Search

CN-122022577-A - Service area operation management level evaluation method, device, medium and program product

CN122022577ACN 122022577 ACN122022577 ACN 122022577ACN-122022577-A

Abstract

The invention discloses a service area operation management level assessment method, equipment, medium and program product, wherein the method comprises the steps of constructing a system covering five assessment dimensions and quantization indexes of an expressway service area, adopting an IoT real-time acquisition and cross-system API docking mode to acquire and verify data of each quantization index, adopting a mixed weight calculation method of a D-AHP and entropy weight method to determine mixed weights of each dimension and index, optimizing the mixed weights according to time periods and types, adopting hierarchical calculation logic to obtain overall assessment scores and grades of the service area, adopting a short-board positioning method based on association rule mining to position short-board indexes and core factors thereof through deviation analysis and an improved Apriori algorithm, and outputting a multi-dimensional assessment report. The invention can construct comprehensive evaluation dimension, accurately quantize evaluation index, improve the efficiency and reliability of data acquisition, improve the objectivity and pertinence of weight, locate the short-circuit cause of each service area and support fine management decision.

Inventors

  • Liu Chayan
  • HONG KE
  • QIU CHENG
  • Qiu Danyue
  • Chen Xunzhi
  • ZENG HANYING
  • WU HANXIANG
  • CHEN QIAN
  • Cai dongxing

Assignees

  • 江西畅行高速公路服务区开发经营有限公司

Dates

Publication Date
20260512
Application Date
20260128

Claims (10)

  1. 1. A service area operation management level assessment method, comprising: Constructing a multidimensional quantitative index system which covers five evaluation dimensions of public basic service capacity, public service capacity, business management capacity, cost management capacity and comprehensive management level of a highway service area and quantitative indexes in each dimension; constructing a hardware acquisition layer and a system docking layer dual acquisition channel by adopting an data acquisition mode of real-time acquisition of an internet traffic (IoT) and cross-system Application Program Interface (API) docking, and acquiring data of each quantization index from an IoT sensor, an intelligent terminal and a service system; Determining the mixed weight of the dimension and the index from the historical operation data of a plurality of service areas by adopting a mixed weight calculation method of a D-AHP and entropy weight method, and optimizing the mixed weight by adopting a weight adjustment mechanism with adaptive time periods and types; based on the optimized mixed weight, adopting hierarchical calculation logic to process the verified data, and obtaining the overall evaluation score and grade of the service area; The short-board index and the core cause thereof are positioned by deviation analysis and an improved Apriori algorithm based on a short-board positioning method of association rule mining, and a multidimensional evaluation report comprising a core data layer, an analysis layer and an application layer is output.
  2. 2. The method of claim 1, wherein, The quantization indexes under the public basic service capability evaluation dimension comprise the number of the saturation accumulated alarms of the lunar parking space, the number of the saturation accumulated alarms of the lunar toilet position and the utilization rate of the lunar average charging pile; quantification indexes under the evaluation dimension of public service capability comprise a monthly average cleaning personnel proportion, a monthly effective complaint number, a monthly average dredger proportion and a monthly average equipment failure rate; quantification indexes under the dimension of business management capability evaluation comprise month average business supply quantity, month average business area plateau effect, month average guest price, month average vehicle average consumption and month average consumption conversion rate; the quantitative index under the evaluation dimension of the cost management capability comprises a monthly output input-output ratio and monthly energy consumption in unit area; The comprehensive management level evaluates quantitative indexes under the dimension including month assessment scores of the operation supervision part.
  3. 3. The method of claim 2, wherein the monthly effective complaint number is calculated as a total number of monthly complaints-a number of repeated complaints-a number of malicious complaints; Wherein, based on the mobile phone number or the ID card number of the complaint after desensitization treatment, double judgment of semantic similarity of the core content of the complaint is combined to determine the repeatedly submitted complaint; the malicious complaints are automatically screened through the following steps: (1) Firstly, collecting complaint text data, including word description, complaint type, submitting time and complaint person information, and then cleaning data, word segmentation and stop word removal; (2) The method comprises the steps of keyword matching and identification, namely constructing a malicious complaint keyword library, storing a plurality of core keywords of an abusive vocabulary, an offensive expression and an imaginary vocabulary without a real basis, supporting dynamic updating of the core keywords, and marking the core keywords as suspected malicious complaints when the frequency of occurrence of the core keywords in a complaint text is more than or equal to 2 times or the abusive vocabulary is contained by adopting a mode of combining accurate matching and fuzzy matching; (3) Semantic analysis verification, namely taking a service area complaint text marked by history as a training data set, comprising malicious and non-malicious labels marked by manual, using a pre-trained BERT language model to perform semantic similarity calculation and emotion polarity analysis on the suspected malicious complaint text, outputting a malicious probability value, setting a malicious threshold value to be 0.8, and judging that the suspected malicious complaint is malicious when the malicious probability value is more than or equal to 0.8.
  4. 4. The method of claim 1, wherein the specific implementation steps of acquiring the data of each quantization index from the IoT sensor, the intelligent terminal and the service system are performed by adopting an IoT real-time acquisition and cross-system API docked data acquisition mode to build a hardware acquisition layer and a system docking layer dual acquisition channel: step one, demand carding and scheme design (1) Data demand carding, namely defining the data type, acquisition frequency, data precision requirement and data source corresponding to each quantization index; (2) Checking the model and communication protocol of the existing IoT sensor and the existing intelligent terminal in the target service area, and combing the interface type, data format and access authority of the business system to be docked; (3) The scheme design comprises an Internet of things (IoT) sensor and intelligent terminal complementary deployment scheme, a cross-system Application Program Interface (API) docking scheme comprising an interface calling rule and a data synchronization mechanism, and a data transmission security scheme comprising an encryption protocol selection and authority management and control strategy aiming at uncovered acquisition points; the equipment is collected, deployed and debugged in real time, and by deploying a plurality of types of IoT sensors and intelligent terminals and combining with edge computing nodes, the real-time collection and local preprocessing of basic operation data are realized, and the specific requirements are as follows: (1) The equipment selection and deployment comprise ① parking space monitoring cameras which are deployed at preset positions of a parking lot, lenses vertically and downwards align with parking space outlines, a dead angle-free parking area covered by parking space, image recognition accuracy less than or equal to +/-2 cm, parking space state recognition response time less than or equal to 1s, support RS485 protocol/Ethernet communication, real-time analysis of vehicle parking outline characteristics through an image algorithm, realization of real-time acquisition of parking space occupation states, ② toilet infrared sensors which are installed above the inner side of a toilet door, detection distance of 0.1-1m, false alarm rate less than or equal to 0.5%, support NB-IoT wireless transmission, acquisition of toilet space occupation states, ③ charging pile intelligent terminals which are integrated in each charging pile control system, support Modbus protocol, acquisition of operation data of charging pile starting and stopping time and charging time and power, ④ intelligent hydroelectric gas meters which are used for replacing original mechanical meters, support NB-IoT wireless transmission, and acquisition of real-time energy consumption data, ⑤ edge computing gateways which are deployed in a service area machine room, support multiprotocol conversion, and conversion of RS485 and Modbus into MQTT; (2) The device debugging comprises ① single device debugging, ② networking debugging, namely connecting all the IoT sensors and the intelligent terminals to an edge computing gateway, verifying a data aggregation function, ensuring that the gateway can stably receive the data of each device and perform format conversion, wherein the power-on test is performed on each IoT sensor and each intelligent terminal, and the data acquisition precision and the communication stability are verified; step three, cross-system API docking development and joint debugging, adopting RESTful API interface development mode to realize bidirectional docking of data of the evaluation platform and each service system and ensure accurate synchronization of structured data, and the specific implementation steps are as follows: (1) The interface development comprises the steps of ① developing a data call script and setting interface call parameters according to an interface document provided by a service system with an existing interface, ② developing a customized API interface according to the service system without the existing interface, deploying the customized API interface on a system server to realize active pushing or passive inquiring of data, ③ unifying interface data formats and standardizing JSON and XML format data returned by each system into a structured data format required by an evaluation platform; (2) The authority configuration, namely applying interface access authorities to each service system management party, configuring an IP white list, setting an interface access key and guaranteeing data transmission safety; (3) The joint debugging test comprises ① single-system joint debugging, ② multi-system linkage test, wherein the joint debugging test comprises the steps of calling an API (application program interface) of each service system, verifying the integrity, accuracy and effectiveness of data acquisition, verifying the consistency of cross-system data and ensuring that the data synchronization has no delay; step four, data transmission and edge preprocessing (1) The data transmission channel is constructed, ① IoT data are transmitted to an edge computing gateway through NB-IoT and 4G networks, the gateway establishes encryption connection with an evaluation platform through a 5G network, the transmission delay is less than or equal to 50ms, ② API interfaces data, the evaluation platform calls API interfaces of all business systems through the 5G network, and the data are directly stored in a cloud database after being obtained; (2) The edge preprocessing comprises that an edge computing gateway preprocesses original data acquired by an IoT sensor and an intelligent terminal, wherein the preprocessing comprises ① data cleaning, removing abnormal values, filtering repeated data, ② data aggregation, ③ protocol conversion, namely converting Modbus and RS485 protocol data into MQTT protocol data, and adapting to an evaluation platform receiving format; Step five, online operation and operation maintenance guarantee (1) Test operation, namely online test operation is carried out for 1 month, the data acquisition state is monitored in real time, and an alarm mechanism is set; (2) Operation and maintenance optimization, namely, performing calibration and maintenance on an IoT sensor and an intelligent terminal at regular intervals to replace fault equipment, and performing operation and maintenance on an ② interface, namely, monitoring the running state of an API interface in real time, and timely processing the problems of interface calling failure and permission expiration; (3) The hardware acquisition layer test comprises deployment of a parking space monitoring camera, deployment of a toilet position infrared sensor, deployment of a charging pile intelligent management terminal, deployment of an RS485 and Modbus protocol, deployment of an intelligent water meter, an electricity meter and a gas meter, and support of NB-IoT wireless transmission, wherein the image recognition precision is less than or equal to +/-2 cm, and the parking space state recognition response time is less than or equal to 1s; (4) The system docking layer tests that a service system is docked through a RESTful API interface, and the interface calling frequency is configurable; (5) And the data transmission test is that a 5G+ edge computing architecture is adopted, an edge node is responsible for local data preprocessing, standardized data is uploaded to an evaluation platform through a 5G network, and the transmission delay is less than or equal to 50ms.
  5. 5. The method of claim 1 wherein the step of determining the hybrid weights of the dimensions and metrics from the historical operating data of the plurality of service areas is performed by using a hybrid weight calculation method of D-AHP, entropy weight method: Step one, constructing a hierarchical structure The target layer is the overall operation level evaluation of the service area, the criterion layer is five evaluation dimensions, and the index layer is a specific quantization index under each dimension; Step two, constructing a data driving type judgment matrix (1) The data sample selection comprises the steps of respectively collecting historical operation data of N different types of service areas for at least 12 months, covering quantitative index data of five evaluation dimensions, forming a sample matrix X of N rows and M columns, wherein M is the total index number; (2) The dimension importance quantization comprises the steps of calculating the influence contribution degree of each dimension to a target layer as a construction basis of a judgment matrix, namely splitting a sample matrix X according to the dimension to obtain an index data subset X 1 -X 5 of each dimension, adopting a principal component analysis method to reduce the dimension of the index data subset of each dimension, extracting a first principal component with the accumulated variance contribution rate of more than or equal to 85 percent as a comprehensive quantized value F 1 -F 5 of the dimension, calculating a correlation coefficient R i of the comprehensive quantized value F i of each dimension and a service area integral operation effect quantized value Y, defining a dimension importance coefficient W i =|R i |/Σ|R i I, wherein i=1-5, constructing a criterion layer judgment matrix A based on the importance coefficient W i , namely a matrix element a ij =W i /W j , and ensuring to meet the 1-9 scale method requirements; (3) The index layer judgment matrix is constructed by calculating the correlation coefficient R ij of each index and the dimension comprehensive quantized value F i for the index C ij in each dimension C i to obtain an index importance coefficient W ij =|R ij |/Σ|R ij I, and constructing an index layer judgment matrix A i and an element a ijk =W ij /W ik ; step three, consistency test and matrix optimization (1) Calculating a maximum eigenvalue lambda max of the judgment matrix and a consistency index CI= (lambda max -n)/(n-1), wherein n is the order of the judgment matrix; (2) Searching an average random consistency index RI based on a standard RI value table of a 1-9 scale method; (3) If CR is more than or equal to 0.1, optimizing the judgment matrix element by a gradient adjustment method, namely carrying out tiny adjustment on a ij , adjusting the step length by 0.1, recalculating CR until CR is less than 0.1, and strictly following the reciprocity requirement of a ij =1/a ji in the adjustment process; D-AHP weight calculation Calculating a feature vector of the qualified judgment matrix by adopting a feature vector method, and normalizing to obtain a criterion layer D-AHP weight W 1 and an index layer D-AHP weight W 1j ; step five, calculating objective weight W by entropy weight method 2 Based on the sample matrix X of N rows and M columns, the objective weight is determined by the index discrete degree, and the steps are as follows: (1) Data standardization, namely distinguishing positive and negative indexes, and standardizing original data to a [0,1] interval by adopting an extremum method to obtain a standardized matrix Z; (2) Calculating index proportion, namely carrying out translation processing on Z, and calculating proportion p ij of an ith sample under the jth index to meet Σp ij =1; (3) Calculating an index entropy value, namely calculating the entropy value according to a formula e j =－[1/lnN]×Σp ij lnp ij ; (4) Calculating a difference coefficient according to a formula g j =1－e j ; (5) Determining objective weights, namely normalizing g j to obtain index layer objective weights W 2j , wherein the criterion layer objective weights are obtained by summing index weights under corresponding dimensions; Step six, determining the mixing weight And a weighted summation method is adopted to fuse the subjective weight and the objective weight, a final mixed weight calculation formula W=alpha×W 1 +(1－α)×W 2 is adopted, and the coefficient alpha is adjusted according to management requirements.
  6. 6. The method of claim 1, wherein the step of processing the verified data using hierarchical computation logic to obtain an overall evaluation score and rank for the service area based on the optimized hybrid weights is performed as follows: Step one, scoring each quantization index Setting grading scoring standards for each quantization index based on industry standards and management requirements, and calculating scores by adopting a piecewise linear interpolation method; Step two, dimension score calculation The scores of all indexes under each evaluation dimension are weighted and summed according to the index weight, wherein the formula is that the dimension score=Σ (index score×optimized index weight), and the score range is 0-100 points; step three, calculating overall evaluation score The total evaluation score = public basic service capability score x W ' 1 + public service capability score x W ' 2 + business management capability score x W ' 3 + cost management capability score x W ' 4 + management level score x W ' 5 , score range 0-100 points, optimized mixed weight W ' = [ W ' 1 ,w' 2 ,w' 3 ,w' 4 ,w' 5 ]; Step four, grading The overall evaluation score is more than or equal to 85 and is classified as excellent, 75-84 is classified as good, 60-74 is classified as qualified, and less than 60 is classified as unqualified.
  7. 7. The method of claim 1, wherein the specific implementation steps for locating the short-board index and its core cause by bias analysis and modified Apriori algorithm based on the short-board locating method of association rule mining are as follows: step one, calculating deviation Calculating the deviation rate of the actual score and the target score (the score corresponding to the industry excellent value) of each quantization index, wherein the deviation rate is = (target score-actual score)/target score multiplied by 100%, and judging the index with the deviation rate more than or equal to 30% as a short-board index; step two, association analysis The association relation between the short-board index and other indexes is mined by adopting an improved Apriori algorithm, and the core cause is positioned, and the specific technical means and implementation steps are as follows: (1) The data preprocessing comprises ① data discretization, namely converting continuous numerical values of all quantized indexes into discrete labels, dividing the discrete labels into high, medium and low 3 types of labels by adopting an equal frequency discrete method, constructing ② transaction sets, namely taking daily operation data of a single service area as 1 transaction record, wherein each record comprises discrete labels of all indexes, screening out a transaction set T comprising at least 1 short-plate index as an improved Apriori algorithm input to ensure a data focusing short-plate related scene, and removing ③ redundant data, namely deleting repeated records in the transaction set T and records of missing key index labels, wherein the sample size of the final transaction set is more than or equal to 1000, and ensuring the reliability of a correlation rule; (2) The core parameter setting comprises the steps of setting ①% of minimum support degree min_sup based on service area operation data characteristics, adjusting the sample size to 3% -4% if the sample size is large, setting 70% of minimum confidence degree min_conf in ②, and setting 4% of maximum item set length in ③; (3) Frequent item set mining Traversing a transaction set T, counting the occurrence frequency of each discrete label of a single index, calculating the support degree, wherein the support degree=the number of transactions containing the label/the total number of transactions, and screening out the label set with the support degree not less than min < as the 1-frequent item set, namely, the single index state of high-frequency occurrence; An iteration generating k-frequent item set, wherein k is more than or equal to 2, a ① connection step is performed, wherein k-1 frequent item sets are generated through self connection, a Hash tree is adopted to store the candidate k-item sets, a parallel computing mechanism is introduced to conduct slicing processing on a transaction set T, ② pruning step is performed, whether all k-1 subsets of the candidate k-item sets belong to the k-1 frequent item sets or not is checked, if not, the candidate sets are removed, ③ support screening is performed, occurrence frequency of the candidate k-item sets in the transaction set T is counted, support is calculated, a set with support of more than or equal to min_sup is screened out to serve as the k-frequent item sets, ④ termination condition is performed, iteration is stopped when a new k-frequent item set cannot be generated, and all the frequent item sets (L1U L2U-shaped Lk) are output; (4) Establishing association rules, namely generating ① rules, namely traversing all non-empty proper subsets for each k-frequent item set (Lk, k is more than or equal to 2), generating association rules of leading parts to trailing parts, wherein the leading parts are required to contain at least 1 short-board index label, the trailing parts are required to be non-short-board index labels or other short-board index labels, the leading parts and the trailing parts have no intersection, ② confidence calculation is that confidence= (support of the leading parts and the trailing parts)/(support of the leading parts), ③ rules screening is that the association rules with the confidence of more than or equal to min_conf are reserved, meanwhile, the promotion degree is calculated, namely the promotion degree = confidence/support of the trailing parts, screening the rules with the promotion degree of more than or equal to 1.2, and ④ rules verification is that the screened core association rules are verified through reserved 10% transaction data, the confidence of the rules on the verification set is calculated, the condition that the confidence of the rule is more than or equal to 65% is required to be met, and the rule is ensured to have generalization capability; (5) Sorting association rules, namely sorting the screened association rules according to a descending order of lifting degree, and preferentially analyzing rules with high lifting degree, wherein the causal association of the rules is stronger; (6) And (3) positioning the core causes, namely analyzing and positioning the short-board causes through a rule chain, wherein the short-board causes comprise direct causes and indirect causes, and the priority ranking is carried out on the causes by combining the confidence level, the lifting degree and the business importance of indexes of the rules, and the core causes of the first three are output.
  8. 8. A computer device comprising a memory, a processor and a computer program stored on the memory, characterized in that the processor executes the computer program to carry out the steps of the method according to any one of claims 1-7.
  9. 9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any of claims 1-7.
  10. 10. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any of claims 1-7.

Description

Service area operation management level evaluation method, device, medium and program product Technical Field The invention relates to the technical field of intelligent traffic, in particular to a service area operation management level assessment method, equipment, medium and program product. Background The expressway service area is used as a core supporting facility of a traffic network, and plays a role in providing basic services such as rest, supply, toilet, energy supply and the like for drivers and passengers, and the running management level of the expressway service area is more directly related to public travel experience, road network traffic efficiency and overall service image of traffic industry. Currently, with the continuous increase of expressway mileage, the increase of the keeping quantity of motor vehicles and the rapid increase of the cross-regional travel demands, the service scenes of the service area are continuously rich, and the function positioning is transformed from the traditional basic supply station to the comprehensive service hub, so that the system comprises multiple dimensions of public basic service, public service, commercial operation, cost control, comprehensive management and the like. Under the background, the service area operation management level is scientifically and comprehensively evaluated, so that the service area operation management level becomes the core requirement of traffic management departments and operation units, the advantages and the short boards of each service area can be accurately mastered through evaluation, decision basis is provided for resource optimal allocation, service quality improvement and management strategy adjustment, and standardized and refined management of the service areas is realized by assistance. Through investigation, the current service area management departments and operation units commonly adopt an empirical evaluation method (hereinafter referred to as a traditional method) combining manual field check and single dimension index statistics, and evaluation is completed by setting basic evaluation items, field check by evaluation staff and combining part of statistical data, and the following links are specifically covered: (one) evaluation dimension and index setting The evaluation dimension is single, the basic running state and the surface service quality of the public infrastructure are focused, and a multi-dimension full coverage system is not formed. The core indexes are classified into two types, namely an infrastructure operation index such as the number of parking spaces, toilet positions, charging piles and basic perfection rate, only statistics of two states of normal operation or failure, no quantitative utilization rate data, and a service quality subjective evaluation index such as on-site health conditions, personnel service attitudes and the like. Some schemes supplement a small number of business related indicators (e.g., number of business stores), but have no business operational efficiency quantified data. (II) data acquisition The method mainly comprises three collection modes, namely, manual on-site checking, on-site observation, on-site counting, recording of information such as parking space use, toilet occupancy, facility perfection, sanitary conditions and the like by an evaluator, and subjective judgment and scoring, and further, simple data statistics, manual recording and summarizing of the number of monthly fault facilities, the number of commercial store operations and the like, and no automatic collection and verification mechanism, and third, complaint data summarizing, namely, only collecting the number of monthly user complaints, not discriminating complaint validity and not associating specific service links. (III) score evaluation A mode of 'scoring of the terms and simple summation' is adopted, fixed scores are set for all evaluation indexes, such as 30 scores for full of the sanitary condition and 20 scores for full of the facility integrity, the evaluation personnel are combined with on-site check results and statistical data to judge and score according to experience, such as 25 scores when the sanitary condition is good and 15 scores when the sanitary condition is general, all the scoring values are directly summed to obtain an overall evaluation score, and a part of schemes set weights for a few indexes, but the weight distribution is random and no data support exists. (IV) evaluation result application The evaluation result is mainly used for basic grading of the service area, such as qualification/disqualification, excellent/good/general/poor, and provides a basic basis for annual assessment of an operation unit. Because the evaluation result lacks accuracy and comprehensiveness, it is difficult to support targeted resource allocation adjustment, service optimization and other fine management decisions. The core contradiction of the traditional method is that the 'compreh