An intelligent allocation method for electric power travel tasks

Abstract

The present invention provides a power task travel allocation method, including: generating a task model based on the current allocation task; screening candidate factors based on the task model; selecting evaluation factors from the candidate factors; classifying the selected evaluation factors into N sets of factors according to vehicle service attributes and driver service attributes; pairwise comparing the importance of the N sets of factors by category to obtain the relative weight of each set of factors relative to other sets of factors; obtaining the weight ratio of each evaluation factor; obtaining the evaluation score of each evaluation factor based on the weight ratio of each evaluation factor; obtaining the total evaluation score corresponding to the vehicle service attributes and each task attribute description, and obtaining the total evaluation score corresponding to the driver service attributes and each task attribute description, and performing driver and vehicle allocation.

Description

Technical Field

[0001] The present invention relates to the field of power distribution, and particularly to an intelligent allocation method for power travel tasks. Background Art

[0002] Power facilities are important infrastructure in our country. The facilities and equipment of power companies are distributed in different spaces. The maintenance and operation of power equipment are one of the most common and important tasks in the power grid. Therefore, power companies need to dispatch a large number of vehicles and drivers to perform tasks every day, with huge investment in human and material resources. The cost of a provincial power company in power task travel exceeds hundreds of millions of yuan. As the complexity of the power system is getting higher and higher, power task travel needs to balance safety and efficiency. How to achieve efficient and safe travel task scheduling of the power system is of great significance.

[0003] Currently, the related technologies in the field mainly focus on vehicle routing optimization methods, with the emphasis on path optimization to improve efficiency, and no in-depth consideration of task safety. Summary of the Invention

[0004] The purpose of the present invention is to propose an intelligent allocation method for power travel tasks based on the integration of multi-attributes of people, vehicles, and roads. Through the integration of driver portraits, vehicle portraits, and path attributes, intelligent allocation of safe and efficient power travel is achieved.

[0005] The technical solution of the present invention is as follows:

[0006] The present invention provides a power task travel allocation method, including:

[0007] Based on the current allocation task, generate a task model including target task attribute descriptions and a target rule set;

[0008] Based on the task model, screen out multiple candidate factors from a preset comprehensive factor set Z;

[0009] According to a preset selection rule, select evaluation factors from multiple candidate factors; classify the selected evaluation factors into N types of factor sets according to vehicle service attributes and driver service attributes;

[0010] Compare the importance of N types of factor sets pairwise by category to obtain the relative weight of each type of factor set relative to other types of factor sets;

[0011] Compare the importance of multiple evaluation factors within each type of factor set pairwise to obtain the relative weight of each evaluation factor within each type of factor set relative to other evaluation factors;

[0012] Based on the relative weights of each type of factor set with respect to other types of factor sets and the relative weights of each evaluation factor within each type of factor set with respect to other evaluation factors, the weight ratio of each evaluation factor is obtained;

[0013] Based on the weight ratio of each evaluation factor, the evaluation score of each evaluation factor is obtained;

[0014] Based on the evaluation scores of each evaluation factor, the total evaluation score corresponding to the vehicle business attributes and each task attribute description, and the total evaluation score corresponding to the driver business attributes and each task attribute description are obtained;

[0015] Based on the first total evaluation score corresponding to the vehicle business attributes and the target task attribute description and the second total evaluation score corresponding to the driver business attributes and the target task attribute description, driver and vehicle allocation are performed.

[0016] Preferably, in the task model, the target task attribute description includes road condition attributes and task priorities. The road condition attributes include road type and weather, and the task priority is one of the three dimensions of economy, efficiency, and safety;

[0017] The target rule set contains the vehicle types selected based on the road type.

[0018] Preferably, the preset comprehensive factor set includes a driver portrait factor set, a vehicle portrait factor set, and a path factor set;

[0019] The driver portrait factor set is derived from the physiological factors, psychological factors, and social activity factors of the driver;

[0020] The vehicle portrait factor set is derived from the attribute factors, health factors, and economic factors of the vehicle;

[0021] The path factor set is derived from the path attribute factors and the weather factors along the path;

[0022] The steps of screening out multiple candidate factors from the preset comprehensive factor set Z based on the task model include:

[0023] Selecting candidate factors corresponding to the task model from the driver portrait factor set, the vehicle portrait factor set, and the path factor set respectively.

[0024] Preferably, the steps of selecting evaluation factors from multiple candidate factors according to the preset selection rules include:

[0025] First, screening out the first type of evaluation factors that have been selected by all pre-determined industry evaluation experts from multiple first candidate factors;

[0026] Then, select the second type of evaluation factors that have been selected by more than half of all the pre-determined industry evaluation experts from the remaining candidate factors;

[0027] The first type of evaluation factors and the second type of evaluation factors together constitute the required evaluation factors.

[0028] Preferably, in the step of pairwise comparing the importance of N types of factor sets by category to obtain the relative weights of each type of factor set relative to other types of factor sets:

[0029] The importance and relative weights of each type of factor set relative to other types of factor sets are determined in advance by the average importance given by all industry evaluation experts;

[0030] In the step of pairwise comparing the importance of multiple evaluation factors within each type of factor set to obtain the relative weights of each evaluation factor within each type of factor set relative to other evaluation factors:

[0031] The importance and relative weights of each evaluation factor relative to other evaluation factors are determined in advance by the average importance given by all industry evaluation experts;

[0032] The steps of obtaining the weight ratio of each evaluation factor based on the relative weights of each type of factor set relative to other types of factor sets and the relative weights of each evaluation factor within each type of factor set relative to other evaluation factors include:

[0033] Multiply the sum of the relative weights of each evaluation factor relative to other evaluation factors by the sum of the relative weights of the type of factor set to which it belongs relative to other types of factor sets to obtain the weight ratio of each evaluation factor.

[0034] Preferably, the steps of obtaining the evaluation score of each evaluation factor based on the weight ratio of each evaluation factor include:

[0035] Multiply the weight ratio of each evaluation factor by the average score given by the corresponding experts to obtain the evaluation score of each evaluation factor;

[0036] Among them, the average score given by the experts corresponding to each evaluation factor is obtained by scoring in advance by business experts.

[0037] Preferably, the steps of performing driver and vehicle allocation based on the vehicle business attributes and the first total evaluation score corresponding to each task attribute description, and the driver business attributes and the second total evaluation score corresponding to each task attribute description include:

[0038] Based on the task focus in the target task attribute description included in the task model, determine the preset evaluation threshold corresponding to the task focus;

[0039] When both the first total evaluation score and the second total evaluation score exceed their respective preset evaluation thresholds, the drivers and vehicles that match the evaluation factors are screened out, and tasks are assigned to the target driver and the target vehicle with the highest scores.

[0040] The technical effects of the present invention are as follows:

[0041] Based on the attributes of power travel tasks, through the integration of driver portraits, vehicle portraits, and path attributes, intelligent allocation for safe and efficient power travel is achieved. Description of the Drawings

[0042] Figure 1 It is a schematic flowchart of the method of the embodiment of the present invention. Detailed Embodiments

[0043] The present invention is based on the power task travel scenario, with safety as the first perspective, and analyzes and designs from three aspects of people, vehicles, and roads that dominate the safe and efficient travel of power tasks. First, through the improved Delphi method (Delphi method) that integrates prior knowledge in the field, element screening is carried out on driver portraits, vehicle portraits, and path factors, and factor sets of driver portraits, vehicle portraits, and path factors are generated through weighted ranking. The three factor sets form a comprehensive factor set, which is represented as Z, where Z = J ∪ C ∪ L, J is the factor set of driver portraits, C is the factor set of vehicle portraits, L is the factor set of path factors, and the path factor set mainly provides relevant support for the description of tasks.

[0044] The comprehensive factor set Z is the quantitative factor basis for intelligent allocation of travel. A representation model of travel tasks is designed. Based on the distribution of quantitative factors in the representation model, a weighting algorithm design is carried out for the quantitative factors. After obtaining the quantitative weights, travel risk calculation is performed to provide support for the intelligent allocation of power task travel.

[0045] 1. Method for generating the factor set J of driver portraits.

[0046] The generation of the factor set J of driver portraits comes from three aspects of information of the driver, namely the "physiological factors", "psychological factors", and "social activity factors" of the driver. The driver portrait factors are represented as the set J:

[0047] J = J1 ∪ J2 ∪ J3, where J1 represents the set of physiological factor portraits, J2 represents the set of psychological factors, and J3 is the set of social activity factor portraits. The relevant information sources or extraction methods of each part are as follows:

[0048] (1) Physiological factors: Based on the annual physical examination reports and daily medical events of driver employees, it mainly includes the driver's age, gender, vision, hearing, etc., as well as related diseases that affect the driver's normal driving. Specifically, it includes those with organic heart diseases, epilepsy, Meniere's disease, vertigo, hysteria, paralysis agitans, mental illness, dementia, and neurological diseases that affect limb movement and other diseases that impede safe driving; all physiological factors form the physiological factor set J1; J1 = {J11, J12, J13, …… J1n}, and the expression examples of each factor are: J11 = age, J12 = gender, J13 = vision, etc.;

[0049] (2) Psychological factors: Based on the relevant conditions of the mental health test of driver employees, the psychological factor portrait of the driver is described mainly from the following 7 aspects: normal intelligence, healthy emotions, sound will, coordinated behavior, adaptation to interpersonal relationships, appropriate reaction, and psychological characteristics conforming to age. All psychological factors form the physiological factor set J2; J2 = {J21, J22, J23, …… J2n}, and the expression examples of each factor are: J21 = intelligence, J22 = healthy emotions, J23 = sound will, etc.;

[0050] (3) Social activity factors: Based on the social activity behaviors of drivers, various relevant factors are refined, mainly including: traffic behavior events, economic events, emotional events, etc. Traffic events include: traffic violation situations, traffic accident situations, those who have engaged in drug use or injection within three years, or those who have not completed three years after being released from compulsory isolation drug rehabilitation measures, or those who are still addicted to dependent psychotropic drugs; those who have committed a crime by fleeing the scene after causing a traffic accident; driving a motor vehicle after drinking or being drunk; Economic events mainly include: whether there is usury, whether there is serious debt, etc.; Emotional events: whether the family is harmonious, whether there is a serious emotional dispute, etc. All social activity factors form the social activity set J3;

[0051] J3 = {J31, J32, J33, …… J3n}, and the expression examples of each factor are: J31 = traffic violation, J32 = traffic accident, J33 = usury, etc.

[0052] 2. Generation method of the vehicle portrait factor set C.

[0053] The generation of the vehicle portrait comes from information in three aspects of the vehicle, namely the "attribute factor", "health factor", and "vehicle economic factor" of the vehicle. The vehicle portrait factor is expressed as a set C: C = C1 ∪ C2 ∪ C3, where C1 represents the attribute factor portrait set, C2 represents the health factor portrait set, and C3 is the vehicle economic factor portrait set. The relevant information sources or extraction methods of each part are as follows:

[0054] (1) Attribute factors: The factors based on vehicle attributes include the brand, type, number of vehicle seats, theoretical cruising range, etc. of the vehicle. The set of vehicle attribute factors is C1, where C1 = {C11, C12, C13,..., C1n}, and the expression examples of each factor are: C11 = brand, C12 = type, C13 = number of seats, etc.;

[0055] (2) Health factors: The factors based on vehicle health include vehicle age, mileage, maintenance records, traffic accident records, etc. The set of vehicle health factors is C2, where C1 = {C21, C22, C23,..., C2n}, and the expression examples of each factor are: C21 = vehicle age, C12 = mileage, C13 = maintenance record, etc.;

[0056] (3) Economic factors: The vehicle economic factors mainly include factors such as fuel consumption per 100 kilometers and operation and maintenance cost per 100 kilometers. The set of vehicle economic factors is C3, where C1 = {C31, C32, C33,..., C3n}, and the expression examples of each factor are: C31 = fuel consumption per 100 kilometers, C32 = operation and maintenance cost per 100 kilometers, etc.;

[0057] 3. Generation method of the path factor set L.

[0058] The path factor set comes from two aspects of information, namely "path attribute factors" and "weather factors along the path". The path factor is expressed as a set L:

[0059] L = L1 ∪ L2, where L1 represents the set of path attribute factor portraits, and L2 represents the set of weather factors along the path. The relevant information sources or extraction methods of each part are as follows:

[0060] (1) Path attribute factors: The path attribute factors mainly include: highway, urban road, rural road, mountain road, daily congestion index, risk index, etc. The set of path attribute factors is L1, where L1 = {L11, L12, L13,..., L1n}, and the expression examples of each factor are: L11 = highway, L12 = urban road, L13 = mountain village road, etc.;

[0061] (2) Weather factors along the path: The weather factors along the path mainly include: rainy day, foggy day, snowy day, windy day, etc. The set of weather factors along the path is L2, where L2 = {L21, L22, L23,..., L2n}, and the expression examples of each factor are: L21 = rainy day, L22 = foggy day, L23 = snowy day, etc.

[0062] 4. Task model construction.

[0063] This part mainly presents a complete business process model representation method for a driver and vehicle allocation task: mainly including rule formulation and task attributes. The task model M is expressed as: M = T ∪ R, where T is the task description set and R is the rule formulation set. The specific model description is as follows:

[0064] (1) Task attribute description: Task attribute description includes two parts: road condition attributes and task focus. Among them, road condition attributes include: which type of road is mainly along the way, such as rural roads, and what is the weather on the task day, such as rainy or snowy days. Based on different task focuses, different intelligent scheduling schemes are recommended. The task focus mainly includes three categories: three dimensions of safety, efficiency, and economy, that is, T = {T1, T2, T3... Tn}, where T1 = safety, T2 = efficiency, T3 = economy, and so on.

[0065] (2) Rule formulation: Rule formulation is mainly based on the intelligent vehicle selection constraints driven by knowledge. For example: if the target task path of electric vehicle dispatch is mainly rural roads, then an off-road vehicle must be selected; the rule set is a limitation on task attributes, and various rules are combined into the rule set R;

[0066] 5. Factor weighting algorithm.

[0067] The factor weighting algorithm refers to selecting relevant candidate factors from the comprehensive factor set Z based on the task attribute description T and rule formulation R in the task representation model M to complete the evaluation of task allocation. It mainly includes two steps:

[0068] (1) Selection of candidate factors

[0069] Based on the task model M, select the corresponding candidate factors from the comprehensive factor set Z (for example, a total of 40);

[0070] For example, M is described as a dispatch task T1 with safety priority, the task path is rural road L13, and traveling on a rainy or snowy day L23. Combine the rule set R to evaluate and select candidate factors. The selection rules are as follows:

[0071] Selection rule: Use the comprehensive factor set Z guided by prior knowledge as the initial factor list, and use the Delphi method to preset the candidate factors as the index list to experts.

[0072] The recommended ranking of indicators refers to experts arranging the assessment indicators in order according to the priority of the indicators.

[0073] The generation process rules mainly include: first, experts independently complete the questionnaire during the indicator generation process; second, the selected assessment indicators are not limited to the comprehensive factor set Z; third, experts use the written online method; fourth, the number of indicators given by experts should not be less than the number in the factor comprehensive set Z.

[0074] After receiving the index research form with expert feedback, conduct the preliminary selection of indexes. If n expert answer sheets are received.

[0075] First step: Extract the k indexes jointly selected by the experts, that is, the number of occurrences of each of the k indexes is n times;

[0076] Second step: Take the average of the sum of the priorities of the k jointly selected indexes and arrange them in ascending order of values;

[0077] Third step: Incorporate the indexes that appear i times in the questionnaire into the assessment scope, where n / 2 > i > n, that is, the common opinions of more than half of the experts are incorporated into the assessment scope, a total of j;

[0078] Fourth step: Arrange the j indexes selected by more than half of the experts in descending order of the number of selected experts;

[0079] Fifth step: Combine the k indexes in the second step with the j indexes in the fourth step to form a new index resource pool (i.e., the evaluation factor resource pool).

[0080] Sixth step: Classify the k + j indexes based on AHP (Analytic Hierarchy Process) according to the six major business attributes of J1, J2, J3, C1, C2, and C3. Among them, J1, J2, and J3 are one of the vehicle business attributes and driver business attributes, and C1, C2, and C3 are the other of the vehicle business attributes and driver business attributes. The six major categories of indexes generate a new set of evaluation factors Z+, expressed as Z+ = {Z1, Z2, …… Zn}, where n = k + j;

[0081] (2) Weight assignment of evaluation factors

[0082] The weight assignment of evaluation factors is carried out based on the selected evaluation factors. The weight assignment of evaluation factors is used as the basis for quantitative calculation. Compare the evaluation factors in the set Z+ pairwise. To improve efficiency, the evaluation factors for vehicle selection and personnel selection are weighted separately:

[0083] The specific steps are as follows: The traditional Delphi method calculates the average of the weights given by experts, but its effectiveness is inferior to the improved AHP algorithm. The importance between factors is better after pairwise comparison and assignment. Specifically as shown in Table 1 below:

[0084] Table 1 Reference table for weight comparison

[0085]

[0086]

[0087] First: Assign weights to the six major categories of evaluation factor sets of J1, J2, J3, C1, C2, and C3, and regard each category of evaluation factor set as a factor.

[0088] The importance of the six categories of evaluation factors is compared pairwise, and a pairwise comparison judgment matrix is constructed and subjected to consistency testing.

[0089] Then: Calculate the relative weights of the compared evaluation factor sets with respect to other categories of evaluation factor sets from the judgment matrix.

[0090] Finally, use the recursive algorithm to calculate the weights of each evaluation factor in the six major categories of J1, J2, J3, C1, C2, and C3 with respect to all evaluation factors.

[0091] For example: If the relative weights of the compared evaluation factors are calculated from the judgment matrix. The weight distribution of {J1, J2, J3, C1, C2, C3} is:

[0092] WSi = {0.1, 0.28, 0.23, 0.16, 0.12, 0.11} (i = 1…6). Using the recursive algorithm, if the local weights of the indicators {J11, J12, J13} in category J1 are WS1i = {0.5, 0.3, 0.2} (i = 1…3) respectively, then their corresponding global weights are

[0093] W S1 multiplied by WS1i, that is, 0.1 * WS1i.

[0094] All (k + j) factors in the set Z+ can be weighted according to this method.

[0095] In this way, a set Z+ and the weight Wn of each of its factors are obtained:

[0096] Among them where n = 1……(k + j).

[0097] Intelligent allocation quantization score function Score: f(n)

[0098] Among them, Zn is the average score of the evaluation factor obtained by the expert scoring in advance by the business expert, and Wn is the weight corresponding to this evaluation factor.

[0099] The score function is used to dynamically evaluate and score the driver and the vehicle, and the scoring situation is for the decision - maker to use.

[0100] 6. Implementation of intelligent allocation.

[0101] The main processes of the implementation of intelligent allocation include task point selection, driver factor generation, vehicle factor generation, driver factor weight generation, and vehicle factor weight generation.

[0102] (1) Task key point selection: For example, in the task model, the focus of the allocation task with safety priority is T1, the road type in the road condition attribute is mountain road L13, and the weather is traveling in rain and snow weather L23, which is the initialization of the task; in addition, there is also the selection of the rule set R. For example, only off-road vehicle models can be selected for mountain roads, and rules such as binding safety modes in rainy and snowy days are used as the initialization of the task. Specific rules can be flexibly formulated according to user needs.

[0103] (2) The generation of driver factors, the generation of vehicle factors, and factor weighting are implemented according to the factor weighting algorithm in Part 5;

[0104] (3) Intelligent allocation based on the scoring function:

[0105] The intelligent allocation mode can be divided into two categories:

[0106] One is the verification score mode: The scores of the vehicle and the driver are compared with the task limit value. If the selected vehicle and person do not reach the score threshold, a prompt to replace the object will be given. An example is shown in Table 2 below:

[0107] Table 2

[0108] Mode Driver score threshold Vehicle score threshold T1 Safety mode 95 95 T2 High-efficiency mode 80 85 T3 Economy mode 85 80

[0109] For example: In the safety mode, the verification values are required: fJ(n)>95; and fC(n)>95, where fJ(n) is the driver score and fC(n) is the vehicle score.

[0110] The other is the score ranking mode within the resource pool (driver pool and vehicle pool): That is, after scoring all drivers and vehicles, the driver and vehicle with the highest score are pushed before allocation without verification.

Claims

1. A power task travel allocation method, characterized in that, Including: Generate a task model based on the current allocation task, which includes target task attribute descriptions and a target rule set; Based on the task model, screen out multiple candidate factors from the preset comprehensive factor set Z; Select evaluation factors from multiple candidate factors according to the preset selection rules; classify the selected evaluation factors into N types of factor sets according to vehicle business attributes and driver business attributes; Compare the importance of the N types of factor sets pairwise by category to obtain the relative weights of each type of factor set relative to other types of factor sets; Compare the importance of multiple evaluation factors within each type of factor set pairwise to obtain the relative weights of each evaluation factor within each type of factor set relative to other evaluation factors; Based on the relative weights of each type of factor set relative to other types of factor sets and the relative weights of each evaluation factor within each type of factor set relative to other evaluation factors, obtain the weight ratio of each evaluation factor; Based on the weight ratio of each evaluation factor, obtain the evaluation score of each evaluation factor; Based on the evaluation scores of each evaluation factor, obtain the first total evaluation score corresponding to the vehicle business attributes and each task attribute description, and obtain the second total evaluation score corresponding to the driver business attributes and each task attribute description; Based on the first total evaluation score corresponding to the vehicle business attributes and each task attribute description and the second total evaluation score corresponding to the driver business attributes and each task attribute description, perform driver and vehicle allocation; In the task model, the target task attribute description includes road condition attributes and task priorities. The road condition attributes include road type and weather, and the task priority is one of the three dimensions of economy, efficiency, and safety; The target rule set contains the vehicle types selected based on the road type; The preset comprehensive factor set includes a driver portrait factor set, a vehicle portrait factor set, and a path factor set; The driver portrait factor set is derived from the physiological factors, psychological factors, and social activity factors of the driver; The vehicle portrait factor set is derived from the attribute factors, health factors, and economic factors of the vehicle; The path factor set is derived from the path attribute factors and the weather factors along the path; The steps of screening out multiple candidate factors from the preset comprehensive factor set Z based on the task model include: Select candidate factors corresponding to the task model from the driver portrait factor set, the vehicle portrait factor set, and the path factor set respectively.

2. The method according to claim 1, wherein The steps of selecting evaluation factors from multiple candidate factors according to the preset selection rules include: First, screen out the first type of evaluation factors selected by all pre-determined industry evaluation experts from multiple first candidate factors; Then, screen out the second type of evaluation factors selected by more than half of all pre-determined industry evaluation experts from the remaining candidate factors; The first type of evaluation factors and the second type of evaluation factors together constitute the required evaluation factors.

3. The method according to claim 1, wherein In the step of comparing the importance of the N types of factor sets pairwise by category to obtain the relative weights of each type of factor set relative to other types of factor sets: The importance and relative weights of each type of factor set relative to other types of factor sets are determined by the average importance given by all industry evaluation experts in advance; In the step of pairwise comparing the importance of multiple evaluation factors within each type of factor set to obtain the relative weight of each evaluation factor within each type of factor set relative to other evaluation factors: The importance and relative weight of each evaluation factor relative to other evaluation factors are both determined in advance by the average importance given by all industry evaluation experts; The step of obtaining the weight ratio of each evaluation factor based on the relative weight of each type of factor set relative to other types of factor sets and the relative weight of each evaluation factor within each type of factor set relative to other evaluation factors includes: Multiplying the sum of the relative weights of each evaluation factor relative to other evaluation factors by the sum of the relative weights of the type of factor set to which it belongs relative to other types of factor sets to obtain the weight ratio of each evaluation factor.

4. The method according to claim 1, wherein The step of obtaining the evaluation score of each evaluation factor based on the weight ratio of each evaluation factor includes: Multiplying the weight ratio of each evaluation factor by the average score given by the corresponding expert to obtain the evaluation score of each evaluation factor; Among them, the average score given by the expert corresponding to each evaluation factor is obtained by scoring in advance by business experts.

5. The method according to claim 1, wherein The step of performing driver and vehicle allocation based on the vehicle business attributes and the first total evaluation score corresponding to each task attribute description, and the driver business attributes and the second total evaluation score corresponding to each task attribute description includes: Based on the task focus in the target task attribute description included in the task model, determining a preset evaluation threshold corresponding to the task focus; When both the first total evaluation score and the second total evaluation score exceed their respective corresponding preset evaluation thresholds, screening out the drivers and vehicles that match the evaluation factors, and assigning tasks to the target driver and target vehicle with the highest scores.

Images