Multi-algorithm dynamic routing processing system and method for intelligent container

Abstract

The invention discloses an intelligent container multi-algorithm dynamic routing processing system, which comprises a routing decision engine module, a configuration management center module and a dynamic weight adjustment module: the routing decision engine module is used for receiving an identification request and analyzing a merchant ID, a container type and a timestamp; acquiring historical performance data and loading personalized configuration; carrying out availability check, basic capability matching and cost threshold screening and filtering on the candidate algorithms to obtain a plurality of candidate algorithms; performing quality dimension, economic dimension and performance dimension real-time scoring on the plurality of candidate algorithms by using a multi-dimensional scoring model; and performing candidate algorithm score sorting according to the weighted total score of each dimension, and finally selecting the candidate algorithm with the highest score in the plurality of candidate algorithms as a target algorithm to execute the commodity identification task.

Description

Technical Field

[0001] This invention relates to information processing for smart vending machines, and more particularly to a multi-algorithm dynamic routing processing method and system for smart vending machines. Background Technology

[0002] In recent years, vending machines have been widely used in the fields of smart retail and automated vending, including unmanned vending, smart retail, and intelligent logistics management. Existing identification systems often employ fixed algorithms or single routing strategies, making it difficult to simultaneously achieve high identification accuracy, low response latency, and cost control in complex environments. Specifically, this manifests in the following aspects: Traditional systems primarily rely on a single recognition algorithm. When faced with containers of different models or configurations, the lack of targeted optimization can easily lead to recognition errors or unstable recognition speeds. In scenarios with high traffic or multiple tasks running concurrently, fixed routing strategies struggle to dynamically allocate resources and adjust strategies, potentially resulting in prolonged response times and impacting the overall user experience. Traditional systems mainly adopt a reactive approach to problems, failing to leverage real-time data and contextual information for proactive prevention and optimization. Consequently, they often cannot meet the demands of complex market environments with diverse scenarios and requirements in practical applications. Summary of the Invention

[0003] To address the aforementioned technical shortcomings, this invention proposes a multi-algorithm dynamic routing processing system for intelligent vending machines. By integrating a routing decision engine, a configuration management center, and a dynamic weight adjustment module, it achieves real-time screening and comprehensive scoring of candidate algorithms. The system adaptively adjusts weights based on different time periods and merchant levels, ensuring both stable recognition and fast response times across various usage scenarios, while also achieving an optimized balance between cost and performance. This system not only boasts high recognition accuracy but also enhances overall system resilience and availability through proactive preventative optimization mechanisms, demonstrating broad market application prospects and practical promotional value.

[0004] On one hand, the present invention provides a smart vending machine multi-algorithm dynamic routing processing system, the system comprising a routing decision engine module, a configuration management center module, and a dynamic weight adjustment module: The routing decision engine module is used to receive identification requests and parse merchant ID, vending machine type, and timestamp; acquire historical performance data and load personalized configurations; perform availability checks, basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; use a multi-dimensional scoring model to score multiple candidate algorithms in real time on quality, economic, and performance dimensions; the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined by response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally the candidate algorithm with the highest score among multiple candidate algorithms is selected as the target algorithm to execute this product identification task; The dynamic weight adjustment module is used to dynamically adjust the scoring weight of the multi-dimensional scoring model based on the preset time period of the request timestamp and / or the merchant level customized by the merchant when using a multi-dimensional scoring model (including quality dimension, economic dimension and performance dimension) to score multiple candidate algorithms in real time. The configuration management center module is used to set personalized configurations for merchants, time period weighting rules, and vending machine types.

[0005] On the other hand, the present invention also provides a multi-algorithm dynamic routing processing method for smart vending machines, including the following operations: The routing decision engine module receives identification requests and parses merchant ID, vending machine type, and timestamp; acquires historical performance data and loads personalized configurations; performs availability checks (including circuit breaker status judgment), basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; uses a multi-dimensional scoring model to score multiple candidate algorithms in real time across quality, economic, and performance dimensions; the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined using response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally, the candidate algorithm with the highest score among multiple candidate algorithms is selected as the target algorithm to execute this product identification task; The dynamic weight adjustment module dynamically adjusts the scoring weights of the multi-dimensional scoring model based on a preset time period and / or the merchant level customized by the merchant, when using a multi-dimensional scoring model (including quality, economic, and performance dimensions) to score multiple candidate algorithms in real time. It should be noted that the above steps are based on a time-based weight adaptation strategy, adjusting the weights of recognition time, cost, and system availability during peak, off-peak, and nighttime periods respectively. A merchant level-based weight customization strategy is also implemented, where high-end merchants and price-sensitive merchants adjust the weights of recognition success rate, dispute rate, cost, response time, and system availability according to their respective needs.

[0006] Preferably, as one possible implementation; before receiving the identification request and parsing the merchant ID, container type, and timestamp, the following is also included: The configuration management center module allows you to set personalized configurations for merchants, time period weighting rules, and vending machine types.

[0007] Preferably, as one possible implementation; receiving an identification request and parsing the merchant ID, container type, and timestamp, including; Receive an identification request and parse the context information in the identification request, wherein the context information includes at least the requesting merchant identifier, the requesting container type, and the request timestamp; After parsing and recognizing the context information in the request, the merchant ID, container type, and timestamp are obtained.

[0008] Preferably, as one possible implementation, the quality dimension scoring of the multi-dimensional scoring model in the routing decision engine includes: Based on the current container type, the preset mapping table between container type and historical recognition accuracy is called to determine the basic success rate score (based on the historical recognition accuracy, corresponding weights are assigned). Based on the current container type, the preset container type and scene adaptation mapping table is called to determine the scene adaptation rate score of the current container (considering the special recognition capabilities under specific container types). Based on the current time period, a preset time stability mapping table is invoked to determine the time stability score (stability is identified based on different time periods). Quality dimension score = Basic success rate score × preset weight + Scene adaptability score × preset weight + Time stability score × preset weight.

[0009] Preferably, as one feasible implementation; the economic dimension score is determined based on the cost of a single identification, tiered pricing, and merchant cost preferences, wherein: Obtain the actual cost of a single recognition operation, as well as the preset minimum and maximum costs; The formula for calculating the cost score for a single recognition operation is as follows: Single recognition cost score = 1 - (actual cost - minimum cost) / (maximum cost - minimum cost); Simultaneously, it obtains the call volume based on monthly sales, determines the tiered price discount coefficient based on the call volume based on monthly sales, multiplies the tiered price discount coefficient by the preset price to equal the tiered price, and obtains the set merchant cost preference value and preset preference weight to achieve a comprehensive score in the economic dimension. The overall score for the economic dimension = single recognition cost score × preset single recognition weight + tiered price × preset price weight + merchant cost preference value × preset preference weight.

[0010] Preferably, as one possible implementation; the performance dimension score is determined based on the response time score grading, and the calculation result is as follows: The preset relationship between the above response time and range includes: ; Obtain the response time and calculate the time score based on the preset relationship between the above response time and range.

[0011] Preferably, as one possible implementation, the dynamic weight adjustment module obtains the time corresponding to the current time period; and adjusts the weights of each scoring dimension according to different time periods.

[0012] Preferably, as one possible implementation, after obtaining multiple candidate algorithms through usability checks, basic capability matching, and cost threshold filtering, the method further includes: Monitor and statistically analyze the recent anomaly rate and response timeout rate of the current candidate recognition algorithms; The current candidate identification algorithm is detected according to the preset circuit breaker rules. If the current candidate identification algorithm meets the preset circuit breaker rules, the circuit breaker elimination algorithm is executed. The preset circuit breaker rules include: if the recent identification anomaly rate of the current candidate identification algorithm exceeds the first threshold, or the response timeout rate of the current candidate identification algorithm exceeds the second threshold, the candidate identification algorithm is marked as circuit breaker and excluded from the set of available algorithms.

[0013] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages: Analysis of the intelligent vending machine multi-algorithm dynamic routing processing method and system provided by this invention reveals that, in specific applications, the method firstly involves the routing decision engine module receiving the identification request and parsing the merchant ID, vending machine type, and timestamp; acquiring historical performance data and loading personalized configurations; performing availability checks (including circuit breaker status judgment), basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; using a multi-dimensional scoring model to score the multiple candidate algorithms in real time across quality, economic, and performance dimensions; where the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined using response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally, the candidate algorithm with the highest score among the multiple candidate algorithms is selected as the target algorithm to execute this product identification task; Meanwhile, when the dynamic weight adjustment module uses a multi-dimensional scoring model (including quality, economic and performance dimensions) to score multiple candidate algorithms in real time, it dynamically adjusts the scoring weight of the multi-dimensional scoring model based on the preset time period of the request timestamp and / or the merchant level customized by the merchant. On the one hand, the above-mentioned intelligent vending machine multi-algorithm dynamic routing processing method adopts a multi-dimensional comprehensive scoring method, that is, through the comprehensive scoring of quality dimension (basic success rate, scenario adaptability rate, time stability), economic dimension (cost, tiered pricing, merchant preference) and performance dimension (response time), it ensures that the optimal algorithm is selected for each recognition task.

[0014] The scoring process also utilizes a dynamic weight adjustment function, which adjusts the weights of each dimension based on different time periods (such as peak / off-peak / nighttime) and merchant level (high-end / price-sensitive) to make the algorithm selection more in line with the needs of actual scenarios, thereby improving the overall recognition accuracy and stability. Attached Figure Description

[0015] Figure 1 This is a schematic diagram illustrating the principle structure of the intelligent vending machine multi-algorithm dynamic routing processing system provided by the present invention. Figure 2 This is an operation flowchart of the intelligent vending machine multi-algorithm dynamic routing processing method provided by the present invention; Figure 3 This is a specific operation flowchart of the intelligent vending machine multi-algorithm dynamic routing processing method provided by the present invention; Figure 4 This is a schematic diagram of the structure of a storage medium for applying the multi-algorithm dynamic routing processing method for smart vending machines provided by the present invention.

[0016] Labels: Memory 1130; Communication interface 1120; Processor 1110; Computer storage medium 1140. Detailed Implementation

[0017] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0019] Example 1 like Figure 1 As shown, this embodiment of the invention provides a smart vending machine multi-algorithm dynamic routing processing system, including a routing decision engine module 10, a dynamic weight adjustment module 20, and a configuration management center module 30. The routing decision engine module 10 is used to receive identification requests and parse merchant ID, vending machine type, and timestamp; acquire historical performance data and load personalized configurations; perform availability checks, basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; use a multi-dimensional scoring model to score multiple candidate algorithms in real time on quality, economic, and performance dimensions; the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined by response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally the candidate algorithm with the highest score among multiple candidate algorithms is selected as the target algorithm to execute this product identification task; The dynamic weight adjustment module 20 is used to dynamically adjust the scoring weight of the multi-dimensional scoring model based on the preset time period of the request timestamp and / or the merchant level customized by the merchant end when using a multi-dimensional scoring model (including quality dimension, economic dimension and performance dimension) to score multiple candidate algorithms in real time. The configuration management center module 30 is used to set personalized configurations for merchants, time period weighting rules, and cabinet types.

[0020] The routing decision engine module receives the identification request and parses the merchant ID, vending machine type, and timestamp; it acquires historical performance data and loads personalized configurations; it performs availability checks (including circuit breaker status judgment), basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; it uses a multi-dimensional scoring model to score the multiple candidate algorithms in real time across quality, economic, and performance dimensions; the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined using response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally, the candidate algorithm with the highest score among multiple candidate algorithms is selected as the target algorithm to execute this product identification task; It scores based on historical performance data, container type mapping relationships, and time stability mapping tables, providing data support for decision-making and reducing the misjudgment rate.

[0021] Example 2 See Figure 2 Embodiment 2 of the present invention provides a multi-algorithm dynamic routing processing method for smart vending machines, including the following operations: The S100 routing decision engine module receives the identification request and parses the merchant ID, vending machine type, and timestamp; it acquires historical performance data and loads personalized configurations; it performs availability checks (including circuit breaker status judgment), basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; it uses a multi-dimensional scoring model to score multiple candidate algorithms in real time on quality, economic, and performance dimensions; the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined by response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally, the candidate algorithm with the highest score among multiple candidate algorithms is selected as the target algorithm to execute this product identification task; S200: When the dynamic weight adjustment module uses a multi-dimensional scoring model (including quality, economic and performance dimensions) to score multiple candidate algorithms in real time, it dynamically adjusts the scoring weight of the multi-dimensional scoring model based on the preset time period of the request timestamp and / or the merchant level customized by the merchant. It should be noted that the above steps are based on a time-based weighted adaptive strategy, which adjusts the weights of recognition time, cost, and system availability scores during peak, off-peak, and nighttime periods respectively; and a merchant-level-based weighted customization strategy, in which high-end merchants and price-sensitive merchants adjust the weights of recognition success rate, dispute rate, cost, response time, and system availability according to their respective needs.

[0022] Preferably, as one possible implementation; before receiving the identification request and parsing the merchant ID, container type, and timestamp, the following is also included: S300, the configuration management center module sets the merchant's personalized configuration, time period weighting rules, and container type.

[0023] Preferably, as one possible implementation; receiving an identification request and parsing the merchant ID, container type, and timestamp, including; Receive an identification request and parse the context information in the identification request, wherein the context information includes at least the requesting merchant identifier, the requesting container type, and the request timestamp; After parsing and recognizing the context information in the request, the merchant ID, container type, and timestamp are obtained.

[0024] Preferably, as one possible implementation, the quality dimension scoring of the multi-dimensional scoring model in the routing decision engine includes: Based on the current container type, the preset mapping table between container type and historical recognition accuracy is called to determine the basic success rate score (based on the historical recognition accuracy, corresponding weights are assigned). Based on the current container type, the preset container type and scene adaptation mapping table is called to determine the scene adaptation rate score of the current container (considering the special recognition capabilities under specific container types). Based on the current time period, a preset time stability mapping table is invoked to determine the time stability score (stability is identified based on different time periods). Quality dimension score = Basic success rate score × preset weight + Scene adaptability score × preset weight + Time stability score × preset weight.

[0025] Furthermore, to improve the above scheme, a dispute rate scoring mechanism can be introduced. For example, the basic success rate score can be multiplied by a dispute rate to reduce the score of the project. This will not be elaborated further.

[0026] Preferably, as one feasible implementation; the economic dimension score is determined based on the cost of a single identification, tiered pricing, and merchant cost preferences, wherein: Obtain the actual cost of a single recognition operation, as well as the preset minimum and maximum costs; The formula for calculating the cost score for a single recognition operation is as follows: Single recognition cost score = 1 - (actual cost - minimum cost) / (maximum cost - minimum cost); Simultaneously, it obtains the call volume based on monthly sales, determines the tiered price discount coefficient based on the call volume based on monthly sales, multiplies the tiered price discount coefficient by the preset price to equal the tiered price, and obtains the set merchant cost preference value and preset preference weight to achieve a comprehensive score in the economic dimension. The overall score for the economic dimension = single recognition cost score × preset single recognition weight + tiered price × preset price weight + merchant cost preference value × preset preference weight.

[0027] Preferably, as one possible implementation; the performance dimension score is determined based on the response time score grading, and the calculation result is as follows: The preset relationship between the above response time and range includes: ; Obtain the response time and calculate the time score based on the preset relationship between the above response time and range.

[0028] Preferably, as one possible implementation, the dynamic weight adjustment module obtains the time corresponding to the current time period; and adjusts the weights of each scoring dimension according to different time periods.

[0029] It should be noted that (for example: peak hours 07:00-09:00 and 17:00-19:00, off-peak hours 09:00-17:00, nighttime hours 22:00-06:00) and different merchant levels (such as high-end merchants and price-sensitive merchants).

[0030] The aforementioned system, by constructing an architecture that integrates a routing decision engine, a configuration management center, and a dynamic weight adjustment module, achieves a shift from "passively responding to problems" to "proactively preventing and optimizing," thus forming a flexible, intelligent, and economical product identification service system.

[0031] See Figure 3 After performing usability checks, basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms, the process also includes: S210. Monitor and statistically analyze the recent recognition anomaly rate and response timeout rate of the current candidate recognition algorithm; S220. Detect the current candidate identification algorithm according to the preset circuit breaker rules. If the current candidate identification algorithm meets the preset circuit breaker rules, execute the circuit breaker elimination algorithm operation. The preset circuit breaker rules include: if the recent identification anomaly rate of the current candidate identification algorithm exceeds the first threshold, or the response timeout rate of the current candidate identification algorithm exceeds the second threshold, the candidate identification algorithm is marked as circuit breaker and excluded from the set of available algorithms.

[0032] The above-mentioned circuit breaker method monitors the anomaly rate and timeout rate of candidate algorithms in real time. Once the threshold is exceeded, the algorithm is eliminated by circuit breaker to avoid the failure of the algorithm affecting the overall system stability.

[0033] Example 2 On the other hand, this second embodiment, based on the intelligent vending machine multi-algorithm dynamic routing processing method provided in the first embodiment, also provides a computer storage medium 1140 (hereinafter referred to as the storage medium). For example... Figure 4 The diagram shown is a schematic of a computer storage medium structure framework provided in Embodiment 3 of the present invention, which includes: Memory 1130 is used to store computer programs; The communication interface 1120 is used to connect the memory 1130 to the processor 1110; Processor 1110 is configured to execute a computer program to implement a multi-algorithm dynamic routing processing method for smart vending machines, as disclosed in any of the above-described embodiments.

[0034] In summary, the intelligent vending machine multi-algorithm dynamic routing processing system and method proposed in this invention employs a multi-dimensional comprehensive scoring approach. This involves a comprehensive score based on quality (basic success rate, scenario adaptability, time stability), economic (cost, tiered pricing, merchant preferences), and performance (response time) dimensions to ensure the optimal algorithm is selected for each identification task. Furthermore, a dynamic weight adjustment function is used during the scoring process. This adjusts the weights of each dimension according to different time periods (e.g., peak / off-peak / nighttime) and merchant levels (high-end / price-sensitive), making the algorithm selection more aligned with actual scenario requirements, thereby improving overall recognition accuracy and stability.

[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; those skilled in the art can modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart vending machine multi-algorithm dynamic routing processing system, characterized in that, The system includes a routing decision engine module, a configuration management center module, and a dynamic weight adjustment module. The routing decision engine module is used to receive identification requests and parse merchant ID, container type and timestamp; obtain historical performance data and load personalized configurations; The candidate algorithms were filtered for usability, basic capabilities, and cost thresholds to obtain multiple candidate algorithms. A multi-dimensional scoring model was used to score the candidate algorithms in real time across quality, economic, and performance dimensions. The quality dimension score considered basic success rate, scenario adaptability, and time stability; the economic dimension score was determined based on single recognition cost, tiered pricing, and merchant cost preferences; and the performance dimension score was determined by response time scoring. The candidate algorithms were ranked according to the weighted total score of each dimension, and the candidate algorithm with the highest score was finally selected as the target algorithm to perform this product recognition task. The dynamic weight adjustment module is used to dynamically adjust the scoring weight of the multi-dimensional scoring model based on the preset time period of the request timestamp and / or the merchant level customized by the merchant when using a multi-dimensional scoring model (including quality dimension, economic dimension and performance dimension) to score multiple candidate algorithms in real time. The configuration management center module is used to set personalized configurations for merchants, time period weighting rules, and vending machine types.

2. A multi-algorithm dynamic routing processing method for intelligent vending machines, characterized in that, This includes the following operations: The routing decision engine module receives identification requests and parses merchant ID, vending machine type, and timestamp; acquires historical performance data and loads personalized configurations; performs availability checks (including circuit breaker status judgment), basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms; uses a multi-dimensional scoring model to score multiple candidate algorithms in real time across quality, economic, and performance dimensions; the quality dimension score considers basic success rate, scenario adaptability, and time stability; the economic dimension score is determined based on single identification cost, tiered pricing, and merchant cost preferences; and the performance dimension score is determined using response time scoring grading; the candidate algorithms are ranked according to the weighted total score of each dimension, and finally, the candidate algorithm with the highest score among multiple candidate algorithms is selected as the target algorithm to execute this product identification task; The dynamic weight adjustment module dynamically adjusts the scoring weights of the multi-dimensional scoring model based on the preset time period of the request timestamp and / or the merchant level customized by the merchant. When using a multi-dimensional scoring model (including quality, economic and performance dimensions) to score multiple candidate algorithms in real time, the module adjusts the scoring weights of the multi-dimensional scoring model based on the merchant level customized by the merchant.

3. The method according to claim 2, characterized in that, Before receiving the identification request and parsing the merchant ID, container type, and timestamp, the following steps are also included: The configuration management center module allows you to set personalized configurations for merchants, time period weighting rules, and vending machine types.

4. The method according to claim 2, characterized in that, Receive identification requests and parse merchant ID, container type, and timestamp, including; Receive an identification request and parse the context information in the identification request, wherein the context information includes at least the requesting merchant identifier, the requesting container type, and the request timestamp; After parsing and recognizing the context information in the request, the merchant ID, container type, and timestamp are obtained.

5. The method according to claim 2, characterized in that, The quality dimension scoring of the multi-dimensional scoring model in the routing decision engine includes: Based on the current container type, the preset mapping table between container type and historical recognition accuracy is called to determine the basic success rate score (based on the historical recognition accuracy, corresponding weights are assigned). Based on the current container type, the preset container type and scene adaptation mapping table is called to determine the scene adaptation rate score of the current container (considering the special recognition capabilities under specific container types). Based on the current time period, a preset time stability mapping table is invoked to determine the time stability score (stability is identified based on different time periods). Quality dimension score = Basic success rate score × preset weight + Scene adaptability score × preset weight + Time stability score × preset weight.

6. The method according to claim 2, characterized in that, The economic dimension score is determined based on the cost of a single identification, tiered pricing, and merchant cost preferences, where: Obtain the actual cost of a single recognition operation, as well as the preset minimum and maximum costs; The formula for calculating the cost score for a single recognition operation is as follows: Single recognition cost score = 1 - (actual cost - minimum cost) / (maximum cost - minimum cost); Simultaneously, it obtains the call volume based on monthly sales, determines the tiered price discount coefficient based on the call volume based on monthly sales, multiplies the tiered price discount coefficient by the preset price to equal the tiered price, and obtains the set merchant cost preference value and preset preference weight to achieve a comprehensive score in the economic dimension. The overall score for the economic dimension = single recognition cost score × preset single recognition weight + tiered price × preset price weight + merchant cost preference value × preset preference weight.

7. The method according to claim 2, characterized in that, The performance dimension score is determined based on the response time score grading, and the calculation result is as follows: The preset relationship between the above response time and range includes: ； Obtain the response time and calculate the time score based on the preset relationship between the above response time and range.

8. The method according to claim 2, characterized in that, The dynamic weight adjustment module obtains the time corresponding to the current time period; and adjusts the weight of each scoring dimension according to different time periods.

9. The method according to claim 1, characterized in that, After performing usability checks, basic capability matching, and cost threshold filtering on candidate algorithms to obtain multiple candidate algorithms, the process also includes: Monitor and statistically analyze the recent anomaly rate and response timeout rate of the current candidate recognition algorithms; The current candidate identification algorithm is detected according to the preset circuit breaker rules. If the current candidate identification algorithm meets the preset circuit breaker rules, the circuit breaker elimination algorithm is executed. The preset circuit breaker rules include: if the recent identification anomaly rate of the current candidate identification algorithm exceeds the first threshold, or the response timeout rate of the current candidate identification algorithm exceeds the second threshold, the candidate identification algorithm is marked as circuit breaker and excluded from the set of available algorithms.

10. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the intelligent vending machine multi-algorithm dynamic routing processing method according to any one of claims 1-8.

Images