Intelligent decision-making method for logistics transportation service

Abstract

The invention belongs to the field of medicine logistics transportation, and particularly relates to a logistics transportation service intelligent decision-making method, which comprises the steps of S1, acquiring industry constraint information of a to-be-delivered medicine order; s2, dynamically determining a decision strategy suitable for the current order based on the industry constraint information, the decision strategy defining a quantitative trade-off relationship among cost, service quality compliance and operation risk; s3, performing multi-target comprehensive evaluation on a plurality of candidate logistics service providers according to the decision strategy, and generating a comprehensive evaluation result fusing predicted distribution cost and historical compliance risk performance of the logistics service providers; s4, according to the comprehensive evaluation result, selecting a logistics service provider to execute order delivery; and S5, based on actual delivery result data fed back after the selected logistics service provider executes the order, performing optimization iteration on the decision strategy determination logic in the step S2.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of medical logistics transportation, more specifically, particularly relates to a logistics transportation service intelligent decision-making method. BACKGROUND

[0002] The logistics distribution of medical products is a key link to ensure the shelf life and quality safety of drugs, which is strictly bound by laws and regulations such as GSP (Good Supply Practice). Unlike ordinary commodity logistics, which pursues the single target optimization of the lowest cost or the fastest time, medical logistics decision-making is essentially a complex optimization problem with multiple targets and strong constraints. The core dilemma is reflected in three aspects:

[0003] First, for medical logistics, business requirements such as ensuring a 2-8℃ cold chain throughout the entire process and delivering to the hospital within 2 hours before surgery are absolute and qualitative commands. Existing logistics optimization systems deal with continuous and negotiable variables such as time windows and ordinary damage rates, which cannot internalize such rigid constraints as quantifiable parameters that the model can understand and weigh. Enterprises usually use manual exclusion methods to first filter out service providers that meet basic conditions, and then make a second selection based on price, which is a cumbersome process and cannot guarantee comprehensive optimization.

[0004] Second, under the premise of meeting the above rigid constraints, decision-makers still need to balance multiple conflicting targets such as economy (transportation cost), compliance quality (such as temperature control compliance rate, invoice accuracy), and operational risk (such as delay risk, damage risk). Existing technical solutions are mostly built around a single target, with the model objective function being cost minimization, and compliance quality not being included as a quantifiable competitive target in the optimization framework. Simply adding medical characteristics to such systems can only get a more optimal cost solution, but cannot automatically identify service providers with better compliance records within an acceptable cost range.

[0005] Third, for orders with different characteristics, the priority of each target should change dynamically. For example, the weight requirements for time efficiency and compliance of urgent vaccine distribution are much higher than those of regular drug replenishment, and the weight of risk control for orders sent to tertiary hospitals is usually higher than that for orders sent to retail pharmacies. The existing lack of automated decision-making mechanisms that can adaptively adjust evaluation standards according to order characteristics.

[0006] In summary, the existing technology lacks a systematic solution that can automatically translate the rigid business constraints of the medical industry into quantifiable decision-making parameters and build an intelligent decision-making framework that supports dynamic weighting of multiple targets. This leads to the current medical logistics service provider selection still highly dependent on experience and judgment, with problems such as opaque decision-making, low overall efficiency, and high implicit risks. SUMMARY

[0007] Based on the problems mentioned in the background art, the present application provides a logistics transportation service intelligent decision-making method.

[0008] In a first aspect, the technical solution adopted by the present application is as follows: a logistics transportation service intelligent decision-making method, comprising the following steps: S1, obtaining industry constraint information of a to-be-delivered medical order; S2, dynamically determining a decision-making strategy suitable for the current order based on the industry constraint information, the decision-making strategy defining a quantitative trade-off relationship between cost, service quality compliance, and operational risk; S3, according to the decision-making strategy, performing multi-objective comprehensive evaluation on multiple candidate logistics service providers, generating a comprehensive evaluation result that integrates predicted delivery cost and historical compliance risk performance of the logistics service provider; S4, selecting a logistics service provider to execute order delivery according to the comprehensive evaluation result; S5, based on the actual delivery result data fed back by the selected logistics service provider after executing the order, optimizing and iterating the decision-making strategy determination logic in step S2.

[0009] Further, the step S2 of dynamically determining a decision-making strategy suitable for the current order specifically includes: according to the temperature control level in the industry constraint information, selecting one from a plurality of pre-defined strategy templates, each strategy template pre-defining a weight coefficient combination of cost, service quality compliance, and operational risk.

[0010] Further, the multi-objective comprehensive evaluation in step S3 specifically includes the following steps:

[0011] S31. Associate the feature information of the current order with the identification information of each candidate service provider, input the cost prediction model, and obtain the predicted delivery cost of each service provider; S32. According to the weight coefficient combination defined by the selected strategy template, calculate the comprehensive evaluation result of each service provider using a scoring function, the scoring function being: comprehensive evaluation result = Wc*F(predicted delivery cost) + Wq*G(service quality compliance) - Wr*H(operation risk); wherein Wc, Wq, and Wr are weight coefficients defined in the strategy template.

[0012] Further, the quantitative basis of the service quality compliance indicator includes at least the historical medicine temperature control transportation compliance rate of the candidate service provider; the quantitative basis of the operational risk indicator includes at least the historical cargo loss rate or transportation abnormal event rate of the candidate service provider.

[0013] Further, the cost prediction model is built by: acquiring a historical medicine distribution data set containing service provider identities, actual total settlement fees, and order characteristics; encoding the order characteristics, wherein the service provider identities are one-hot encoded or embedded encoded to form training features; using the actual total settlement fees as a prediction target, using gradient boosting decision trees or neural network models to perform supervised learning on the training features to obtain the cost prediction model.

[0014] Further, the optimization iteration in the step S5 includes: using newly added data containing actual distribution costs and actual compliance results to perform incremental training or retraining on the cost prediction model; based on the actual effects reflected by the newly added data, performing regression calibration and updating on the weight combination in the pre-set strategy template.

[0015] In a second aspect, the technical solution adopted by the present application is as follows: a logistics transportation service intelligent decision system, comprising a constraint analysis module for acquiring and analyzing industry constraint information of a to-be-distributed medicine order; a strategy decision module for dynamically outputting a corresponding decision strategy based on the industry constraint information, the decision strategy comprising weight definitions for multiple evaluation targets; a multi-target evaluation engine for performing comprehensive evaluation on multiple candidate service providers based on the decision strategy and generating a ranking result; a decision execution module for triggering service provider selection and order distribution process according to the ranking result; and a strategy optimization module for receiving actual result data after order distribution and optimizing the internal logic of the strategy decision module based on the actual result data.

[0016] Further, the multi-target evaluation engine comprises: a cost prediction unit for calling a pre-trained cost prediction model to output a predicted cost based on order characteristics and service provider identities; and a comprehensive score unit for calculating comprehensive scores of each service provider according to the weight definitions output by the strategy decision module, in combination with the predicted cost and historical compliance and risk indicators obtained from a service provider performance database.

[0017] Further, the strategy optimization module comprises: a model updating unit for retraining the cost prediction model using cost and characteristic data in the actual result data; and a weight tuning unit for analyzing the relationship between compliance achievement and selected service provider scores in the actual result data and automatically adjusting the pre-set weight combination in the strategy decision module.

[0018] The present application has the following beneficial effects:

[0019] 1、The technical features of the present application form a close synergistic effect, and together solve the multi-constraint multi-target dynamic optimization problem proposed in the background art.

[0020] Firstly, the business constraints are converted into policy weights, providing a dynamic evaluation scale to achieve scene adaptation; secondly, the multi-objective scoring function is the only algorithm implementation of executing the policy defined by S2, otherwise the policy output by S2 cannot act on the service provider selection; finally, the policy optimization of S5 depends on the comparison and analysis of the scoring results and actual results of S3, and the optimization goal is to improve the future decision effectiveness of the policy determined by S2. The three constitute an indivisible optimization closed loop.

[0021] 2. In the actual operation and procurement decision-making scene of medical B2B logistics, the control of transportation cost and the guarantee of distribution compliance often constitute a direct resource allocation contradiction in the short term. Selecting a higher specification of temperature control transportation service will inevitably result in higher immediate transportation expenses; while pursuing the lowest transportation cost may mean compromising on equipment, processes or personnel qualifications, thereby introducing potential compliance risks.

[0022] The multi-objective dynamic trade-off function invented by the present application is just to systematically deal with this realistic contradiction. The function places the predicted cost representing short-term economy and the compliance quality and operational risk representing long-term reliability in the same mathematical framework, and through configurable weight coefficients: Wc, Wq, Wr, it explicitly quantifies the current inclination of the decision maker to this contradiction. For example, increasing Wq means giving higher priority to compliance in the current decision, and being willing to bear higher cost for this; on the contrary, increasing Wc means that cost control is the primary consideration; more importantly, these weights are not fixed but can be intelligently adjusted according to the customer type and other situational information of the order. BRIEF DESCRIPTION OF DRAWINGS

[0023] The present application can be further illustrated by the non-limiting embodiments shown in the accompanying drawings;

[0024] Figure 1 The method flowchart of the present application;

[0025] Figure 2 The system schematic diagram of the present application; DETAILED DESCRIPTION

[0026] Example 1:

[0027] The method of the present embodiment mainly includes the following steps:

[0028] S1, obtaining industry constraint information of the medical order to be distributed;

[0029] Among them, the industry constraint information, the system receives the delivery request from the order management system, and extracts the key industry constraint information, which contains a data set of multiple completed delivery records, each record at least includes: shipping warehouse location, customer location, cargo weight, volume / piece, service level, such as ordinary / urgent / cold chain, customer type, transport company identification, actual settlement transportation cost, delivery date / time, loading and unloading conditions, whether remote area, temperature control level, destination type and other characteristic variables. For example:

[0030] Temperature control level (T_level): room temperature (RT), refrigeration (2-8C), freezing (FZ).

[0031] Destination type (D_type): tertiary hospital (H3), community center (CC), retail pharmacy (RP). These information collectively define the core business constraint set that the current order faces.

[0032] S2, based on the industry constraint information, dynamically determine the decision strategy suitable for the current order, specifically including: according to the temperature control level in the industry constraint information, select one from a plurality of pre-set strategy templates, each strategy template pre-defines a weight coefficient combination of cost, service quality compliance and operation risk;

[0033] Specifically, a plurality of strategy templates are pre-set in the system, each template is a triple {Wc, Wq, Wr}, representing the basic weight of cost, compliance quality and operation risk in the comprehensive score. Among them, the process of obtaining the weight coefficient is: collect a large amount of historical order data, i.e. the industry constraint information mentioned in S1, select the decision data including temperature control level, destination type, transport company, total cost, etc., through statistical analysis and machine learning method, identify which weight distribution can produce the optimal comprehensive operation effect under the combination of specific order characteristics. Secondly, an optimization model is established to minimize the long-term comprehensive cost under the condition of meeting business constraints, including quality loss cost, by solving the optimization problem, the recommended weight combination for different combinations is obtained, i.e. the strategy template, finally the recommended weight is audited, combined with business experience to fine-tune, form the built-in strategy template library in the system.

[0034] The above implementation based on the strategy template is only one preferred embodiment of the present application. Those skilled in the art can understand that the function of dynamically determining the decision strategy can also be implemented by other technical means, for example: based on a machine learning classifier: training a classification model to directly predict the optimal weight coefficients {Wc, Wq, Wr} according to the industry constraint information of the order; based on reinforcement learning: constructing an agent to continuously optimize the decision strategy through interaction with the environment, such as order dispatching-feedback; based on real-time optimization solution: each time a decision is made, a multi-objective optimization problem is solved according to the current order characteristics and real-time service provider data to dynamically generate weights. Regardless of the specific technical means adopted, the core is to realize the idea of dynamically adjusting the decision focus according to the industry constraint information of the order.

[0035] Specifically, for example:

[0036] Strategy matching logic: trigger strategy selection according to T_level and D_type.

[0037] Rule R1: if T_level = refrigerated and D_type = tertiary hospital, select strategy template A, where Wq (compliance quality weight) is set to a high value.

[0038] Rule R2: if T_level = room temperature and D_type = retail pharmacy, select strategy template B, where Wc (cost weight) is relatively high. The weight setting in rules R1 and R2 is a technical product of combining data-driven and domain knowledge, rather than purely subjective experience. This strategy template optimized based on historical data can ensure the scientificity and effectiveness of the decision. Through this step, rigid business requirements are converted into quantifiable strategy instructions that guide subsequent evaluation.

[0039] S3, according to the decision strategy, performing multi-objective comprehensive evaluation on the plurality of candidate logistics service providers, to generate a comprehensive evaluation result that integrates the predicted distribution cost and the historical compliance risk performance of the logistics service providers;

[0040] This step quantitatively compares and selects the candidate service providers under the guidance of the selected strategy.

[0041] Step S31: associate the feature information of the current order with the identification information of each candidate service provider, input the cost prediction model, and obtain the predicted distribution cost of each service provider. For each candidate service provider Carrier_i, the system constructs a feature vector X_i, which includes order-specific features such as weight, volume, distance, and service provider identification one-hot encoding. Input X_i into the cost prediction model to obtain the predicted cost C_pred_i. The cost prediction model can learn the different pricing of different service providers for the same order.

[0042] From the service provider performance database, real-time access to Carrier_i key historical indicators:

[0043] Compliance quality indicators Q_i: such as: cold chain transportation compliance rate in the past half year, etc.

[0044] Operational risk indicators R_i: such as: cargo loss rate in the past half year, etc.

[0045] Step S1033: According to the weight coefficient combination defined in the selected strategy template, the comprehensive evaluation result of each service provider is calculated by using the scoring function, and the scoring function is: comprehensive evaluation result=Wc*F (predicted distribution cost)+Wq*G (service quality compliance)-Wr*H (operation risk); Wherein, Wc, Wq, Wr are respectively the weight coefficients defined in the strategy template.

[0046] The scoring function is used to calculate the comprehensive score Score_i of each service provider: Score_i=Wc*F(C_pred_i)+Wq*G(Q_i)-Wr*H(R_i) Wherein, Wc, Wq, Wr come from the strategy template selected in step S102. F(), G(), H() are respectively the standardization function. In a preferred embodiment, these standardization functions are normalized functions that map the original values to the [0,1] interval. The score Score_i intuitively reflects the comprehensive advantages and disadvantages of the service provider under this particular strategy.

[0047] All candidate service providers are ranked in descending order of Score_i, and the ranking list is output. The highest score is the optimal service provider recommended by the system under the current strategy.

[0048] Wherein the cost prediction model is constructed as follows:

[0049] First, from the enterprise's historical business system, all completed pharmaceutical distribution order data in a period of time, such as the past 24 months, is obtained to construct a training data set. Each data record contains at least the following fields:

[0050] Order features: cargo weight, volume, temperature control level, delivery warehouse code, customer type code, distribution distance, etc.

[0051] Participant identification: unique identification of transportation service providers (carrier_id);

[0052] Target variable: actual total settlement cost, including basic transportation cost, additional cost, tax, etc.

[0053] Second, the original data is cleaned and converted to generate a feature vector available for the model:

[0054] Numerical Feature Standardization: Standardize continuous features like weight, volume, distance, etc. to have a mean of 0 and a variance of 1.

[0055] Categorical Feature Encoding:

[0056] Service Provider Identification Encoding: This is crucial for the model to learn the pricing differences among different service providers. Use one of the following two encoding methods:

[0057] One-hot Encoding: Create a binary feature column for each service provider. For example, if there are N service providers, create N feature columns. When an order is shipped by a certain service provider, the corresponding column value is 1, and the others are 0.

[0058] Embedding Encoding: Map service provider identities to a low-dimensional dense vector space. This method is particularly suitable for cases with a large number of service providers, as it can capture potential similarities between service providers.

[0059] Other Categorical Feature Encoding: For features like temperature control level, customer type, warehouse encoding, etc., also use one-hot encoding or ordinal encoding.

[0060] Feature Combination: According to business knowledge, construct interaction features, such as charging weight: take the larger value of actual weight and volumetric weight, interaction terms between distance and temperature control level, etc.

[0061] Finally, Model Selection and Training

[0062] Among them, the algorithm selection uses gradient boosting decision trees, such as XGBoost, LightGBM, or deep neural networks as the basic learning algorithm. These two algorithms have advantages in handling structured data and capturing nonlinear relationships.

[0063] Divide the processed dataset into training set, validation set and test set in chronological order or randomly; the proportion is 70%, 15% and 15% respectively.

[0064] Take the total actual settlement fee as the prediction target, and train the model on the training set. Use the validation set for hyperparameter tuning during training to prevent overfitting; the goal of the model is to learn the complex mapping relationship between order features including service provider identification and actual transportation cost, especially the different pricing patterns of different service providers for the same order features.

[0065] Evaluate the prediction performance of the model on the test set using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) as evaluation indicators.

[0066] The model meeting the performance requirements, such as MAPE < 15%, is serialized into a file format and deployed to a prediction service in a production environment. A model version management mechanism is established to support iterative updating and A / B testing of the model.

[0067] Regarding model optimization and updating, as new order data accumulates, periodically, such as monthly, update model parameters using incremental learning methods to adapt to market changes; the difference between actual transportation costs and predicted costs is used as a feedback signal for continuous optimization of the model.

[0068] The cost prediction model constructed through the above process can accurately predict the transportation costs of different service providers for specific pharmaceutical orders, providing reliable cost input for subsequent multi-objective comprehensive evaluation.

[0069] S4, according to the comprehensive evaluation result, select a logistics service provider to execute the order delivery;

[0070] S5, based on the actual delivery result data fed back by the selected logistics service provider after executing the order, optimize and iterate the decision strategy determination logic in step S2.

[0071] This step ensures the self-evolution of the system.

[0072] Examples of strategy optimization are as follows:

[0073] The system analyzes the actual effect of this decision, for example, suppose the service provider A with the highest score is selected, but there is a cargo loss. The system will judge that under the current strategy {Wc, Wq, Wr}, the punishment for the risk term R_i may be insufficient. An automatic tuning algorithm can be: calculate an ideal score Score_ideal, which should be able to make the underperforming service provider A drop to a reasonable position in the ranking. By solving backward, the value of Wr can be fine-tuned. Through the accumulation of a large number of orders, the system can automatically calibrate the weight coefficients in each strategy template, making them more suitable for actual business results, forming a complete closed loop of decision-making, execution, feedback, and optimization.

[0074] The technical features of the present application form a close synergistic effect, and together solve the multi-constraint multi-objective dynamic optimization problem proposed in the background art.

[0075] Firstly, business constraints are converted into strategy weights, providing a dynamic evaluation scale and realizing scene adaptation; secondly, the multi-objective scoring function is the only algorithm implementation of the strategy defined in S2, otherwise the strategy output by S2 cannot be applied to service provider comparison and selection; finally, the strategy tuning of S5 relies on the comparison and analysis of the scoring results and actual results of S3, and its optimization goal is to improve the future decision effectiveness of the strategy determined by S2. The three constitute an indivisible optimization closed loop.

[0076] Embodiment two:

[0077] The system of the embodiment comprises:

[0078] A constraint analysis module is configured to acquire and analyze industry constraint information of a medical order to be delivered.

[0079] A strategy decision module is configured to dynamically output a corresponding decision strategy based on the industry constraint information, wherein the decision strategy comprises weight definition of multiple evaluation targets.

[0080] A multi-target evaluation engine comprises: a cost prediction unit configured to call a pre-trained cost prediction model to output a predicted cost based on order features and service provider identifiers; and a comprehensive score unit configured to calculate a comprehensive score of each service provider according to the weight definition output by the strategy decision module, in combination with the predicted cost and historical compliance and risk indicators obtained from a service provider performance database. The multi-target evaluation engine performs comprehensive evaluation on multiple candidate service providers based on the decision strategy and generates a ranking result.

[0081] A decision execution module is configured to trigger service provider selection and order delivery process according to the ranking result.

[0082] A strategy optimization module comprises: a model updating unit configured to retrain the cost prediction model using cost and feature data in actual result data; and a weight tuning unit configured to analyze the relationship between compliance achievement and selected service provider score in actual result data and automatically adjust the weight combination preset in the strategy decision module.

[0083] Each service communicates through an API gateway and adopts a micro-service architecture to ensure high availability and scalability.

[0084] The present application has been described in detail. The description of the specific embodiments is only used to help understand the method of the present application and its core idea. It should be noted that for those skilled in the art, without departing from the principles of the present application, some improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A smart decision-making method for logistics transportation services, characterized in that: include S1. Obtain industry constraint information for pharmaceutical orders awaiting delivery; S2. Based on the industry constraint information, dynamically determine the decision strategy applicable to the current order, wherein the decision strategy defines a quantitative trade-off between cost, service quality compliance, and operational risk; S3. Based on the decision-making strategy, conduct a multi-objective comprehensive evaluation of multiple candidate logistics service providers to generate a comprehensive evaluation result that integrates predicted delivery costs and the logistics service providers' historical compliance risk performance. S4. Based on the comprehensive evaluation results, select a logistics service provider to execute order delivery; S5. Based on the actual delivery results data fed back by the selected logistics service provider after executing the order, optimize and iterate the decision-making strategy determination logic in step S2.

2. The intelligent decision-making method for logistics transportation services according to claim 1, characterized in that: The step S2 of dynamically determining the decision strategy applicable to the current order specifically includes: selecting one from multiple preset strategy templates based on the temperature control level in the industry constraint information. Each strategy template predefines a combination of weight coefficients for cost, service quality compliance, and operational risk.

3. The intelligent decision-making method for logistics transportation services according to claim 2, characterized in that: The multi-objective comprehensive evaluation in step S3 specifically includes the following steps: S31. Associate the feature information of the current order with the identification information of each candidate service provider, input the information into the cost prediction model, and obtain the predicted delivery cost of each service provider; S32. Based on the weight coefficient combination defined in the selected strategy template, calculate the comprehensive evaluation result of each service provider using a scoring function, wherein the scoring function is: comprehensive evaluation result = Wc*F (predicted delivery cost) + Wq*G (service quality compliance) - Wr*H (operational risk); where Wc, Wq, and Wr are the weight coefficients defined in the strategy template, respectively.

4. The intelligent decision-making method for logistics transportation services according to claim 3, characterized in that: The quantitative basis for the service quality compliance indicators includes at least the candidate service provider's historical compliance rate for temperature-controlled drug transportation; the quantitative basis for the operational risk indicators includes at least the candidate service provider's historical damage rate or transportation anomaly rate.

5. The intelligent decision-making method for logistics transportation services according to claim 3, characterized in that: The cost prediction model is constructed in the following way: Obtain a historical pharmaceutical delivery dataset that includes service provider identifiers, total actual settlement costs, and order characteristics; The order features are encoded, wherein the service provider identifier is one-hot encoded or embedded encoded to form training features; Using the actual total settlement cost as the prediction target, the training features are supervised learning using a gradient boosting decision tree or neural network model to obtain the cost prediction model.

6. The intelligent decision-making method for logistics transportation services according to claim 5, characterized in that: The optimization iteration in step S5 includes: The cost prediction model is incrementally trained or retrained using new data that includes actual delivery costs and actual compliance results. Based on the actual effects reflected by the new data, regression calibration and updates are performed on the weight coefficient combinations in the preset strategy template.

7. An intelligent decision-making system for logistics transportation services, characterized in that: include The constraint parsing module is used to obtain and parse industry constraint information for pharmaceutical orders to be delivered; The strategy decision module is used to dynamically output corresponding decision strategies based on the industry constraint information. The decision strategies include weight definitions for multiple evaluation objectives. A multi-objective evaluation engine is used to comprehensively evaluate multiple candidate service providers and generate ranking results based on the decision-making strategy. The decision execution module is used to trigger the service provider selection and order delivery process based on the sorting results; The strategy optimization module is used to receive the actual result data after order delivery and optimize the internal logic of the strategy decision module based on this data.

8. The intelligent decision-making system for logistics transportation services according to claim 7, characterized in that: The multi-objective evaluation engine includes: The cost prediction unit is used to call a pre-trained cost prediction model and output the predicted cost based on order characteristics and service provider identification. The comprehensive scoring unit is used to calculate the comprehensive score of each service provider based on the weight definition output by the strategy decision module, combined with the predicted cost and historical compliance and risk indicators obtained from the service provider performance database.

9. The intelligent decision-making system for logistics transportation services according to claim 8, characterized in that: The strategy optimization module includes: The model update unit is used to retrain the cost prediction model using cost and feature data from the actual result data. The weighting optimization unit is used to analyze the relationship between compliance achievement in actual result data and the scores of selected service providers, and to automatically adjust the pre-set weight combinations in the strategy decision module.

Images