ANN-based community low-value recoverable delivery quantity prediction method

Abstract

The invention provides an ANN-based community low-value recoverable delivery quantity prediction method, and aims to optimize resource allocation and layout strategies of an intelligent recovery system. Historical delivery data and external feature data of low-value recoverable delivery sites covering a plurality of communities are collected, data preprocessing and training are performed to generate an ANN model, the model can capture a complex mode relationship of resident delivery behaviors, and short-term accurate prediction is provided for the delivery amount of each recoverable delivery site in the next 3-7 days. Besides, community comprehensive quality scores and convenience features are selected as two basis of community classification, and community differentiated clearance and resource scheduling strategy formulation is realized based on a K-means clustering method, so that the operation efficiency and sustainable development capability of the intelligent recovery system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of solid waste management and intelligent recycling systems, and in particular to a method for predicting the delivery volume of low-value recyclables in a community based on an ANN. Background Art

[0002] In recent years, the rapid development of cities and the improvement of residents' living standards have led to a sharp increase in garbage production, and garbage disposal and recycling are facing great pressure. Low-value recyclables refer to solid waste with low economic value and easy to be discarded, such as waste paper boxes, waste plastics, glass bottles, foam, etc. This type of solid waste can be seen everywhere in daily life, but its economic value is low and the recycling cost is relatively high. At the same time, the traditional recycling method of low-value recyclables is inefficient, with serious waste of resources, which has caused considerable pressure on the environment and resources. In response to the above problems, the intelligent recycling system based on the Internet of Things has been widely used in recent years. The system promotes residents to actively participate in recycling by providing a reward mechanism. By setting up intelligent garbage recycling bins in the community, residents can deliver low-value recyclables by themselves, realizing the automation of the recycling process. However, the intelligent recycling system usually uses a fixed time and space layout to optimize the recycling strategy, and it is difficult to achieve the optimal configuration in equipment layout and transportation scheduling. Therefore, how to use advanced data prediction algorithms to accurately predict the delivery volume of low-value recyclables has become a key link in the optimization management of intelligent recycling systems.

[0003] In the field of intelligent recycling systems, existing technologies still have many limitations in predicting the delivery volume of low-value recyclables. Although traditional prediction methods, such as ARIMA (Autoregressive Integrated Moving Average) models and simple linear regression models, have been widely used in some fields, these methods essentially assume that there is a linear relationship between data. However, the delivery volume of low-value recyclables is affected by many complex factors, such as weather, holidays, etc., and the relationship between these factors and the delivery volume is often nonlinear. Therefore, linear models show obvious deficiencies in dealing with such complex relationships, and it is difficult to capture the inherent patterns of the data, resulting in low prediction accuracy. At the same time, with the popularization of intelligent recycling systems, the amount of data and data dimensions continue to increase. Traditional methods are prone to "dimensionality disasters" when processing high-dimensional data, and the model complexity and computational cost increase significantly. At the same time, the prediction accuracy will also decrease due to data sparsity.

[0004] In summary, the existing technology has obvious shortcomings in predicting the delivery volume of low-value recyclables, mainly reflected in the insufficient prediction accuracy caused by the difficulty of traditional methods in capturing complex relationships and processing high-dimensional data. These shortcomings limit the optimization and promotion of intelligent recycling systems, and a new prediction method that can effectively handle complex nonlinear relationships is urgently needed. Summary of the invention

[0005] The purpose of the present invention is to provide a method for predicting the delivery volume of low-value recyclables in the community based on ANN, which uses ANN (Artificial Neural Network) to process and analyze community multi-site delivery data to achieve the prediction of daily delivery volume, so as to optimize the resource allocation and layout strategy of the intelligent recycling system, improve operational efficiency and promote the sustainable development of the community, and solve the problem of insufficient prediction accuracy of the existing intelligent recycling system in the optimization of layout and transportation strategy.

[0006] On the one hand, the present invention provides a method for predicting the delivery amount of low-value recyclables in a community based on ANN, comprising the following steps: Data collection: Obtain historical delivery data, external feature data, and community feature data of low-value recyclables delivery sites covering multiple communities; wherein the external feature data includes weather information, weekday status, and holiday promotion information, and the community feature data includes community economic characteristics and convenience characteristics; Data preprocessing: For historical delivery data and external feature data, Lagrange interpolation method is used to fill missing values, box plot method is used to detect and remove outliers, and Max-Min method is used for normalization to ensure data quality and consistency; Model establishment: Design a multi-level ANN structure, which includes an input layer, two hidden layers and an output layer. The input layer receives multi-dimensional feature vectors, the hidden layer uses the nonlinear activation function ReLU to capture the complex pattern relationship of residents' delivery behavior, and the output layer provides the prediction results of the daily delivery volume of specific low-value recyclables delivery stations. Model training: During the training process, the Adam optimizer is used to adjust the parameters of the ANN model, and the hyperparameter configuration is optimized through cross-validation. At the same time, a variety of prediction evaluation indicators are used to comprehensively evaluate the prediction accuracy of the ANN model. Multi-site forecasting: The sliding window method is used to generate time series samples, and the trained ANN model is used to independently forecast the delivery volume of each low-value recyclables delivery site in the next 3 to 7 days to obtain short-term forecast data; Community difference analysis: First, based on the community's economic characteristics, the comprehensive quality score of each community is calculated using the entropy method. Secondly, the convenience characteristics are combined and the K-means clustering algorithm is used to calculate the comprehensive quality score of the community. Finally, the community type is classified according to the score. Based on the classification results, according to historical delivery data and short-term forecast data, targeted resource scheduling strategies are formulated to optimize the frequency of transportation and facility layout.

[0007] Optionally, the method for obtaining external characteristic data and community characteristic data in the data collection is specifically to obtain detailed weather conditions, temperature, wind force level, community housing price, community housing age, community greening rate and community floor area ratio, as well as customized working days, holidays or promotional days from open data sources on the Internet.

[0008] Optionally, the data preprocessing also includes encoding conversion of text data or category data to facilitate model input.

[0009] Optionally, in the model establishment, the first hidden layer contains 256 neurons, the second hidden layer contains 128 neurons, and the initial learning rate is set to 0.001.

[0010] Optionally, the batch size of the training process in the model training is set to 32, and the maximum number of iterations is 500 to ensure that the model is fully trained while avoiding overfitting.

[0011] Optionally, the convenience feature is measured by the number of people served by a single device.

[0012] In a second aspect, the present invention also provides a computer device comprising a memory, a processor and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of the aforementioned ANN-based method for predicting the delivery volume of low-value recyclables in the community.

[0013] In a third aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned method for predicting the delivery volume of low-value recyclables in a community based on an ANN.

[0014] One or more technical solutions provided by the present invention have at least the following technical effects or advantages: First, the present invention covers the historical delivery data and external feature data of low-value recyclables delivery stations in multiple communities, forming a complete multi-dimensional feature matrix, including external feature variables such as weather information, weekday status, and holiday promotion information. In the model construction stage, a multi-level ANN structure is adopted, and the nonlinear activation function ReLU is used to capture the complex pattern of residents' delivery behavior, reducing the prediction error.

[0015] In addition, the ANN model developed by the present invention can effectively process high-dimensional data through feature input and parameter optimization, avoiding the "dimensionality disaster" and ensuring that the model can still maintain a high prediction accuracy under data sparsity conditions. The model uses a cross-validation method to optimize the model parameter relationship during the training process, and uses a variety of prediction evaluation indicators to comprehensively evaluate the prediction accuracy of the model. Finally, by predicting the delivery volume of each low-value recyclable delivery station in the next 3 to 7 days, it assists in adjusting the resource scheduling strategy, optimizing the frequency of transportation and facility layout, so as to improve operational efficiency and reduce resource waste.

[0016] Therefore, the present invention adopts an ANN-based prediction model to significantly improve the prediction accuracy of the delivery amount of low-value recyclables. Compared with other models in the prior art, the ANN model shows obvious advantages in prediction error. The MAPE value of the ANN model developed by the present invention is only 0.38, which is significantly lower than the LSTM (Long Short-Term Memory) model, random forest model, XGBoost (eXtreme Gradient Boosting) model, and SVR (Support Vector Regression) model. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a method for predicting the delivery volume of low-value recyclables in a community based on ANN in an embodiment of the present invention; Figure 2 A network structure diagram of the ANN model in an embodiment of the present invention; Figure 3 A comparison chart of the predicted delivery volume and the actual delivery volume of a specific low-value recyclables delivery site in an embodiment of the present invention; Figure 4 It is a bubble chart of the classification of five types of communities in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to overcome the limitations of traditional prediction methods, the present invention introduces ANN on the basis of traditional methods, and constructs a multi-site level ANN model that can predict the delivery volume by integrating historical delivery volume and external feature data (weather information, weekday status, holiday promotion information, etc.). The ANN model captures the complex pattern relationship of residents' delivery behavior through a multi-layer network, can provide high-precision prediction results, support dynamic optimization of resource allocation, and thus achieve efficient and sustainable operation management.

[0019] The present invention provides a method for predicting the delivery volume of low-value recyclables in a community based on ANN, which includes six steps: data collection, data preprocessing, model establishment, model training, multi-site prediction and community difference analysis.

[0020] Please refer to the following Figure 1 , Example 1 of the present invention is described in detail.

[0021] 1. Data collection; In the present invention, the historical delivery data of low-value recyclables from January 1, 2023 to March 31, 2024 at 436 low-value recyclables delivery stations covering 173 communities in a certain city are collected through IoT smart devices, mainly including the following core data: 1) Site number: used to identify each intelligent recycling device; 2) Daily delivery volume: the total weight of low-value recyclables received by each device every day, in kilograms; 3) Other residents’ delivery behavior characteristics: such as user number, street, community, delivery frequency, compensation received, etc.

[0022] In order to capture the external factors that affect delivery behavior, the present invention obtains external feature data in combination with open data sources. Among them, weather information is obtained from the "China Weather Network" through Internet crawlers, including descriptive information of weather conditions (such as sunny, cloudy, overcast, light rain, heavy rain, etc.), temperature (average value, unit is ℃) and wind force level (level 1-6). In addition, date information is also obtained through Internet crawlers, including whether it is a weekday, holiday or promotion day (such as "618", "Double 11", "Double 12"), as well as the periodic characteristics of years, months and seasons.

[0023] Finally, in order to apply the ANN model of the present invention to each community, it is necessary to realize community clustering based on community economic characteristics and convenience characteristics, so as to realize the difference analysis of the historical value and predicted value of the delivery volume of low-value recyclables in different categories of communities. The community economic characteristics and convenience characteristics are collected with 173 communities as the minimum analysis unit. The community economic characteristics are obtained from the Anjuke website through an Internet crawler, including community housing prices, community house age, community greening rate and community volume ratio. The convenience characteristics are calculated from the data provided by the community property, including the number of people in the community and the number of recycling stations laid in the community, and then the number of people served by a single device is obtained, and the convenience characteristics are measured using the number of people served by a single device. Finally, according to the comprehensive quality score and convenience characteristics of each community, the 173 communities are divided into 5 categories based on the K-means cluster analysis method.

[0024] After collection and processing, the individual external feature data and historical delivery volume data are integrated into a complete set of feature vectors, represented as X={x1,x2,…,x d}, where d is the number of features. Combined with the data entries, a multidimensional feature matrix D∈X is finally formed n×d, where n is the total number of records. Each record includes the daily delivery volume of low-value recyclable delivery stations and the input external feature variable data for that day.

[0025] 2. Data preprocessing; To ensure the quality of the model input data, a series of preprocessing is required for the collected data.

[0026] First, for the descriptive information of weather conditions collected (such as sunny, cloudy, overcast, light rain, heavy rain, etc.), it is converted into a numerical form during the data preprocessing stage. The specific coding rules are as follows: "1" represents sunny, cloudy, overcast or changing from cloudy to overcast; "2" represents fog + cloudy or fog + overcast; "3" represents light snow, light rain or shower; "4" represents heavy rain, rainstorm, moderate snow, heavy snow or blizzard. For the date feature, whether it is a working day, a festival or a promotional day is also numerically processed, where "yes" is marked as 0 and "no" is marked as 1. In the year feature, "2023" is marked as 0 and "2024" is marked as 1; the month is marked as 1 - 12; the season is marked as 1 - 4 according to the international astronomical division standard.

[0027] Secondly, missing values are filled. For the occasionally missing numerical values in weather information (such as precipitation or temperature), Lagrange interpolation method is used for time series completion.

[0028] For outliers (such as extreme values in daily delivery volume), the present invention detects abnormal data through the IQR (Interquartile Range) box plot method and eliminates or smooths the abnormal data. Set the interquartile range as ΔQ = Q3 - Q1, where Q3 is the third quartile and 75% of the data in the dataset is less than or equal to this value; Q1 is the first quartile and 25% of the data in the dataset is less than or equal to this value. When a data point x satisfies x > Q3 + 1.5×ΔQ or x < Q1 - 1.5×ΔQ, that is, when the daily delivery volume exceeds 1.5 times the interquartile range, it is regarded as an outlier and eliminated.

[0029] Finally, since the data value ranges of different features vary greatly, before inputting into the ANN model, it is also necessary to normalize the data to eliminate the dimensional differences between different features. The normalization process uses the Max - Min method to normalize the data into the interval [0, 1]. The specific formula is as follows: , where, is the normalized feature data, is the original feature data, and are the minimum and maximum values of this feature respectively.

[0030] 3. Model building and training; The present invention uses ANN to predict the delivery volume of multiple sites. Figure 2 As shown in the figure, the model consists of an input layer, two hidden layers and an output layer. The main purpose of its network structure design is to capture the complex nonlinear relationship of the delivery amount of low-value recyclables.

[0031] The input of the input layer is a multidimensional feature vector D∈X n×d , where n is the total number of records and d is the number of features. Let the weight matrix of the hidden layer be , the bias term is , the mapping from input to output is realized through nonlinear activation function. The calculation formula of the hidden layer is: , , in, Represents the layer index of the neural network; It is The linear transformation result of the layer (the value without activation function); It is The weight matrix of a layer, the size of which depends on the number of input neurons and output neurons in the layer; It is The output of the layer; It is The output of the layer (the value after the activation function); For the The bias term of the layer is used to adjust the activation value of the neuron; is an activation function used to introduce nonlinear characteristics.

[0032] The hidden layer uses the ReLU (Rectified Linear Unit) activation function. ReLU can effectively avoid the gradient vanishing problem and improve training efficiency. Its form is: , in, is the input value (i.e. ). ReLU is passed Ensure that negative values ​​become zero to avoid the gradient vanishing problem and improve training efficiency.

[0033] The output layer of the model is a one-dimensional prediction value, which represents the delivery volume of a specific low-value recyclables delivery station on a certain day. The output layer uses a linear activation function: , in, is the final output of the model, i.e., the predicted value of the delivery volume of a specific low-value recyclables delivery station on a certain day. Represents the last layer (output layer) of the neural network.

[0034] The model is optimized using the Adam optimizer, and its update rule is: , , , in, For the Model parameters at each iteration; and are the first and second order momentum respectively; Represents the loss function with respect to the parameter The gradient of is the learning rate; , is the momentum hyperparameter, usually set to =0.9, =0.999; To prevent the denominator from being a decimal of zero, the value is usually 10 -8 .

[0035] 4. Model performance evaluation; During the training process, the pre-processed data is divided into a training set and a test set, with 70% of the data used for training and 30% for testing. The training goal is to enable the model to accurately predict the delivery volume on the test set. During training, the batch size is set to 32 and the maximum number of iterations is 500 to ensure that the model is fully trained. The standardized data is used as input, and the prediction results are restored to the original scale through inverse transformation.

[0036] The five-fold cross-validation method was used to evaluate the model performance to ensure the generalization ability of the prediction results. The grid search method was used to tune the hyperparameters such as the number of hidden layer neurons and the learning rate. Finally, the number of hidden layer neurons was determined to be {256, 128} (the first hidden layer contained 256 neurons and the second hidden layer contained 128 neurons), and the initial learning rate was 0.001.

[0037] In order to more comprehensively evaluate the performance of the model for predicting the delivery volume of low-value recyclables, four widely used prediction evaluation indicators were used: fit coefficient R 2 , mean error MAE, mean absolute percentage error MAPE and root mean square error RMSE. 2 The closer the value is to 1, the closer the distance between the predicted value and the true value in the test set is, and the better the fitting performance of the prediction model is; the lower the values ​​of the other three prediction evaluation indicators MAE, MAPE, and RMSE are, the higher the prediction accuracy of the model is. The calculation formula of the prediction evaluation index is as follows: , , , , in, is the actual delivery volume, To predict the delivery volume, is the mean of all actual delivery amounts, is the sample size.

[0038] Through training, the ANN model of the present invention has a high fitting coefficient on the test set and excellent error indicators, with a MAPE of 0.38 and a RMSE of 13.14, which are significantly better than traditional time series models and other machine learning algorithms.

[0039] like Figure 3 The comparison chart of the predicted delivery volume and the actual delivery volume of specific low-value recyclables shown in the figure shows that the horizontal axis represents time in days, the vertical axis represents the delivery volume in kg, the solid line with dots represents the actual delivery volume (true value), and the dotted line with a cross represents the predicted delivery volume (predicted value). From the overall trend, the change trend of the predicted value and the actual value is basically the same, indicating that the prediction model performs well in capturing the overall trend. At some obvious peaks and valleys, the predicted value and the actual value fit well, indicating that the model can also predict well in these extreme cases. Overall, the ANN prediction model can reflect the actual changes in the delivery volume to a large extent, reasonably grasp the data trend, and the actual value and the predicted value are relatively close, indicating that the prediction accuracy of the ANN prediction model is relatively high.

[0040] 5. Multi-site prediction; In the multi-site prediction of the present invention, the delivery volume of each site in the next 3 to 7 days is predicted by constructing an ANN model for 436 low-value recyclables delivery sites covering multiple communities. The present invention takes the historical delivery data and external feature data of the site as input. After the model is trained, it can capture the complex dependency between delivery behavior and external feature data. After generating time series samples using the sliding window method, the ANN model performs independent predictions based on site-specific input features and generates daily delivery volume forecasts for the next few days.

[0041] 6. Community difference analysis; In the present invention, in order to comprehensively analyze the differences between communities, it is necessary to effectively classify the communities. The present invention selects two main classification bases: the comprehensive quality score of each community and the convenience characteristics. These two bases reflect the economic level of the community and the convenience of participating in smart recycling, respectively. The community economic quality score is obtained by comprehensively scoring the four community economic characteristics of community housing prices, community housing age, community greening rate and community floor area ratio through the entropy method. The convenience feature, that is, the number of people served by a single device, is directly related to the convenience of residents using smart recycling facilities and has an important impact on the recycling efficiency of the community.

[0042] The entropy method is a method that can objectively assign weights and is suitable for measuring the economic characteristics of different communities. The comprehensive score of the four community economic characteristics is defined as the comprehensive quality score of each community to reflect the economic level of the community. The entropy method assigns weights by calculating the degree of dispersion of community economic characteristics, ensuring the relative objectivity of each community's economic characteristics in the comprehensive score.

[0043] First, the economic characteristics of each community are standardized to eliminate the dimension effect. For each community economic characteristic, its entropy value is calculated as follows: , in, For the The entropy value of the economic characteristics of each community; It is The community in Standardized values ​​under the economic characteristics of each community; For the The community in The original value of each community's economic characteristics; is the total number of communities.

[0044] Finally, the weight of each community economic feature is calculated based on the entropy value. The specific calculation formula is as follows: , in, No. The weight of the economic characteristics of each community, is the total number of community economic characteristics.

[0045] Therefore, the comprehensive quality score of each community obtained by the entropy method can objectively reflect the economic level of each community, and then be used for subsequent cluster analysis. Subsequently, the K-means clustering algorithm is used to classify the communities using the comprehensive quality scores and convenience characteristics of each community. The K-means clustering algorithm performs clustering by minimizing the intra-cluster square error SSE, and its calculation formula is: , in, is the number of clusters, It is Clusters, is the cluster center, is a data point.

[0046] In the clustered community groups, the collection of low-value recyclables in each type of community will show similar characteristics. Figure 4 As shown in the figure, the relationship between the comprehensive quality score of different community types and the number of people served by a single device is shown. The horizontal axis represents the comprehensive quality score, which reflects the economic level of the community; the vertical axis represents the number of people served by a single device, that is, the average number of residents served by each recycling device, reflecting the convenience of delivery. The size of the bubble reflects the number of people served by a single device, and the color of the bubble reflects the different community types. Figure 4 It can be seen that the present invention obtained 5 types after clustering 173 communities. Combining historical data and prediction results, the following descriptions of these 5 types of communities are obtained: Type 1 communities (low quality-low convenience) have low economic levels and poor delivery convenience, and are relatively scarce in resource allocation; Category 2 communities (low quality-high convenience) have lower economic levels but higher delivery convenience and relatively better facility coverage; Category 3 communities (medium quality-medium convenience) are at a medium level in terms of economic level and delivery convenience, and are relatively balanced; The fourth type of community (high quality-low convenience) has a higher economic level, but the delivery convenience is insufficient and the facility coverage is short. Category 5 communities (high quality-high convenience) excel in both economy and delivery convenience, with complete recycling facilities and good resource allocation.

[0047] In summary, the high-precision ANN model developed in this invention not only accurately predicts the delivery volume of each station, but also classifies and summarizes the prediction results of the station delivery volume according to the community type, thereby providing targeted optimization suggestions for the resource allocation of the mid-term removal link of the intelligent recycling system. Specifically, different types of communities can adopt different removal strategies and resource allocation plans according to their economic level and delivery convenience to improve the overall operational efficiency and recycling effect.

[0048] Example 2: The embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of a method for predicting the delivery volume of low-value recyclables in a community based on an ANN in the aforementioned embodiment.

[0049] Example 3. The embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of a method for predicting the delivery volume of low-value recyclables in a community based on an ANN in the aforementioned embodiment are implemented.

[0050] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for predicting the delivery volume of low-value recyclables in a community based on ANN, characterized in that: The following steps are involved: Data collection: Obtain historical delivery data, external feature data, and community feature data of low-value recyclables delivery sites covering multiple communities; wherein the external feature data includes weather information, weekday status, and holiday promotion information, and the community feature data includes community economic characteristics and convenience characteristics; Data preprocessing: For historical delivery data and external feature data, Lagrange interpolation method is used to fill missing values, box plot method is used to detect and remove outliers, and Max-Min method is used for normalization to ensure data quality and consistency; Model establishment: Design a multi-level ANN structure, which includes an input layer, two hidden layers and an output layer. The input layer receives multi-dimensional feature vectors, the hidden layer uses the nonlinear activation function ReLU to capture the complex pattern relationship of residents' delivery behavior, and the output layer provides the prediction results of the daily delivery volume of specific low-value recyclables delivery stations. Model training: During the training process, the Adam optimizer is used to adjust the parameters of the ANN model, and the hyperparameter configuration is optimized through cross-validation. At the same time, a variety of prediction evaluation indicators are used to comprehensively evaluate the prediction accuracy of the ANN model. Multi-site forecasting: The sliding window method is used to generate time series samples, and the trained ANN model is used to independently forecast the delivery volume of each low-value recyclables delivery site in the next 3 to 7 days to obtain short-term forecast data; Community difference analysis: First, based on the community's economic characteristics, the comprehensive quality score of each community is calculated using the entropy method. Secondly, the convenience characteristics are combined and the K-means clustering algorithm is used to calculate the comprehensive quality score of the community. Finally, the community type is classified according to the score. Based on the classification results, according to historical delivery data and short-term forecast data, targeted resource scheduling strategies are formulated to optimize the frequency of transportation and facility layout.

2. The method according to claim 1, characterized in that The method for obtaining external characteristic data and community characteristic data in the data collection is specifically to obtain detailed weather conditions, temperature, wind force level, community housing prices, community housing ages, community greening rates and community floor area ratios, as well as customized working days, holidays or promotional days from open data sources on the Internet.

3. The method according to claim 1, characterized in that The data preprocessing also includes encoding conversion of text data or category data to facilitate model input.

4. The method according to claim 1, characterized in that In the model establishment, the first hidden layer contains 256 neurons, the second hidden layer contains 128 neurons, and the initial learning rate is set to 0.

001.

5. The method according to claim 1, characterized in that The batch size of the training process in the model training is set to 32, and the maximum number of iterations is 500 to ensure that the model is fully trained while avoiding overfitting.

6. The method according to claim 1, characterized in that The convenience feature is measured by the number of people served by a single device.

7. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images