Game lineup balance test method, equipment, medium and product

Abstract

The embodiment of the invention relates to the large field of computer technology, and discloses a game lineup balance test method and device, a medium and a product. The method comprises the steps of determining a target game role combination in response to a target instruction; determining target equipment configuration according to the virtual equipment of the game role and a preset algorithm; the preset algorithm is used for recording historical attempt information of equipment configuration, and determining a selection strategy of the equipment configuration based on historical performance so as to assist in determining target equipment configuration; and carrying out a balance test according to the target game role combination and the target equipment configuration. The method and the device can be used at least for solving the technical problem of determining how to combine and match which weapon equipment and attributes which the equipment should select so as to realize better game experience and numerical value balance aiming at numerous game roles which are continuously increased along with versions in the related technology.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device, medium, and product for testing the balance of a game lineup. Background Art

[0002] RPGs (Role-Playing Games) are a genre centered around players taking on characters and engaging in adventures in fictional worlds. They emphasize character development, immersive storytelling, and the player's decision-making and interactions within the game world. In these games, players typically control one or more characters, advancing the plot through completing quests, engaging in combat, and exploring maps. Characters typically possess distinct attributes (such as health, attack power, and mana), skill sets, and growth paths (such as leveling up, changing classes, and enhancing equipment). Players must strategically enhance their characters' abilities through resource management to meet the challenges of the game.

[0003] Among related technologies, deep reinforcement learning has been applied to game balance testing. Reinforcement learning algorithms can control one or more characters in a game, learning through trial and error through interaction with the game environment. For a specific game lineup, reinforcement learning can be used to train an agent to master that lineup's combat strategies. This allows the agent to analyze whether the lineup's numerical configuration meets the designer's expectations, or to adjust the relevant character values ​​based on the analysis results to achieve numerical control goals.

[0004] However, the inventors have discovered that there are at least the following technical problems in the relevant technologies: when planners need to determine how to combine and match what weapons and equipment, as well as the attributes that should be selected for the numerous game characters that continue to increase with each version in the current game version's game character pool, in order to achieve a better gaming experience and numerical balance, conventional reinforcement learning technology is difficult to solve such problems. Summary of the Invention

[0005] One purpose of this application is to provide a balance testing method, equipment, medium and product for a game lineup, at least to solve the technical problem in related technologies of determining how to combine and match what weapons and equipment and the attributes that should be selected for the equipment to achieve a better gaming experience and numerical balance for the numerous game characters that continue to increase with each version.

[0006] To achieve the above objectives, some embodiments of the present application provide the following aspects:

[0007] In the first aspect, some embodiments of the present application also provide a balance testing method for a game lineup, the method comprising: determining a target game character combination in response to a target instruction; determining a target equipment configuration based on the virtual equipment of the game characters and a preset algorithm; the preset algorithm is used to record historical attempt information of the equipment configuration, and determine a selection strategy for the equipment configuration based on historical performance to assist in determining the target equipment configuration; and performing a balance test based on the target game character combination and the target equipment configuration.

[0008] In a second aspect, some embodiments of the present application further provide an electronic device comprising: one or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, cause the processor to perform the steps of the method described above.

[0009] In a third aspect, some embodiments of the present application further provide a computer-readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method described above.

[0010] In a fourth aspect, some embodiments of the present application further provide a computer program product, comprising a computer program / instruction, which implements the steps of the above-described method when executed by a processor.

[0011] Compared with related technologies, the solution provided in the embodiment of the present application can effectively solve the limitations of traditional reinforcement learning technology in dealing with complex combination search problems. The core lies in breaking through the exponentially growing combination space barrier through structured process design and intelligent algorithm support. Specifically, in this embodiment, the target game character combination is first determined by responding to the target instruction, and the test range is accurately defined from the huge game character pool to avoid indiscriminate search of all possible combinations, greatly reducing the initial search dimension; then the target equipment configuration is determined in combination with a preset algorithm: the algorithm records historical attempt information and generates a selection strategy based on historical performance. It can quickly converge to a reasonable configuration in multi-dimensional selections such as light cones, relic sets, accessories and attribute values, significantly improving the screening efficiency of equipment combinations and avoiding the dilemma of combination explosion. On this basis, the balance test based on the target game character combination and equipment configuration is greatly improved in terms of pertinence and accuracy, and can efficiently cover a variety of key combinations of game character pools, light cone pools, relic pools and corresponding attributes, thereby quickly locating the strong lineup of the current version. At the same time, this embodiment simplifies the planning process. You only need to fill in the test information and wait for the results. It is also convenient for configuring training and obtaining data analysis, further reducing the time and operation cost of processing complex combination problems, and meeting the planning needs for optimal combination search and numerical balance control. In other words,

[0012] Planners can use the algorithm to determine whether the team combinations found meet expectations. If there are any anomalies, they can conduct in-depth analysis of the causes (such as program bugs or numerical setting issues) to achieve precise control of overall balance. In this way, various unexpected abnormal combination problems can be discovered and resolved in advance, thereby preventing them from impacting game balance, reducing public opinion crises, and maintaining player retention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings. These exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are represented as similar elements. Unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0014] Figure 1 An exemplary flow chart of a method for testing the balance of a game lineup provided in some embodiments of the present application;

[0015] Figure 2 An exemplary structural diagram of an electronic device provided for some embodiments of the present application. DETAILED DESCRIPTION

[0016] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0017] The following terms are used in this document.

[0018] Genetic Algorithm (GA): A randomized global search optimization method whose core principle is to simulate natural selection and genetic phenomena such as replication, crossover, and mutation. Starting from an arbitrary initial population, the algorithm uses random selection, crossover, and mutation to select and generate a group of individuals that are better adapted to the environment, driving the population to evolve towards more optimal areas in the search space. After multiple generations of continuous reproduction and evolution, it eventually converges on a group of individuals that are most adapted to the environment, thereby obtaining a high-quality solution to the problem.

[0019] Reinforcement Learning (RL), also known as reinforcement learning, evaluation learning, or enhanced learning, is a key paradigm and methodology in machine learning. It focuses on the interaction between an agent and its environment, with the core goal of optimizing strategies through learning to maximize rewards or achieve specific goals.

[0020] Bandit Optimization: This is a type of single-step optimization algorithm. It uses continuous trial and error to ultimately select the optimal solution. It can be thought of as reinforcement learning that only performs a single step.

[0021] Balance testing: This is primarily used to test the capabilities of different characters and equipment combinations in the game. This testing reveals how the performance of different characters or equipment will change after attribute adjustments, thereby increasing the appeal of specific characters and controlling overall stat inflation.

[0022] First embodiment

[0023] The first embodiment of the present application relates to a method for testing the balance of a game lineup. It is understandable that the numerical balance of game characters and equipment in the game is crucial to the continued development of the game. Once a numerical imbalance occurs, it is very easy to impact operational work, trigger negative public opinion, and even lead to a decline in revenue. For example, in an open world game, the mismatch between the strength of a certain character and its background setting caused widespread dissatisfaction among players; a newly launched game character in a turn-based game had a numerical imbalance, and with its powerful output capabilities and special mechanisms, the experience of players who owned the game character was significantly different from that of players who did not own it, causing a lot of dissatisfaction. These cases fully illustrate that if the character abilities and related numerical values ​​are not designed to be controllable before the game goes online, it may cause a series of irreparable adverse consequences after going online.

[0024] Based on this, combined Figure 1 As shown, the method may include the following steps:

[0025] Step S101, in response to a target instruction, determining a target game character combination;

[0026] Step S102, determining a target equipment configuration based on the virtual equipment of the game character and a preset algorithm; the preset algorithm is used to record historical equipment configuration attempt information and determine an equipment configuration selection strategy based on historical performance to assist in determining the target equipment configuration;

[0027] Step S103: Perform a balance test based on the target game character combination and the target equipment configuration.

[0028] The following describes each of the above steps in detail.

[0029] For step S101, illustratively, in response to a target instruction, the skill characteristics, role positioning and synergy effects of different game characters can be analyzed, and a game character lineup with test value can be selected from a large number of possible combinations.

[0030] The target instruction may be, but is not limited to, a user input instruction, a system automatically generated instruction, a configuration file reading instruction, or an instruction triggered by another module, and this embodiment does not impose any specific limitations. For example, the user input instruction may be: a tester manually selecting or configuring a lineup; the system automatically generated instruction may be a test task automatically generated according to a test plan, a random algorithm, or preset rules; the configuration file reading instruction may be loading a lineup combination to be tested from an external configuration file; and the instruction triggered by another module may be a test instruction issued by a test management module.

[0031] The game character may be used to represent a character controlled by a player, or a character not controlled by a player, such as a monster, an NPC, etc., which is not specifically limited in this embodiment.

[0032] Regarding step S102, for example, a target equipment configuration can be determined based on the determined game character, the virtual equipment that can be used with the game character, and a preset algorithm. The preset algorithm can record historical trial information for various equipment configurations, such as the number of times different virtual equipment has been used in past tests and the matching effects. Based on this historical performance, an equipment configuration selection strategy can be formulated to assist in selecting the optimal equipment combination suitable for the game character, thereby enabling the game character's abilities to be properly displayed.

[0033] For step S103, illustratively, after obtaining the target game character combination and target equipment configuration, a team can be formed based on these configurations, and the team can actually operate in the game, such as fighting monsters in specific levels, challenging dungeons, etc. By observing the team's combat performance, win rate, output data, etc., it can be judged whether the lineup is balanced, providing a basis for adjusting and optimizing the game.

[0034] Regarding steps S101-S103, for example, take the balance test of an RPG game as an example: assuming that the target game character combination of "warrior + mage + priest" needs to be tested, for the warrior's virtual equipment, the preset algorithm will retrieve the historical attempt information of past configurations such as "heavy armor + battle axe" and "light armor + longsword", and combine its output and survival performance in actual combat to determine the priority selection of the "heavy armor (+15% defense) + battle axe (+20% critical strike)" strategy, and at the same time match the mage with the "flame robe (+30% spell damage)" and the priest with the "holy scepter (+25% healing amount)" to form the target equipment configuration; then, through simulated combat, task completion evaluation, etc., the overall combat effectiveness, professional coordination balance, etc. of the game character combination and equipment configuration are tested to determine whether a certain character is too strong or too weak.

[0035] Among related technologies, conventional reinforcement learning techniques have limitations when faced with specific problems: When game designers need to determine the optimal character combinations, weapon and equipment combinations, and equipment attribute selection for a large and growing number of characters in the current version of the game to achieve a better gaming experience and numerical balance, such problems are difficult to directly solve. Essentially, this has evolved into a complex combinatorial search problem: with the expansion of the game character pool, the richness of equipment types, and the diversity of attribute options, the combinatorial search space has grown exponentially, making it difficult for traditional evaluation techniques to quickly and accurately identify the optimal solution within this vast array of possibilities. For example, in a certain game, there are currently approximately 40 characters, and there are approximately 90,000 possible combinations for a four-person team alone. Furthermore, each character can have an average of 10 light cones, 10 relic sets, and 10 accessories. Each of the four relic positions has 7-8 attribute value options. Multiplying all these combinations together, the total number can reach a staggering scale.

[0036] It is not difficult to find that compared with related technologies, the solution provided by the embodiment of the present application can effectively solve the limitations of traditional reinforcement learning technology in dealing with complex combination search problems. The core lies in breaking through the exponentially growing combination space barrier through structured process design and intelligent algorithm support. Specifically, in this embodiment, the target game character combination is first determined by responding to the target instruction, and the test range is accurately defined from the huge game character pool to avoid indiscriminate search of all possible combinations, greatly reducing the initial search dimension; then the target equipment configuration is determined in combination with a preset algorithm: the algorithm records historical attempt information and generates a selection strategy based on historical performance. It can quickly converge to a reasonable configuration in multi-dimensional selections such as light cones, relic sets, accessories and attribute values, significantly improving the screening efficiency of equipment combinations and avoiding the dilemma of combination explosion. On this basis, the balance test based on the target game character combination and equipment configuration has greatly improved its pertinence and accuracy, and can efficiently cover a variety of key combinations of game character pools, light cone pools, relic pools and corresponding attributes, thereby quickly locating the strong lineup of the current version. At the same time, this embodiment simplifies the planning process. It only needs to fill in the test information and wait for the results. It is also convenient for configuring training and obtaining data analysis, further reducing the time and operating costs for handling complex combination problems, and meeting the planning needs for optimal combination search and numerical balance control. In other words, the planner can judge whether the team combination searched by the algorithm meets the expectations. If there are anomalies, the cause can be deeply analyzed (such as program loopholes or numerical setting problems) to achieve precise control of the overall balance. In this way, various unexpected abnormal combination problems can be discovered and solved in advance, thereby avoiding their impact on the balance of the game, reducing public opinion crises, and maintaining the player's willingness to stay.

[0037] Second embodiment

[0038] The second embodiment of the present application relates to a method for testing the balance of a game lineup. The second embodiment is an improvement on the first embodiment. The specific improvement is that: in this embodiment, a specific implementation method for determining a target game character combination in response to a target instruction is provided.

[0039] In some embodiments, the step of determining the target game character combination in response to the target instruction, i.e., step S101, may include:

[0040] Step S1011, determining an initial team lineup in response to the target instruction;

[0041] Step S1012: Determine a target game character combination based on the genetic algorithm and the game characters in the initial team lineup.

[0042] Specifically, the initial team lineup may include several game characters with different skill characteristics or career positioning; the initial team lineup is iteratively optimized based on a genetic algorithm: the genetic algorithm simulates the selection, crossover, mutation and other processes in biological evolution, and adjusts the combination of game characters in the initial lineup. For example, retain the game character combination with excellent performance as the "parent generation", "cross" the game character combination of different lineups to generate a new lineup, or randomly replace some game characters to achieve "mutation", and screen out a better combination based on the actual combat performance of the lineup (such as winning rate, output efficiency, survivability, etc.), so as to iteratively obtain a target game character combination with balanced performance and strong performance.

[0043] Optionally, in some embodiments, the method for determining the initial team lineup, i.e., step S1011, may include:

[0044] Step S10111: determining the number of parallel test groups based on a predefined population size parameter; the parallel test groups are used to execute balance tests in parallel in the game simulator;

[0045] Step S10112: determining an initial configuration of each test group based on the number of the parallel test groups; the initial configuration at least includes a game character and virtual equipment;

[0046] Step S10113: Determine the initial team lineup based on the initial configuration of all test groups.

[0047] Regarding step S10111, the value of the population size parameter can, for example, determine the number of parallel test groups. For example, when the population size parameter is set to 80, the system can determine the number of parallel test groups to be 80. These 80 test groups will simultaneously perform balance tests in the game simulator, significantly improving test efficiency and shortening the overall test cycle through parallel processing.

[0048] Regarding step S10112, for example, the elements of the initial configuration are directly related to the composition of the game lineup, and primarily include game characters and virtual equipment (specifically, types such as light cones and artifacts). For example, when there are 80 parallel test groups, each test group can clearly specify the game character participating in the test, as well as the light cones, artifacts worn by the character, and their main attributes, to ensure that each test group has a clear and specific initial state.

[0049] For step S10113, illustratively, by integrating the initial configurations of all test groups, a complete set of initial team lineups can be formed, which not only facilitates overall control of the test scope, but also provides basic data for subsequent iterative optimization based on genetic algorithms, so that the genetic algorithm can start from the initial team lineup and gradually optimize a better game character combination.

[0050] Take the initial team lineup determination process of an RPG game as an example: first execute step S10111. If the predefined population size parameter is 8, the number of parallel test groups is determined to be 8, and these test groups will conduct balance tests simultaneously in the game simulator; then enter step S10112, and set the initial configuration for each group according to the needs of the 8 parallel tests - for example, the first group is "Warrior A + Mage B + Priest C" with "Heavy Armor + Flame Staff + Holy Scepter", the second group is "Assassin D + Archer E + Tank F" with "Agility Cloak + Precision Bow + Solid Shield", and the remaining groups are also equipped with different game characters and virtual equipment; finally, through step S10113, the initial configurations of these 8 test groups are summarized to form an initial team lineup that includes a variety of game characters and equipment combinations.

[0051] Optionally, in some embodiments, determining the initial configuration of each test group may include: determining the game characters and virtual equipment of the test group according to a random selection scheme; or determining the game characters and virtual equipment of the test group according to preconfigured weight parameters.

[0052] Specifically, two methods can be used to determine the initial configuration of each test group: one is to randomly select game characters and corresponding virtual equipment from the game character library and virtual equipment library in the game, so as to reduce human intervention, cover a wider range of lineup combination possibilities through randomness, and provide diverse initial samples for testing; the other is to incorporate human experience into the initial configuration determination process based on the weight parameters pre-configured by the planner. For example, if the planner judges based on experience that weapon A performs better than weapon B, weapon A can be given a higher weight, so that the probability of weapon A being selected in the virtual equipment selection is relatively higher, so that the determination of the initial configuration has both algorithmic logic support and reflects the planner's experience judgment.

[0053] Optionally, in some embodiments, the genetic algorithm and the game characters in the initial team lineup determine the target game character combination, that is, step S1012 may include:

[0054] Step S10121, according to predefined genetic algorithm parameters, the initial team lineup enters the iteration phase;

[0055] Step S10122, in the iterative stage, the game character combination in the initial team lineup is optimized according to the crossover operation and mutation operation of the genetic algorithm to determine the target game character combination.

[0056] Specifically, after the initial team lineup is configured, the system can initiate an iterative process based on predefined genetic algorithm parameters, driving the initial lineup from a static initial state to a dynamic optimization phase. During the iterative process, the game character combinations in the initial team lineup are optimized through the genetic algorithm's crossover and mutation operations. The crossover operation can exchange some gene segments of the game character combinations that perform well in different test groups, while the mutation operation can randomly adjust the characters in some game character combinations to introduce new pairing possibilities. By combining these two operations, the game character combinations are continuously optimized. After multiple rounds of iteration, the combination with the best performance is screened out to obtain the target game character combination. The genetic algorithm's crossover and mutation operations can be implemented using existing technologies, but to avoid redundancy, they are not described in detail in this embodiment.

[0057] Optionally, in some embodiments, for the mutation operation, the method may further include:

[0058] Step S201: determining a high-quality lineup resource library based on historical balance test results; the high-quality lineup resource library is used to store high-scoring lineup combinations of game characters;

[0059] Step S202: In subsequent iterations, a historically strong combination is selected from the high-quality lineup resource library as a benchmark lineup according to a preset probability;

[0060] Step S203 , performing a mutation operation based on the benchmark lineup to generate derivative combinations, so as to improve the pertinence and efficiency of the game character combination optimization.

[0061] In traditional technology, as the project iterates, the pool of game characters will continue to expand. For example, dozens of new game characters will be added compared to the initial stage, which will cause the overall number of combinations to increase by several orders of magnitude. When the iteration time remains unchanged, the algorithm is prone to falling into the local optimal problem during the search process: it may only search for some strong new game characters, while ignoring the old game characters that were previously searchable. For example, under the same computing power and time conditions, the difficulty of selecting 4 teams from 20 game characters to search for strong combinations is significantly different from selecting 4 teams from 60 game characters to search: if the previously strong game character A is replaced by a stronger new game character B, the algorithm may tend to focus on searching for teams related to B due to limited resources, and thus ignore game character A; but for planners, the relevant teams of game character A are still valuable for analysis.

[0062] To solve the above problems, in this embodiment, a high-quality lineup resource library can be determined based on the results of historical balance tests to store high-scoring lineup combinations of game characters, that is, automatically extracting and recording strong pairings for specific game characters in each set of experimental search results; in subsequent iterations, historical strong combinations are selected from the resource library as benchmark lineups according to a preset probability, so that the old strong combinations become "anchor points" in the huge search space, allowing the algorithm to introduce historical experience in new search experiments in a timely manner; then, with the benchmark lineup as a reference, mutation operations are performed to generate derivative combinations, and the search is expanded around the historical strong combinations, helping the algorithm to escape from local optimal solutions and approach the global optimal solution, thereby improving the pertinence and efficiency of game character combination optimization, significantly increasing the diversity of search results, and meeting the analysis needs of planners.

[0063] Suppose that in a historical balance test, the combination of "Game Character A + Game Character B + Game Character C + Game Character D" was included in the high-quality lineup resource library with a high score. In subsequent iterations, the algorithm selects this combination as the baseline lineup according to a preset probability, and then performs a mutation operation: for example, replacing one of the game characters with game character E with the same position, or adjusting the virtual equipment configuration of some game characters to generate derivative combinations such as "Game Character A + Game Character B + Game Character C + Game Character E" and "Game Character A + Game Character B + Game Character F + Game Character D". In this way, the proven strong combination logic in the baseline lineup can be retained, and a variety of new combinations can be developed based on it. This eliminates the need for the algorithm to blindly explore the vast search space, but instead accurately optimizes based on valuable historical experience, effectively improving the pertinence and efficiency of game character combination optimization.

[0064] It is not difficult to find that in the embodiments of this application, by first responding to the target instruction and determining the initial team lineup, a diverse base sample can be provided for subsequent combination optimization, avoiding the search scope being limited to a local area. Then, a genetic algorithm is used to iteratively optimize the game characters in the initial lineup, using mechanisms such as selection, crossover, and mutation to screen out the better-performing combinations, and efficiently converge to a balanced and strong target combination within the vast pool of game characters. This approach not only ensures the comprehensiveness of the search, but also improves the efficiency and accuracy of combination screening through algorithmic iteration, thereby effectively addressing the combination complexity brought about by the increasing number of game characters.

[0065] Third embodiment

[0066] The third embodiment of this application relates to a method for testing the balance of a game lineup. The third embodiment is an improvement on the first embodiment. Specifically, the improvement is that this embodiment provides a specific implementation method for determining the target equipment configuration based on the virtual equipment of the game character and a preset algorithm.

[0067] In some embodiments, determining the target equipment configuration based on the virtual equipment of the game character and a preset algorithm, that is, step S102, may include:

[0068] Step S1021, determining a first attempt record for each type of virtual equipment and a second attempt record for attribute configuration corresponding to each game character; the first attempt record is used to dynamically track the number of attempts for each type of virtual equipment configuration during an iterative process; the second attempt record is used to dynamically track the number of attempts for attribute configuration during an iterative process;

[0069] Step S1022: Determine the target equipment configuration according to the first attempt record and the second attempt record.

[0070] Specifically, but not limited to, a gambling machine optimization algorithm can be used to determine the first attempt record of each type of virtual equipment and the second attempt record of attribute configuration for each game character, thereby dynamically tracking the number of attempts of each type of virtual equipment configuration and attribute configuration during the iteration process. For example, records can be maintained for each game character's light cone, artifact, and attribute, tracking the number of attempts during the iteration. Based on these records, the attempts of various virtual equipment and attribute configurations can be comprehensively analyzed to determine the target equipment configuration.

[0071] For example, taking a certain game character as an example, during the iteration process, the system can establish a first attempt record for each type of virtual equipment such as light cones and relics that can be matched with the game character, such as recording "Light Cone A was tried 15 times, Light Cone B was tried 8 times", "Relic Set C was tried 20 times" and other information; at the same time, a second attempt record is established for the attribute configuration of the game character (such as the numerical combination of attributes such as attack power and defense power), such as recording "The attribute combination with attack power priority was tried 12 times, the attribute combination with defense priority was tried 10 times" and other information. Based on these records, the system can comprehensively analyze the attempts of various virtual equipment and attribute configurations. For example, if Light Cone A has been tried more times and has stable performance, while a new relic set has been tried fewer times but has higher potential, the two can be combined to determine the final target equipment configuration of the game character.

[0072] Optionally, in some embodiments, the first attempt record of the virtual equipment may include an attempt record of a first type of virtual equipment and / or an attempt record of a second type of virtual equipment; wherein the first type of virtual equipment and the second type of virtual equipment are used to represent different types divided according to the game equipment system.

[0073] Specifically, the first category of virtual equipment primarily refers to equipment that is closely tied to the game character and plays a decisive role in that character's core abilities. For example, light cones often directly enhance a character's core skills or output attributes, providing a key support for their combat capabilities. The second category of virtual equipment primarily refers to equipment that can be flexibly matched and used to refine character attributes, such as relic artifacts. Through different set effects and attribute bonuses, they can specifically address character weaknesses or enhance performance in specific combat scenarios. These two types of equipment are divided according to the functional positioning and mechanism of the game equipment system, and together they constitute a complete system of virtual equipment.

[0074] Optionally, in some embodiments, determining the first attempt record and the second attempt record of the attribute configuration of each type of virtual equipment corresponding to each game character, i.e., step S1021, may include:

[0075] Step S10211, determining a target hierarchical architecture; wherein the first layer of the target hierarchical architecture is a game character combination optimization layer based on a genetic algorithm; the second layer is an optimization layer for the first type of virtual equipment; and the third layer is an optimization layer for the second type of virtual equipment and attribute configuration.

[0076] Step S10212, separate the decision variables according to the alternating optimization strategy, determine the number of attempts of the first type of virtual equipment through the second layer, and determine the number of attempts of the second type of virtual equipment and attribute configuration through the third layer to form the first attempt record and the second attempt record.

[0077] Specifically, the target hierarchical architecture can be divided into three layers. The bottom layer is a genetic algorithm-based game character combination optimization layer, which serves as the foundation because weapons and equipment must be attached to game characters. The second layer is an optimization layer for the first type of virtual equipment, such as the light cone (weapon) optimization layer. The third layer is an optimization layer for the second type of virtual equipment and attribute configuration, such as the relic and main attribute optimization layer. Then, according to the alternating optimization strategy, the decision variables are separated. During the optimization process, the second layer determines the number of attempts for the first type of virtual equipment, and the third layer determines the number of attempts for the second type of virtual equipment and attribute configuration, thereby forming the first attempt record and the second attempt record.

[0078] Optionally, in some embodiments, the decision variables are separated according to the alternating optimization strategy, the number of attempts for the first type of virtual equipment is determined through the second layer, and the number of attempts for the second type of virtual equipment and attribute configuration is determined through the third layer to form the first attempt record and the second attempt record. That is, step S10212 may include:

[0079] During the second-layer optimization process, the second-class virtual equipment and attribute configuration of the third layer are kept at the current optimal state, the first-class virtual equipment is optimized and the number of attempts is recorded;

[0080] During the third-level optimization process, the first-class virtual equipment of the second-level is kept in the current optimal state, and the second-class virtual equipment and attribute configuration are optimized and the number of attempts is recorded;

[0081] The decision variables are separated by the above-mentioned alternating optimization method to form a first attempt record and a second attempt record respectively.

[0082] Specifically, under the alternating optimization strategy, decision variables can be separated through hierarchical processing: when optimizing the second layer (such as first-class virtual equipment such as light cones), the second-class virtual equipment and attribute configurations of the third layer (such as relics and their main attributes) are kept in the current optimal state (or the configuration with the highest current score is selected), and the focus is on optimizing the first-class virtual equipment and recording its number of attempts; when optimizing the third layer, the first-class virtual equipment of the second layer is kept as the currently explored optimal configuration, and the focus is on optimizing the second-class virtual equipment and attribute configurations and recording the number of attempts. This alternating optimization method can effectively reduce the credit allocation problem and avoid the inability to clarify the specific role of each layer of equipment in high-scoring results due to simultaneous optimization of multiple layers, thereby reducing the ambiguity of the algorithm's understanding of the score, and ultimately forming a first attempt record recording the attempts of the first-class virtual equipment and a second attempt record recording the attempts of the second-class virtual equipment and attribute configurations.

[0083] For example, consider the optimization of light cones (the first category of virtual equipment) and relics and primary attributes (the second category of virtual equipment and attribute configuration): when optimizing light cones on the second level, keep the highest-scoring relic combination and primary attribute configuration on the third level unchanged, and focus on testing the effects of different light cones paired with this optimal relic configuration, while recording the number of attempts for each light cone. When optimizing relic and primary attributes on the third level, use the optimal light cone explored on the second level, test different relic combinations and primary attributes paired with this light cone, and simultaneously record the number of attempts for each relic and attribute configuration. This alternating optimization approach, for example, by first fixing "relic A + primary attribute B" to test the effects of light cones C, D, and E, and then fixing light cone C to test the effects of "relic F + primary attribute G" and "relic H + primary attribute I," allows us to clearly identify the independent impacts of different light cones, relics, and attributes on lineup performance, while also separately recording the first attempt record for light cones and the second attempt record for relic and attribute configurations.

[0084] It should be noted that this embodiment may also be an improvement based on the second embodiment.

[0085] It is not difficult to find that in the embodiment of the present application, by dynamically tracking the attempt records of various virtual equipment configurations and attribute configurations corresponding to each game character, the frequency of attempts of different equipment and attributes in the iterative process can be accurately grasped, providing data support for subsequent configuration selection; determining the target equipment configuration based on these records can avoid repeated attempts on inefficient combinations, and at the same time, can give priority to focusing on equipment and attribute combinations with better performance, significantly improving the pertinence and efficiency of equipment configuration screening, reducing invalid computing consumption, and thus providing a more reasonable equipment configuration basis for the overall balance test.

[0086] Fourth embodiment

[0087] The fourth embodiment of this application relates to a method for testing the balance of a game lineup. The fourth embodiment is an improvement on the third embodiment, specifically comprising: providing a specific implementation method for determining a target equipment configuration based on the first and second attempt records.

[0088] In some embodiments, the determining the target equipment configuration according to the first attempt record and the second attempt record, i.e., step S1022 may include:

[0089] Based on the first attempt record and the second attempt record, the configuration with the current highest score is selected as the target equipment configuration with a first probability; or, based on the first attempt record and the second attempt record, the configuration with the number of attempts lower than a preset threshold is selected as the target equipment configuration with a second probability; the sum of the first probability and the second probability is 1.

[0090] Specifically, when determining the target equipment configuration based on the first and second trial records, this can be achieved in two ways: selecting the current highest-scoring configuration with a first probability, or selecting a configuration with fewer attempts than a preset threshold with a second probability, where the sum of the first and second probabilities is 1. For example, during the next optimization, a 50% probability can be set to select the current highest-scoring configuration, while a 50% probability can be set to explore configurations with fewer attempts. This approach allows for balanced exploration of more equipment combinations during the iterative process, effectively preventing the algorithm from falling into local optima.

[0091] Taking the equipment configuration optimization of a certain game character as an example, the system knows from the first attempt record that the combination of "Light Cone A + Relic Set B" has been tried 30 times and currently has the highest score. From the second attempt record, it is found that the combination of "Light Cone C + Relic Set D" has only been tried 5 times (lower than the preset threshold of 10 times). At this time, if the first probability is set to 60% and the second probability is set to 40% (the sum of the two is 1), there is a 60% probability of choosing the high-scoring configuration of "Light Cone A + Relic Set B", and a 40% probability of choosing the low-attempt configuration of "Light Cone C + Relic Set D". This can not only maintain the stability of the current optimal solution, but also reserve exploration space for potential high-quality combinations.

[0092] It should be noted that this embodiment may also be an improvement based on the first embodiment and / or the second embodiment.

[0093] It is not difficult to find that in the embodiment of the present application, by combining the first attempt record and the second attempt record, the configuration with the highest current score and the configuration with the number of attempts below the threshold are selected with different probabilities. This can not only focus on high-quality configurations verified by practice with a high probability, ensuring the basic performance and reliability of the target equipment configuration, but also explore configurations that have not been fully tried with a certain probability, avoiding the limitations of local optimal solutions. This strategy of taking into account both utilization and exploration can not only reduce invalid trial and error, but also retain the possibility of discovering potentially better configurations. Thus, while improving the efficiency of equipment configuration screening, it enhances the comprehensiveness and applicability of the target equipment configuration and provides a more scientific equipment foundation for subsequent balance testing.

[0094] Fifth embodiment

[0095] The fifth embodiment of this application relates to a method for testing the balance of a game lineup. The fifth embodiment is an improvement on the first embodiment. Specifically, this embodiment provides a method for performing a balance test by receiving a specified solution through a configuration selection interface after determining the target equipment configuration, combining the solution generated by a preset algorithm with the final target configuration.

[0096] In some embodiments, after determining the target equipment configuration based on the virtual equipment of the game character and a preset algorithm, the method may further include:

[0097] Step S301, receiving a specified equipment configuration plan according to a configuration selection interface;

[0098] Step S302, determining a target configuration according to the equipment configuration plan and the configuration generated by the preset algorithm;

[0099] Step S303: determining the target equipment configuration according to the target configuration to perform the balance test.

[0100] Specifically, after determining the target equipment configuration based on the game character's virtual equipment and the preset algorithm, the configuration determination process can be optimized by first receiving a specified equipment configuration scheme through the configuration selection interface, then combining this scheme with the configuration generated by the preset algorithm to jointly determine the target configuration, and finally determining the final target equipment configuration based on this target configuration, and then performing balance testing. This process incorporates both externally specified configuration schemes and algorithm-generated configuration results, making the target equipment configuration more comprehensive and reasonable, providing a more accurate foundation for subsequent balance testing.

[0101] For example, suppose the system has determined a target equipment configuration based on the game characters' virtual equipment and a preset algorithm: "Game character A with equipment A, game character B with equipment B." Then, through the configuration selection interface, it receives a planner's specified solution: "Game character A with equipment C, game character B with equipment D." The system can combine this specified solution with the algorithm-generated configuration for analysis. For example, retaining the algorithmic configuration of "Game character A with equipment A" while adopting the specified solution of "Game character B with equipment D" creates a target configuration of "Game character A + equipment A, game character B + equipment D." This determines the target equipment configuration and conducts balance testing, balancing the algorithmic optimization results with the manually specified configuration requirements.

[0102] It should be noted that this embodiment may also be an improvement based on any one or more of the second to fourth embodiments.

[0103] It is not difficult to find that in the embodiments of this application, by adding a step of receiving a specified equipment configuration scheme and determining the target configuration based on the algorithm configuration after the preset algorithm generates the equipment configuration, it can not only leverage the algorithm's advantage in efficiently screening high-quality configurations from massive amounts of data, but also incorporate the experience and targeted needs of manually specified schemes, thus achieving the synergy between machine intelligence and artificial intelligence. This approach can avoid the possible limitations of the algorithm and reduce the blindness and inefficiency of purely manual configuration, making the final target equipment configuration more in line with actual test scenarios and planning expectations, thereby improving the accuracy and practicality of balance testing.

[0104] Sixth embodiment

[0105] The sixth embodiment of this application relates to a method for testing the balance of a game lineup. The sixth embodiment is an improvement on the first embodiment. The specific improvement is that this embodiment provides a specific implementation method for performing a balance test based on the target game character combination and the target equipment configuration.

[0106] In some embodiments, the balance test is performed according to the target game character combination and the target equipment configuration, that is, step S103 may include:

[0107] Step S1031, determining the number of parallel test groups according to a predefined population size parameter;

[0108] Step S1032: inputting each parallel test group formed by the target game character combination and the target equipment configuration into a game simulator, and driving each test group to execute a preset level challenge;

[0109] Step S1033, by controlling the action strategies of the game characters in each test group, outputting the corresponding lineup test scores, so as to implement a balance assessment of the target game character combination and target equipment configuration based on the lineup test scores.

[0110] Specifically, when conducting a balance test based on the target game character combination and target equipment configuration, the number of parallel test groups can be determined first based on the predefined population size parameters; then, each parallel test group formed by the target game character combination and target equipment configuration is input into the game simulator, and each test group is driven to execute the preset level challenge. Then, the action strategy of the game characters in each test group is controlled by the algorithm to replace manual operation and avoid the time-consuming and labor-intensive problems of manual operation, and the corresponding lineup test score is output. Based on the lineup test score, the balance assessment of the target game character combination and target equipment configuration is completed.

[0111] For example, assuming a predefined population size parameter of 80, 80 parallel test groups are identified. Each test group consists of a specific target character combination (e.g., "Character A + Character B + Character C + Character D") and a corresponding target equipment configuration (e.g., Character A with Light Cone X and Relic Set Y, Character B with Light Cone M and Relic Set N, etc.). After entering these 80 test groups into the game simulator, each group automatically performs a pre-set challenge, "Abyss Dungeon Level 10." During this process, the system uses an algorithm to control the action strategies of the characters within each group (e.g., Character A prioritizes group attack skills, Character D focuses on healing teammates, etc.). After the battle, each group's lineup score is output (covering metrics such as clearance time, remaining health, and skill linking efficiency). These scores are ultimately used to evaluate the balance of different character combinations and equipment configurations. For example, combinations scoring between 80 and 90 are considered highly balanced, while combinations scoring below 60 or well above 95 require further adjustment and optimization.

[0112] Furthermore, by controlling the action strategies of the game characters in each test group, the lineup test scores output can be used as a criterion for judging previously generated team lineups and equipment configurations. For example, based on these lineup test scores, the algorithm can be used to further optimize and iterate, for example, stopping after 100 iterations. This can then be used to output results containing basic table information as well as various bar charts and graphs. These results can be provided to planners for further balance analysis and corresponding numerical adjustments.

[0113] Optionally, in some embodiments, the action strategy of controlling the game characters in each test group includes any of the following methods:

[0114] In step S1033A, the optimal action strategy is dynamically learned through a reinforcement learning algorithm to control the actions of the game characters in each test group. Alternatively, in step S1033B, a predefined strategy is executed through the game's built-in behavior tree to control the actions of the game characters in each test group. The output results of the reinforcement learning algorithm and the behavior tree strategy are both used to calculate the lineup test score.

[0115] Specifically, this embodiment supports two methods of controlling the actions of game characters, and the output results of both methods are used to calculate the lineup test score.

[0116] The first approach uses a reinforcement learning algorithm to dynamically learn the optimal action strategy to control the actions of the game characters in each test group. This approach uses reinforcement learning algorithms to automatically learn the team's operating strategy. The advantage is that it can fully explore the upper limit of the team's configuration, but the training process is generally slow.

[0117] Another approach is to use predefined strategies to control the actions of the game characters in each test group through built-in behavior trees. This allows the team's capabilities to be evaluated directly using the game's built-in behavior trees. Developer-written behavior trees are more efficient, but they have certain limitations, and strategies often perform worse than those trained using reinforcement learning.

[0118] Optionally, in some embodiments, the step S1033A may include dynamically learning the optimal action strategy through a reinforcement learning algorithm to control the actions of the game characters in each test group:

[0119] Step S1033A1, determining a matching reinforcement learning model based on the game characters and virtual equipment configurations of the player's current lineup;

[0120] Step S1033A2, when automatic combat is enabled, the character's actions are controlled according to the decision sequence generated by the reinforcement learning model to simulate the player's actual combat scenario, providing a reference basis close to real player behavior for determining the target equipment configuration and balance testing.

[0121] Specifically, while dynamically learning the optimal action strategy to control the actions of the game characters in each test group through a reinforcement learning algorithm, a matching reinforcement learning model can be determined based on the player's current lineup of game characters and virtual equipment configurations. During this process, the system can automatically choose whether to use the game character equipment configuration provided by the planner or conduct subsequent searches based on the algorithm's own iterative results. This not only further increases the diversity of results, but also more efficiently focuses on and showcases the core factors that planners care about, providing assistance in matching the appropriate reinforcement learning model.

[0122] Furthermore, when auto-combat is enabled, character actions can be controlled according to the decision sequence generated by the established reinforcement learning model, thereby simulating actual player combat scenarios and providing a reference basis for determining target equipment configurations and balance testing that is close to real player behavior. This can address the technical issue of "weak performance of behavior tree combat logic (i.e., auto-combat functionality)" reported by players. The reinforcement learning model serves as a stronger strategic solution. By deploying the trained reinforcement learning model, players can be provided with more intelligent auto-combat options. At the same time, the system's search results, after feedback from planners, can also provide support for the character development and team formation suggestions provided by the operator within the game.

[0123] For example, assuming the player's current lineup is "Game Character X (Equipment Set A) + Game Character Y (Equipment Set B) + Game Character Z (Equipment Set C)", the system can first match the corresponding reinforcement learning model from the model library based on this specific game character and equipment configuration. The model has been trained for similar combinations. When automatic combat is enabled, the model can generate a specific decision sequence, such as "At the beginning of the battle, game character X releases a group buff skill → game character Y connects with an AOE damage skill → game character Z focuses single-target fire on the enemy leader → cycle skills until the end of the battle." By controlling the character actions through such decisions, it accurately simulates the strategic logic that players may adopt in actual combat. The generated combat data can not only provide a reference for optimizing the target equipment configuration, but also make the balance test results more closely reflect the behavior of real players.

[0124] Optionally, in some embodiments, the method for determining the lineup test score may include: calculating based on at least one indicator of clearance time, team survival index, and damage output efficiency; wherein, the shorter the clearance time, the higher the team survival rate, and the higher the damage output efficiency, the higher the corresponding lineup test score.

[0125] Specifically, the scoring rules for the lineup test score can be as follows: the faster the clearance speed (i.e., the shorter the clearance time), the fewer team injuries (i.e., the higher the team survival index), and the higher the output damage (i.e., the higher the damage output efficiency), the higher the corresponding lineup test score.

[0126] Taking a preset level challenge as an example, the calculation of the lineup test score will refer to at least one of the clearance time, team survival index, and damage output efficiency. For example, test group A clears the level in 3 minutes, the team has 80% health remaining, and the total damage output is 500,000; test group B clears the level in 5 minutes, the team has 50% health remaining, and the total damage output is 300,000. According to the scoring rules, since test group A has a shorter clearance time, a higher team survival rate, and a higher damage output efficiency, its lineup test score will be higher than test group B. If another test group C has the same clearance time as A (3 minutes), but the team only has 30% health remaining, its score will be lower than A. In this way, the performance gap between different lineups is quantified through the difference in specific indicators.

[0127] It should be noted that this embodiment may also be an improvement based on any one or more of the second to fourth embodiments.

[0128] It is not difficult to find that in the embodiment of the present application, by determining the number of parallel test groups based on the predefined population size, multiple groups of test content formed by the target game character combination and equipment configuration are input into the game simulator to execute the preset level challenge, and the lineup test score is output by controlling the character action strategy, it is possible to achieve simultaneous verification of multiple groups of configurations under the same test conditions, greatly improving the test efficiency; at the same time, balance assessment based on standardized scores can not only ensure the objectivity and comparability of the test results, but also accurately capture the possible imbalance problems in the lineup combination and equipment configuration through comparative analysis of multiple groups of parallel data, thereby providing a comprehensive and reliable basis for the balance optimization of the game lineup.

[0129] Seventh embodiment

[0130] The seventh embodiment of this application relates to a method for testing the balance of a game lineup. The seventh embodiment is an improvement on the first embodiment. The specific improvement is that: in this embodiment, a method is provided for determining the test environment update cycle based on the game version update rhythm, determining the initial search configuration based on historically strong combinations, and completing the test iteration within a preset period to output a version-synchronized lineup test report, thereby achieving a specific implementation method for balance assessment synchronized with version updates.

[0131] Optionally, in some embodiments, the method further comprises:

[0132] Step S401, determining the update cycle of the test environment according to the game version update rhythm;

[0133] Step S402, determining an initial search configuration based on historically strong combinations in the high-quality lineup resource library within the update period;

[0134] Step S403, completing the balance test iteration based on the target game character combination and target equipment configuration within a preset period, and outputting a version-synchronized lineup test report based on the test results to achieve a balance evaluation synchronized with the game version update.

[0135] Specifically, the update cycle of the test environment can be determined according to the rhythm of game version updates. During this process, the team continuously balances the relationship between game version updates and algorithm iteration time. For example, in order to follow up on the weekly version updates of the game, the iteration cycle is proactively compressed to weekly; and then the initial search configuration is determined based on the historical strong combinations in the high-quality lineup resource library within the update cycle, so as to enhance the targetedness of the initial search when the complexity of the combination problem continues to increase while the algorithm iteration time does not increase; and complete the balance test iteration based on the target game character combination and target equipment configuration within the preset cycle, and output the version synchronized lineup test report based on the test results. By iterating based on the latest game simulator every week, more timely and valuable search feedback is provided to the planners, and a balance evaluation synchronized with the game version update is achieved.

[0136] For example, consider an RPG game with a monthly update: first, based on the monthly update rhythm, the test environment update cycle can be set to 30 days. Then, the "Warrior + Priest + Mage" combination and corresponding equipment configuration that performed well in the previous update cycle are retrieved from the high-quality lineup resource library as the initial search configuration. Within the preset 30-day period, multiple rounds of balance testing iterations based on the target game character combination and equipment configuration are completed for the newly added game characters and equipment in the new version. This allows for a test report to be output before the version update, including analysis of the current version's strong lineups, numerical balance risk points, and other content, to ensure that the balance assessment is completed synchronously with the game version update.

[0137] It should be noted that this embodiment may also be an improvement based on any one or more of the second to fifth embodiments.

[0138] It is not difficult to find that in the embodiment of the present application, by setting the test environment update cycle according to the game version update rhythm, the test work can be synchronized with the version iteration rhythm to avoid the test lagging behind the version change; the initial search configuration is determined based on the historical strong combination, and the experience basis of the existing high-quality lineup can be used to reduce the cost of repeated exploration and improve the efficiency of test iteration; the test iteration is completed within the preset period and the version synchronization report is output, which can not only ensure the timeliness of the balance evaluation and version update, but also allow the planner to obtain the balance status of the current version of the lineup in time, so that the value can be adjusted in a targeted manner when the version is updated, effectively preventing the balance problems caused by version iteration, and enhancing the stability of the game version update and the consistency of the player experience.

[0139] The step division of the above various methods is only for the purpose of clear description. During implementation, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application; adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the scope of protection of this application.

[0140] In addition, some embodiments of the present application further provide an electronic device. The electronic device may be various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device may also be various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices.

[0141] The electronic device includes: one or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, enable the processor to perform the steps of the method provided in any one or more of the above embodiments. Figure 2 An exemplary structural diagram of the electronic device is disclosed. The electronic device includes: one or more processors 1101, a memory 1102, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations. Among them, the components shown in this article, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0142] The electronic device may further include an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103 and the output device 1104 may be connected via a bus or other means, with the bus connection being used as an example in the figure.

[0143] The input device 1103 can receive input digital or character information and generate key signal input related to user settings and function control of the electronic device, such as input devices such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, an indicator stick, one or more mouse buttons, a trackball, and a joystick. The output device 1104 may include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The display device may include, but is not limited to, a liquid crystal display, a light emitting diode display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0144] To provide interaction with a user, the electronic device may be a computer. The computer may include a display device (e.g., a cathode ray tube or LCD monitor) for displaying information to the user, and a keyboard and pointing device (e.g., a mouse) through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with the user. For example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback), and input from the user may be received in any form (e.g., voice input or tactile input).

[0145] In the embodiments of the present application, a computer program / instruction is stored on a computer-readable medium. When executed by a processor, the computer program / instruction implements the steps of the method provided in any one or more of the above embodiments. The computer-readable medium may be included in the electronic device described in the above embodiments, or it may exist independently and not be incorporated into the device. The computer-readable medium carries one or more computer-readable instructions.

[0146] The memory 1102 can be used as a non-transitory computer-readable storage medium to store non-transitory software programs, non-transitory computer executable programs, and modules. The processor 1101 executes the non-transitory software programs, instructions, and modules stored in the memory 1102 to execute various functional applications and data processing of the server, thereby implementing the program instructions / modules corresponding to the method provided in any one or more of the above embodiments of the present application.

[0147] The memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device, etc. In addition, the memory 1102 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 1102 may optionally include a memory remotely located relative to the processor 1101, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0148] It should be noted that the computer-readable medium described in this application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above. Computer-readable media may be, for example, but not limited to: electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0149] Computer-readable media includes both permanent and non-permanent, removable and non-removable media, and can be implemented using any method or technology for information storage. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory, static random access memory, dynamic random access memory, other types of random access memory, read-only memory, electrically erasable programmable read-only memory, flash memory or other memory technology, compact discs, digital versatile discs or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0150] Computer program code for performing the operations of the present application can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as C or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network or a wide area network, or can be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0151] In the above-described embodiment, can realize wholly or in part by software, hardware, firmware or its arbitrary combination.For example, can adopt application-specific integrated circuit, general-purpose computer or any other similar hardware device to realize.In certain embodiments, the software program of the present application can be carried out to realize above steps or function by processor.Similarly, the software program of the present application (comprising relevant data structure) can be stored in computer-readable recording medium, for example, RAM memory, magnetic or optical drive or floppy disk and similar device.In addition, some steps or functions of the present application can adopt hardware to realize, for example, as the circuit that cooperates with processor to perform each step or function.

[0152] The computer program product provided by the embodiment of the present application includes one or more computer programs / instructions, and when the computer program / instructions are executed by the processor, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instruction can be stored in a computer-readable storage medium, or transmitted from a computer-readable storage medium to another computer-readable storage medium. For example, the computer instruction can be transmitted from a website, a computer, a server or a data center by wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.) mode to another website, a computer, a server or a data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, a data center that includes one or more available media integrations. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state hard disk) etc.

[0153] The flowcharts or block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the devices, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-specific system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0154] The scope of this application is defined by the appended claims rather than the foregoing description and is therefore intended to encompass within this application all changes that come within the meaning and range of equivalents of the claims. Any reference signs in the claims should not be construed as limiting the claims to which they relate. In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim may also be implemented by one unit or device through software or hardware. Words such as "first" and "second" are only used to distinguish the description and do not indicate any particular order, nor should they be understood as indicating or implying relative importance.

[0155] The above descriptions are merely specific embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art may easily propose variations or substitutions within the technical scope disclosed in the present application, and such variations or substitutions shall be encompassed within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the scope of protection of the claims, and the above descriptions shall be regarded as exemplary and non-limiting.

Claims

1. A method for testing the balance of a game lineup, characterized in that: The method comprises: In response to the target instruction, determining a target game character combination; Determining a target equipment configuration based on the game character's virtual equipment and a preset algorithm; the preset algorithm is used to record historical equipment configuration attempts and determine an equipment configuration selection strategy based on historical performance to assist in determining the target equipment configuration; A balance test is performed based on the target game character combination and the target equipment configuration.

2. The method according to claim 1, characterized in that Determining the target game character combination in response to the target instruction includes: In response to the target instruction, determining an initial team lineup; A target game character combination is determined based on the genetic algorithm and the game characters in the initial team lineup.

3. The method according to claim 1, characterized in that The method for determining the initial team lineup includes: Determining the number of parallel test groups according to a predefined population size parameter; wherein the parallel test groups are used to execute balance tests in parallel in the game simulator; Determining an initial configuration of each test group based on the number of the parallel test groups; the initial configuration includes at least a game character and virtual equipment; Determine the initial team lineup based on the initial configuration of all test groups.

4. The method according to claim 3, characterized in that Determining the initial configuration of each test group includes: Determine the game characters and virtual equipment of the test group according to a random selection scheme; or, The game characters and virtual equipment of the test group are determined according to the preconfigured weight parameters.

5. The method according to claim 2, characterized in that Determining the target game character combination based on the genetic algorithm and the game characters in the initial team lineup includes: According to predefined genetic algorithm parameters, the initial team lineup enters an iterative phase; In the iterative stage, the game character combination in the initial team lineup is optimized according to the crossover operation and mutation operation of the genetic algorithm to determine the target game character combination.

6. The method according to claim 5, characterized in that For the mutation operation, the method further includes: Determine a high-quality lineup resource library based on historical balance test results; the high-quality lineup resource library is used to store high-scoring lineup combinations of game characters; In subsequent iterations, a historically strong combination is selected from the high-quality lineup resource library as a benchmark lineup based on a preset probability; Using the benchmark lineup as a reference, mutation operations are performed to generate derivative combinations to improve the pertinence and efficiency of game character combination optimization.

7. The method according to claim 1, characterized in that Determining the target equipment configuration according to the virtual equipment of the game character and a preset algorithm includes: Determine a first attempt record for each type of virtual equipment and a second attempt record for attribute configuration corresponding to each game character; the first attempt record is used to dynamically track the number of attempts for each type of virtual equipment configuration during an iterative process; the second attempt record is used to dynamically track the number of attempts for attribute configuration during an iterative process; A target equipment configuration is determined according to the first attempt record and the second attempt record.

8. The method according to claim 7, characterized in that The first attempt record of the virtual equipment includes an attempt record of the first type of virtual equipment and / or an attempt record of the second type of virtual equipment; The first type of virtual equipment and the second type of virtual equipment are used to represent different types divided according to the game equipment system.

9. The method according to claim 8, characterized in that The first attempt record of determining various types of virtual equipment corresponding to each game character and the second attempt record of attribute configuration include: Determine a target hierarchical architecture; wherein the first layer of the target hierarchical architecture is a game character combination optimization layer based on a genetic algorithm; the second layer is an optimization layer for the first type of virtual equipment; and the third layer is an optimization layer for the second type of virtual equipment and attribute configuration; The decision variables are separated according to the alternating optimization strategy, the number of attempts of the first type of virtual equipment is determined through the second layer, and the number of attempts of the second type of virtual equipment and attribute configuration is determined through the third layer to form the first attempt record and the second attempt record.

10. The method according to claim 9, characterized in that The step of separating the decision variables according to the alternating optimization strategy, determining the number of attempts for the first type of virtual equipment through the second layer, and determining the number of attempts for the second type of virtual equipment and attribute configuration through the third layer to form the first attempt record and the second attempt record includes: During the second-layer optimization process, the second-class virtual equipment and attribute configuration of the third layer are kept at the current optimal state, the first-class virtual equipment is optimized and the number of attempts is recorded; During the third-level optimization process, the first-class virtual equipment of the second-level is kept in the current optimal state, and the second-class virtual equipment and attribute configuration are optimized and the number of attempts is recorded; The decision variables are separated by the above-mentioned alternating optimization method to form a first attempt record and a second attempt record respectively.

11. The method according to claim 7, characterized in that Determining the target equipment configuration according to the first attempt record and the second attempt record includes: According to the first attempt record and the second attempt record, selecting the configuration with the highest current score as the target equipment configuration with a first probability; or, selecting, according to the first attempt record and the second attempt record, a configuration with a number of attempts lower than a preset threshold as a target equipment configuration with a second probability; The sum of the first probability and the second probability is 1.

12. The method according to claim 1, characterized in that After determining the target equipment configuration according to the virtual equipment of the game character and the preset algorithm, the method further includes: According to the configuration selection interface, receive the specified equipment configuration plan; Determining a target configuration based on the equipment configuration plan and the configuration generated by the preset algorithm; The target equipment configuration is determined according to the target configuration to perform the balance test.

13. The method according to any one of claims 1 to 12, characterized in that The performing of a balance test according to the target game character combination and the target equipment configuration includes: Based on the determined number of parallel test groups, each parallel test group formed by the target game character combination and the target equipment configuration is input into the game simulator, and each test group is driven to execute a preset level challenge; By controlling the action strategies of the game characters in each test group, the corresponding lineup test scores are output, so as to achieve a balance assessment of the target game character combination and target equipment configuration based on the lineup test scores.

14. The method according to claim 13, characterized in that The action strategy for controlling the game characters in each test group includes any of the following methods: Dynamically learn the optimal action strategy through reinforcement learning algorithms to control the actions of game characters in each test group; or Execute predefined strategies through the game's built-in behavior tree to control the actions of the game characters in each test group; The output results of the reinforcement learning algorithm and the behavior tree strategy are both used to calculate the lineup test score.

15. The method according to claim 14, characterized in that The method of dynamically learning the optimal action strategy through the reinforcement learning algorithm to control the actions of the game characters in each test group includes: Determine the matching reinforcement learning model based on the player's current lineup of game characters and virtual equipment configurations; When automatic combat is enabled, the character's actions are controlled according to the decision sequence generated by the reinforcement learning model to simulate the player's actual combat scenario, providing a reference basis for determining the target equipment configuration and balance testing that is close to real player behavior.

16. The method according to claim 13, characterized in that The method for determining the lineup test score includes: Calculation is based on at least one of the following indicators: clearance time, team survival index, and damage output efficiency; Among them, the shorter the clearance time, the higher the team survival rate, and the higher the damage output efficiency, the higher the corresponding lineup test score.

17. The method according to claim 13, wherein The method further comprises: Determine the update cycle of the test environment based on the game version update rhythm; Determine the initial search configuration based on the historically strong combinations in the high-quality lineup resource library within the update cycle; Complete the balance test iteration based on the target game character combination and target equipment configuration within the preset period, and output the version-synchronized lineup test report based on the test results to achieve balance evaluation synchronized with the game version update.

18. An electronic device, characterized in that: The electronic device comprises: one or more processors; and A memory storing computer program instructions, which, when executed, cause the processor to perform the steps of the method according to any one of claims 1 to 17.

19. A computer readable medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 17 are implemented.

20. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 17 are implemented.

Images