaning on the useful data to obtain valid data, and associating and merging the valid data to obtain a prediction associated data table.

2. The method for predicting the preheating time of vehicle drying equipment according to claim 1, characterized in that: The above-mentioned screening of the current or historical predicted drying equipment preheating time parameters to obtain useful data includes screening the predicted drying equipment preheating time parameters according to correlation analysis and then selecting useful data with high correlation with the heating time, including external temperature data, drying equipment residual temperature data and drying equipment startup data.

3. The method for predicting the preheating time of vehicle drying equipment according to claim 2, characterized in that: The above-mentioned data cleaning of useful data to obtain valid data includes comparing the useful data with the standard data after obtaining the useful data, and obtaining valid data including outside temperature data, drying equipment residual temperature data and drying equipment startup data after clearing the data different from the standard data.

4. The method for predicting the preheating time of vehicle drying equipment according to claim 3, characterized in that: The above-mentioned associating and merging of valid data to obtain a predicted associated data table includes associating valid data including external temperature data, drying equipment residual temperature data and drying equipment startup data at the same time to form a predicted associated data group, and then sorting all the predicted associated data groups in chronological order to form a predicted associated data table.

5. The method for predicting the preheating time of vehicle drying equipment according to any one of claims 1 to 4, characterized in that: The outside air temperature data includes outside temperature, outside humidity, outside enthalpy value and outside moisture content data; the drying equipment residual temperature data includes one to six return air temperature data; the drying equipment startup data includes whether there was production the previous day and historical heating time data.

6. The method for predicting the preheating time of vehicle drying equipment according to claim 5, characterized in that: The preheating time prediction model is: where f t (x) is the current prediction-related data group, T is the historical heating time, and f(x) is the preheating time prediction value. When the current prediction-related data group is obtained and substituted into the preheating time prediction model, the current drying equipment preheating time prediction value is obtained based on the average of the historical heating time at the same time.

Images