Iron tower sharing state judgment method and system based on multi-index classification model

Abstract

The application provides a tower sharing state judgment method and system based on a multi-index classification model. The method comprises the following steps: acquiring power index data and traffic index data of a tower in different sharing states to construct sharing state feature samples, labeling sharing state labels for the sharing state feature samples, wherein the sharing state labels comprise exclusive sharing, two-party sharing and three-party sharing; constructing a random forest classification model, training the random forest classification model by using the sharing state feature samples, and obtaining a tower sharing state classification model; acquiring power index data and traffic index data of a target tower to construct a sharing state feature of the target tower; inputting the sharing state feature of the target tower into the tower sharing state classification model to obtain a sharing state of the target tower. The method can solve the problems of poor timeliness, low accuracy and high cost of traditional schemes, and provides support for tower resource scheduling and communication network planning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication infrastructure management, in particular to a tower sharing state judgment method and system based on a multi-index classification model. BACKGROUND

[0002] As a key infrastructure in the communication industry, the tower carries base station equipment of each operator, and has a sharing mode, including exclusive sharing, two-party sharing and three-party sharing. Through tower sharing, construction cost can be reduced and resource utilization can be improved. Currently, tower companies mainly manage it. In the scenarios of tower resource management and operator cooperation mode, it is necessary to judge the sharing state of the tower. The existing judgment method mainly relies on a manual record method, which has the following defects:

[0003] 1. Poor information timeliness: after the operator increases or reduces equipment, the sharing state record cannot be updated in real time, and needs to rely on manual registration or periodic investigation, resulting in a lag between system information and actual state, usually 1-2 months.

[0004] 2. Low judgment accuracy: manual input is prone to misfilling (such as misfilling "two-party sharing" as "exclusive sharing") and omission. Field investigation is affected by personnel experience and investigation frequency, and it is difficult to cover all towers, with a high error rate. The average error rate in the industry is about 15%.

[0005] 3. No objective operation data to support sharing state judgment, only relying on subjective records, resulting in inaccurate assessment of load when scheduling tower resources, such as misjudging a "three-party sharing" tower as "exclusive sharing", causing equipment expansion planning errors.

[0006] 4. High labor cost: a large amount of manpower is needed for contract checking, field investigation and data correction. SUMMARY

[0007] The purpose of the present application is to provide a tower sharing state judgment method and system based on a multi-index classification model to at least partially overcome the defects of the prior art.

[0008] To achieve the above purpose, the present application proposes the following technical solutions:

[0009] In a first aspect, a tower sharing state judgment method based on a multi-index classification model is provided, comprising:

[0010] Obtaining power index data and traffic index data of the tower in different sharing states to construct sharing state feature samples, and labeling sharing state labels for the sharing state feature samples, the sharing state labels including exclusive sharing, two-party sharing and three-party sharing;

[0011] A random forest classification model is constructed, the random forest classification model is trained by using the shared state feature sample, and a tower sharing state classification model is obtained.

[0012] The power index data and the traffic index data of the target tower are obtained to construct a shared state feature of the target tower.

[0013] The shared state feature of the target tower is input into the tower sharing state classification model to obtain a sharing state of the target tower.

[0014] As an optional implementation of the method of the first aspect, the power index data includes total power consumption, daily average power consumption, peak period power consumption, non-peak period power consumption, and each operator equipment branch power consumption of the tower in a preset time period.

[0015] The traffic index data includes total data traffic, daily average data traffic, operator traffic proportion, peak period traffic peak value / average value, and user access amount in the preset time period.

[0016] As an optional implementation of the method of the first aspect, the power index data and the traffic index data of the tower in different sharing states are obtained to construct a shared state feature sample, specifically including:

[0017] The power index data is converted into a distribution with a mean of 0 and a standard deviation of 1.

[0018] ;

[0019] Wherein, represents the distribution of the power index data, represents the power index data, represents the mean of the power index data, represents the standard deviation of the power index data.

[0020] The traffic index data is normalized:

[0021] ;

[0022] Wherein, represents the normalized value of the traffic index data, represents the traffic index data, represents the minimum value in the traffic index data, represents the maximum value in the traffic index data.

[0023] The shared state feature sample is constructed according to the distribution data of the power index data and the normalized value of the traffic index data.

[0024] In a second aspect, a tower sharing state judgment system based on a multi-index classification model is provided, comprising:

[0025] A first data acquisition module is configured to acquire power index data and traffic index data of a tower in different sharing states to construct sharing state feature samples, and label sharing state labels for the sharing state feature samples, wherein the sharing state labels include exclusive sharing, two-party sharing, and three-party sharing.

[0026] A training module is configured to construct a random forest classification model, train the random forest classification model using the sharing state feature samples, and obtain a tower sharing state classification model.

[0027] A second data acquisition module is configured to acquire power index data and traffic index data of a target tower to construct sharing state features of the target tower.

[0028] A judgment module is configured to input the sharing state features of the target tower into the tower sharing state classification model to obtain a sharing state of the target tower.

[0029] As an optional implementation of the system of the second aspect, the power index data includes total power consumption, daily average power consumption, peak period power consumption, non-peak period power consumption, and each operator equipment branch power consumption of the tower in a preset time period.

[0030] The traffic index data includes total data traffic, daily average data traffic, operator traffic proportion, peak period traffic peak value / average value, and user access amount in the preset time period.

[0031] As an optional implementation of the system of the second aspect, the first data acquisition module is specifically configured to:

[0032] convert the power index data into a distribution with a mean value of 0 and a standard deviation of 1.

[0033] ;

[0034] wherein, represents the distribution of the power index data, represents the power index data, represents the mean value of the power index data, represents the standard deviation of the power index data.

[0035] normalize the traffic index data.

[0036] ;

[0037] wherein, a normalized value of the flow index data, the flow index data, a minimum value in the flow index data, a maximum value in the flow index data;

[0038] constructing the sharing state feature sample according to the distribution data of the power index data and the normalized value of the flow index data.

[0039] In a third aspect, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program. When the computer program is executed on an electronic device, the electronic device is caused to perform the iron tower sharing state judgment method based on the multi-index classification model.

[0040] In a fourth aspect, an electronic device is provided, and the electronic device comprises:

[0041] at least one memory configured to store a program;

[0042] at least one processor configured to execute the program stored in the memory, and when the program stored in the memory is executed, the processor is configured to perform the iron tower sharing state judgment method based on the multi-index classification model.

[0043] Advantages: Compared with the prior art, the iron tower sharing state judgment method and system based on the multi-index classification model have the following advantages:

[0044] The method constructs a sharing state feature sample by mining the internal correlation between power, flow index and sharing mode of the iron tower, and uses a random forest model to learn the differences of the sharing state feature sample under different sharing modes. The trained iron tower sharing state classification model is used to identify the sharing state of the iron tower, solves the problems of poor timeliness, low accuracy and high cost of the traditional scheme, realizes the transformation from "manual recording" to "data driving", and provides support for iron tower resource scheduling and communication network planning. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 A flowchart of an iron tower sharing state judgment method based on a multi-index classification model according to an embodiment.

[0046] Figure 2 A structural diagram of an iron tower sharing state judgment system based on a multi-index classification model according to an embodiment. DETAILED DESCRIPTION

[0047] The application will be further described below with reference to the drawings and specific embodiments. It should be understood, however, that the application can be implemented in various forms and that the specific examples and embodiments shown in the drawings and described below are not intended to limit the application to the specific embodiments described.

[0048] It should be understood that the technical features listed above for different embodiments can be combined with each other to form further embodiments within the scope of the application, if technically feasible. Furthermore, the specific examples and embodiments described in the application are non-limiting and corresponding modifications can be made to the structures, steps, sequences described above without departing from the scope of the application.

[0049] Reference is made to Figure 1 , Figure 1 An exemplary flowchart of a method for judging the sharing state of a tower based on a multi-index classification model is shown. As shown in Figure 1 , the method comprises steps S100 to S106.

[0050] S100: Obtain the power index data and traffic index data of the tower in different sharing states to construct sharing state feature samples, and label the sharing state labels for the sharing state feature samples.

[0051] The sharing state of the tower includes exclusive sharing, two-party sharing and three-party sharing.

[0052] It has been found through research that the power index data and traffic index data of the tower are significantly different under different sharing states of the tower:

[0053] The total power consumption and total traffic of a three-party sharing tower are usually higher than those of a two-party sharing tower, and much higher than those of an exclusive sharing tower.

[0054] The traffic composition of a multi-sharing tower presents a multi-operator feature, for example, the traffic proportion of a three-party sharing tower is usually divided equally by 2-3 operators.

[0055] Based on the above characteristics, the power index data and traffic index data of the tower under different sharing states are selected in the present embodiment to construct the sharing state feature samples.

[0056] The power index data includes the total power consumption, daily average power consumption, peak period power consumption, off-peak period power consumption and each operator device branch power consumption associated with the target tower in a preset time period.

[0057] The traffic index data includes the total data traffic, daily average data traffic, operator traffic proportion, peak period traffic peak value / average value and user access amount in a preset time period.

[0058] For power indicator data, intelligent power monitoring devices equipped on the tower can be used to collect, record total power supply and branch power in real time, and aggregate by 1 hour as granularity.

[0059] For traffic indicator data, base station network management system can be used for collection, and based on operator network slicing technology to identify uplink and downlink traffic of each operator, to calculate peak and average value by 1 hour as granularity.

[0060] The collection period of power indicator data and traffic indicator data can be set according to demand, such as collecting by day, week, month as period, and the total duration of each collection is usually not less than 30 days, to ensure that the data has statistical representativeness, and at the same time, mark invalid data caused by abnormal power failure and short circuit.

[0061] Before constructing the shared state feature sample based on the collected power indicator data and traffic indicator data, the power indicator data and traffic indicator data need to be preprocessed, including data cleaning, missing value processing, standardization and normalization, so as to generate features that can be input into the random forest classification model.

[0062] Data cleaning mainly includes:

[0063] Eliminate outliers: delete data beyond reasonable range, such as single-day power consumption >1000kWh, traffic <0.

[0064] Delete redundant data: remove duplicate collection of the same period indicator data, such as repeated reporting data caused by equipment failure.

[0065] Mark missing period: record the interruption period in data collection caused by equipment failure and signal interruption.

[0066] Missing value processing mainly adopts hierarchical processing strategy according to missing proportion:

[0067] Missing proportion <5%: adopt adjacent period data mean filling, such as missing single time point, take the arithmetic mean of the previous valid period and the next valid period data.

[0068] Missing proportion 5%-30%: adopt linear interpolation method to fill, the formula is:

[0069] ;

[0070] Where, t1 is the time of the last valid data before missing, x1 is the last valid data before missing, t2 is the time of the last valid data after missing, x2 is the last valid data after missing, .

[0071] Missing proportion >30%: mark the tower data as invalid sample, which does not participate in model training.

[0072] The standardization and normalization processing refers to standardizing the power index data and normalizing the flow index data.

[0073] Specifically, the power index data is standardized, specifically including:

[0074] The power index data is converted into a distribution with a mean of 0 and a standard deviation of 1:

[0075]

[0076] wherein, represents the distribution of the power index data, represents the power index data, represents the mean of the power index data, represents the standard deviation of the power index data.

[0077] The normalization of the flow index data mainly includes calculating:

[0078]

[0079] wherein, represents the normalized value of the flow index data, represents the flow index data, represents the minimum value in the flow index data, represents the maximum value in the flow index data.

[0080] The different processing methods for the power index data and the flow index data can avoid interference between the indexes.

[0081] According to the distribution data of the power index data and the normalized value of the flow index data, a shared state feature sample can be constructed.

[0082] S102: Construct a random forest classification model, train the random forest classification model using the shared state feature sample, and obtain a tower sharing state classification model.

[0083] Specifically, the entire training process can adopt the following steps:

[0084] 1) Model parameter setting

[0085] Number of decision trees: 50-200, preferably 100;

[0086] Maximum depth of a single tree: 5-20 layers, preferably 10 layers;

[0087] Sample sampling: a sub-sample set is extracted from the shared state feature sample through bootstrap sampling;​​

[0088] Feature selection: 60%-80% of the total number of features are randomly selected for node splitting when each tree is constructed.

[0089] 2) Shared state feature sample processing

[0090] Sample label generation: based on operator contract, field survey results, historical research and judgment + artificial review three ways to mark shared state label, including exclusive / two shared / three shared;

[0091] Sample division: divided into training set in the ratio of 7:3, used for model training, and validation set, used for performance evaluation;

[0092] Training stop condition: 5-fold cross-validation is used to optimize parameters, and the training is stopped when the classification accuracy of the validation set is greater than or equal to 85%.

[0093] S104: Obtain the electricity index data and traffic index data of the target tower to construct the shared state features of the target tower.

[0094] For the target tower to be identified for sharing state transition, obtain the electricity index data and traffic index data of the target tower, and preprocess the electricity index data and traffic index data of the target tower according to the specific processing mode in the above step S100, to obtain the shared state features of the target tower.

[0095] S106: Input the shared state features of the target tower into the tower sharing state classification model to obtain the shared state of the target tower.

[0096] The trained tower sharing state classification model is used to classify the shared state features of the target tower to obtain the shared state of the target tower.

[0097] Corresponding to the above tower sharing state judgment method based on a multi-index classification model, the embodiment also provides a tower sharing state judgment system based on a multi-index classification model, which is used to realize the above tower sharing state judgment method based on a multi-index classification model. Please refer to Figure 2 , Figure 2 The structure of a tower sharing state judgment system based on a multi-index classification model is schematically shown, as shown in Figure 2 The system comprises:

[0098] A first data acquisition module is configured to acquire electricity index data and traffic index data of a tower in different sharing states to construct shared state features, and to mark shared state labels for shared state feature samples, wherein the shared state labels include exclusive, two shared, and three shared.

[0099] The training module is configured to construct a random forest classification model, train the random forest classification model by using the shared state feature sample, and obtain a tower sharing state classification model.

[0100] The second data acquisition module is configured to acquire power index data and flow index data of the target tower to construct a shared state feature of the target tower.

[0101] The judgment module is configured to input the shared state feature of the target tower into the tower sharing state classification model to obtain a sharing state of the target tower.

[0102] Specifically, the power index data includes total power consumption, daily average power consumption, peak period power consumption, non-peak period power consumption and operator equipment branch power consumption of the tower in a preset time period. The flow index data includes total data flow, daily average data flow, operator flow proportion, peak period flow peak value / average value and user access amount in a preset time period.

[0103] Optionally, the first data acquisition module is specifically configured to:

[0104] The power index data is converted into a distribution with a mean value of 0 and a standard deviation of 1.

[0105] ;

[0106] wherein, represents the distribution of the power index data, represents the power index data, represents the mean value of the power index data, represents the standard deviation of the power index data.

[0107] The flow index data is normalized.

[0108] ;

[0109] wherein, represents the normalized value of the flow index data, represents the flow index data, represents the minimum value in the flow index data, represents the maximum value in the flow index data.

[0110] The shared state feature sample is constructed according to the distribution data of the power index data and the normalized value of the flow index data.

[0111] The embodiment further provides a computer readable storage medium storing a computer program, when the computer program runs on an electronic device, the computer program causes the electronic device to execute the tower sharing state judgment method based on the multi-index classification model.

[0112] The embodiment also provides an electronic device, comprising:

[0113] at least one memory for storing a program;

[0114] at least one processor for executing the program stored in the memory, and when the program stored in the memory is executed, the processor is configured to execute the method for judging the sharing state of a tower based on a multi-index classification model.

[0115] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present disclosure.

[0116] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for judging a sharing state of a tower based on a multi-index classification model, characterized in that, The method comprises the following steps: obtaining power consumption index data and traffic index data of a tower in different sharing states to construct a sharing state feature sample, and labeling a sharing state label for the sharing state feature sample, wherein the sharing state label comprises exclusive sharing, two-party sharing and three-party sharing; constructing a random forest classification model, training the random forest classification model by using the sharing state feature sample, and obtaining a tower sharing state classification model; obtaining power consumption index data and traffic index data of a target tower to construct a sharing state feature of the target tower; inputting the sharing state feature of the target tower into the tower sharing state classification model to obtain a sharing state of the target tower.

2. The method of claim 1, wherein, The power consumption index data comprises total power consumption in a preset time period, daily average power consumption, peak period power consumption, off-peak period power consumption and operator equipment branch power consumption of the tower. The traffic index data comprises total data traffic in the preset time period, daily average data traffic, operator traffic proportion, peak period traffic peak value / average value and user access amount.

3. The method of claim 1, wherein, The method comprises the following steps: converting the power consumption index data into a distribution with a mean value of 0 and a standard deviation of 1; ； wherein, represents a distribution of the electricity amount index data, represents the electricity amount index data, represents a mean of the electricity amount index data, represents a standard deviation of the electricity amount index data; normalizing the traffic index data; ； wherein, denotes a normalized value of the flow indicator data, denotes the flow indicator data, denotes a minimum value in the flow indicator data, denotes a maximum value in the flow indicator data; constructing the sharing state feature sample according to the distribution data of the power consumption index data and the normalized values of the traffic index data.

4. A tower sharing state judgment system based on a multi-index classification model, characterized in that, The method comprises the following steps: a first data acquisition module is configured to obtain power consumption index data and traffic index data of a tower in different sharing states to construct a sharing state feature, and label a sharing state label for the sharing state feature sample, wherein the sharing state label comprises exclusive sharing, two-party sharing and three-party sharing; a training module is configured to construct a random forest classification model, train the random forest classification model by using the sharing state feature sample, and obtain a tower sharing state classification model; a second data acquisition module is configured to obtain power consumption index data and traffic index data of a target tower to construct a sharing state feature of the target tower; a judgment module is configured to input the sharing state feature of the target tower into the tower sharing state classification model to obtain a sharing state of the target tower.

5. The system of claim 4, wherein, The power consumption index data comprises total power consumption in a preset time period, daily average power consumption, peak period power consumption, off-peak period power consumption and operator equipment branch power consumption of the tower. The traffic index data comprises total data traffic in the preset time period, daily average data traffic, operator traffic proportion, peak period traffic peak value / average value and user access amount.

6. The system of claim 4, wherein, The first data acquisition module is specifically configured to: convert the power consumption index data into a distribution with a mean value of 0 and a standard deviation of 1; ； wherein, represents a distribution of the electricity amount index data, represents the electricity amount index data, represents a mean of the electricity amount index data, represents a standard deviation of the electricity amount index data; normalize the traffic index data; ； wherein, denotes a normalized value of the flow indicator data, denotes the flow indicator data, denotes a minimum value in the flow indicator data, denotes a maximum value in the flow indicator data; construct the sharing state feature sample according to the distribution data of the power consumption index data and the normalized values of the traffic index data.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and when the computer program runs on the electronic device, the electronic device executes the method in any one of claims 1 to 3.

8. An electronic device, comprising: comprising: at least one memory for storing a program; at least one processor for executing the program stored in the memory, and when the program stored in the memory is executed, the processor is configured to execute the method in any one of claims 1 to 3.

Images