Method, device and system for equipment collaborative scheduling and storage medium

Abstract

The invention relates to the technical field of intelligent equipment, and discloses a method, device and system for equipment collaborative scheduling and a storage medium. The method comprises the following steps: acquiring current environment information corresponding to a current scene and current equipment state information corresponding to two or more scene equipment in the current scene; based on an inference engine, an optimal scheduling strategy matched with the current equipment state information and the current environment information is determined from a scheduling strategy candidate set, and each strategy in the scheduling strategy candidate set is obtained through a multi-objective optimization algorithm according to two or more optimization objectives of equipment cooperative operation; and performing cooperative scheduling of two or more scene devices according to the optimal scheduling strategy. Therefore, global search is carried out in multiple target spaces through a multi-target optimization algorithm, an optimal equilibrium solution among multiple targets is found, and rapid real-time decision making is realized based on an inference engine.

Description

Technical Field

[0001] This application relates to the field of smart device technology, such as methods, apparatus, systems, and storage media for device collaborative scheduling. Background Technology

[0002] With the widespread application of smart devices in communities and homes, device collaborative scheduling has become increasingly important. In practical applications, multiple objectives often need to be considered simultaneously, such as security performance, energy consumption, and device lifespan. Currently, device collaborative scheduling can be based on rule engines. For example, in security scenarios, it can be configured to immediately activate two or more cameras for monitoring when an anomaly is detected. While this approach can meet security requirements to some extent, it doesn't consider issues like energy consumption and device lifespan. Moreover, these rules are usually set in isolation, failing to comprehensively consider the interactions between multiple objectives, making them prone to getting stuck in local optima and unable to fully balance the various objectives. For instance, to ensure a fast security response, cameras might operate at high power consumption for extended periods, leading to excessive energy consumption. Furthermore, in actual operation, rapid decision-making for device collaborative scheduling is required to meet real-time requirements, but current decision-making methods struggle to generate effective strategies that balance multiple objectives in a short time. Therefore, the device collaborative scheduling process needs further optimization.

[0003] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments, but rather as a prelude to the detailed description that follows.

[0005] This disclosure provides a method, apparatus, system, and storage medium for collaborative device scheduling, in order to address the technical problem of needing optimization in scenario-based collaborative device scheduling.

[0006] In some embodiments, the method includes: Obtain the current environment information corresponding to the current scene, as well as the current device status information corresponding to two or more scene devices in the current scene; Based on the inference engine, the best scheduling strategy that matches the current device state information and the current environment information is determined from the scheduling strategy candidate set. Each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives for device cooperative operation. Based on the optimal scheduling strategy, coordinate the scheduling of two or more scenario devices.

[0007] In this way, in the scenario of collaborative device scheduling, the optimal balance solution among multiple objectives is found by performing a global search in multiple objective spaces. This comprehensively considers multiple objectives such as security response and energy consumption control, and generates an overall optimal scheduling strategy. Furthermore, based on the inference engine, the computational load and memory usage are significantly reduced and the inference speed is improved by optimizing the inference model. The best scheduling strategy can be selected from the candidate set of scheduling strategies in a short time, enabling rapid real-time decision-making and meeting the real-time requirements of collaborative device scheduling.

[0008] In some embodiments, it also includes: Based on the scenario requirements, the corresponding environmental information, and the device status information of each scenario device, determine two or more optimization objectives for the collaborative operation of two or more scenario devices. Based on the third-generation non-dominated sorting genetic algorithm (NSGA-III), a set of scheduling strategy candidates corresponding to two or more optimization objectives is obtained and stored. Each strategy in the set of scheduling strategy candidates is a device cooperative scheduling scheme that achieves a balance between two or more optimization objectives.

[0009] It is evident that by comprehensively considering scenario requirements, equipment status, and environmental information to determine optimization objectives, and by using the NSGA-III algorithm to generate solution sets, the diversity of solution sets and the balance of different optimization objectives are ensured, providing a solid foundation for subsequent real-time scheduling.

[0010] In some embodiments, obtaining and storing the scheduling policy candidate set corresponding to two or more optimization objectives includes: Determine the initial population, where each individual in the initial population represents a cooperative scheduling strategy, which includes two or more decision variables; Calculate the performance value of each individual in the initial population on two or more optimization objectives; Based on the performance values, the initial population is iteratively evolved through non-dominated sorting, selection, and improved chromosome crossover and mutation operations to obtain a candidate set of scheduling strategies.

[0011] By initializing the population, calculating performance values, and evolving the population based on genetic operations such as non-dominated sorting, the quality of the solution set and the optimization effect are guaranteed. This enables the generated scheduling strategy to effectively cover the balance relationship between multiple optimization objectives, thereby enhancing the practicality and reliability of the scheduling strategy candidate set.

[0012] In some embodiments, obtaining and storing the scheduling policy candidate set corresponding to two or more optimization objectives further includes: The candidate set of scheduling strategies is stored using a hash table or database.

[0013] Using hash tables or databases to store solution sets improves the management and retrieval efficiency of solution sets, ensures that solution set data can be quickly obtained when real-time scheduling is required, and supports stable operation and rapid response of collaborative device operation.

[0014] In some embodiments, determining the optimal scheduling strategy that matches the current device state information and the current environment information includes: The current device status information, current environment information, and the scheduling policy candidate set are input into the lightweight inference engine. Through forward propagation, the fitness value of each policy in the scheduling policy candidate set corresponding to the current device status information and current environment information is determined. The strategy corresponding to the maximum fitness value is determined as the optimal scheduling strategy.

[0015] The specific implementation method of using a lightweight inference engine to determine the optimal scheduling strategy was clarified. By calculating the fitness value of each strategy through forward propagation and selecting the optimal strategy, the scientificity and accuracy of the scheduling decision-making process were ensured, and the system's rapid decision-making capability and adaptability in real-time scheduling were further enhanced.

[0016] In some embodiments, determining the fitness value corresponding to each policy in the candidate scheduling policy set and the current device state information and current environment information includes: Based on the current device status information and the current environment information, and using a lightweight inference engine, each policy (p) in the candidate scheduling policy set is determined. i The performance value f1(p) on each optimization objective i f2(p) i ), ..., f m (p i ); According to formula (1), the fitness of each strategy is obtained; F i =w1×f1(p i )+ w2×f2(p i )+…+ w m ×f m (p i (1) Among them, w1, w2, ..., w m The weights corresponding to each optimization objective.

[0017] By combining current device status and environmental information, the performance value of each strategy on each optimization objective is calculated, and the fitness value is obtained according to the weighted summation formula. This provides a quantitative basis for the lightweight inference engine to accurately evaluate and select the best scheduling strategy, ensuring the accuracy and rationality of scheduling decisions.

[0018] In some embodiments, after performing collaborative scheduling of two or more scene devices, the method further includes: Based on the obtained operating status information and target achievement information of the scene devices, adjust and optimize the targets or update the candidate set of scheduling strategies; among them, adjusting the optimization targets includes adjusting the weight of each optimization target based on the operating status information and target achievement information.

[0019] In this way, the system can adjust and optimize objectives or update the solution set based on the equipment's operating status and the results of achieving the objectives, enabling the system to have adaptive capabilities. It can dynamically optimize the scheduling strategy based on the actual operating effect, continuously improve the overall benefits of equipment collaborative scheduling, and ensure that the collaborative operation of equipment can continuously meet optimization needs and adapt to changes in the environment and requirements in the long-term operation.

[0020] In some embodiments, the apparatus includes a processor and a memory storing program instructions, wherein the processor is configured to perform the above-described method for device cooperative scheduling when executing the program instructions.

[0021] In some embodiments, the system includes: a decision-making device, and two or more scene devices; wherein, The decision-making device is configured to acquire the current environment information corresponding to the current scene, and the current device status information corresponding to two or more scene devices in the current scene; based on the inference engine, it determines the best scheduling strategy that matches the current device status information and the current environment information from the scheduling strategy candidate set, wherein each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives of device collaborative operation; and performs collaborative scheduling of two or more scene devices according to the best scheduling strategy.

[0022] In some embodiments, the storage medium stores program instructions that, when executed, perform the above-described method for device collaborative scheduling.

[0023] The method, system, and storage medium for device collaborative scheduling provided in this disclosure can achieve the following technical effects: By employing a multi-objective optimization algorithm to simulate the biological evolution process and perform a global search, a set of strategy solutions that achieve optimal balance among multiple objectives is found, significantly improving the overall efficiency of equipment collaborative scheduling. Furthermore, the inference engine, combined with optimization techniques, enables rapid collaborative scheduling decisions, greatly enhancing the response speed and practicality of equipment collaborative operation.

[0024] The above general description and the description below are exemplary and illustrative only and are not intended to limit this application. Attached Figure Description

[0025] One or more embodiments are illustrated by way of example with reference to the accompanying drawings. These illustrations and drawings do not constitute a limitation on the embodiments. Elements having the same reference numerals in the drawings are shown as similar elements. The drawings are not to be scaled. And wherein: Figure 1 This is a schematic diagram of a device collaborative scheduling system provided in an embodiment of this disclosure; Figure 2 This is a schematic flowchart of a device collaborative scheduling method provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram of a process for obtaining a candidate set of scheduling strategies according to an embodiment of this disclosure; Figure 4 This is a schematic flowchart of a device collaborative scheduling method provided in an embodiment of this disclosure; Figure 5 This is a schematic diagram of a device for collaborative scheduling of equipment provided in an embodiment of this disclosure; Figure 6 This is a schematic diagram of a device for collaborative scheduling of equipment provided in an embodiment of this disclosure; Figure 7 This is a schematic diagram of a device for collaborative scheduling of equipment provided in an embodiment of this disclosure. Detailed Implementation

[0026] To provide a more detailed understanding of the features and technical content of the embodiments of this disclosure, the implementation of the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for illustrative purposes only and are not intended to limit the embodiments of this disclosure. In the following technical description, for ease of explanation, several details are used to provide a full understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be simplified in their depiction to simplify the drawings.

[0027] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this disclosure described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0028] Unless otherwise stated, the term "multiple" means two or more.

[0029] In this embodiment of the disclosure, the character " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B means: A or B.

[0030] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0031] In this embodiment, a multi-objective optimization algorithm, such as the third-generation Non-dominated Sorting Genetic Algorithm III (NSGA-III), is employed to comprehensively consider multiple objectives in device collaborative scheduling. By simulating the process of biological evolution for global search, a series of strategy solutions that achieve optimal balance among multiple objectives can be found, rather than being limited to the optimality of a single objective, significantly improving the overall efficiency of device collaborative scheduling. Real-time performance is crucial in device collaborative scheduling scenarios. The inference engine, combined with TensorRT optimization technology, enables the system to complete the process of selecting the optimal strategy from the strategy solution set in an extremely short time, within milliseconds, making rapid decisions. This ensures that in emergency situations such as sudden security incidents, devices can respond quickly and adopt reasonable collaborative scheduling strategies. Furthermore, the architecture design and feedback mechanism of the multi-objective Pareto optimization decision engine endow the system with high adaptability and flexibility. The multi-objective settings can be dynamically adjusted according to device status, environmental information, and user needs, adjusting the optimization objectives and weights. Based on the feedback of device execution results, the system can continuously adjust its decisions to adapt to various complex and changing real-world scenarios.

[0032] The device collaborative scheduling method based on a multi-objective Pareto optimization decision engine, as disclosed in this embodiment, is not only applicable to smart communities and homes, but also to numerous other fields such as industrial production and intelligent transportation. In industrial production, it can balance multiple objectives such as production efficiency, product quality, and energy consumption; in intelligent transportation, it can consider objectives such as traffic flow optimization, safety assurance, and energy consumption. This versatility helps to promote the improvement of intelligence levels in various fields and facilitates the rational utilization and efficient allocation of resources.

[0033] Figure 1 This is a schematic diagram of a device collaborative scheduling system provided in an embodiment of this disclosure. Figure 1 As shown, the system includes a decision-making device 100 and a scene device 200. The decision-making device 100 can be a standalone device, a scene device with control functions, or a scene device with control functions and a cloud device with computing and storage capabilities. The decision-making device 100 can communicate with each scene device 200.

[0034] Because this system can be applied to numerous scenarios and fields such as smart communities, homes, industrial production, and intelligent transportation, the scene device 200 can be intelligent devices such as surveillance cameras, streetlights, home appliances, and traffic lights. For example, in a home environment, there are multiple home appliances, such as televisions, refrigerators, air conditioners, and access control devices. If the television can communicate with other home appliances, then the television can be the decision-making device 100. If the current scenario is a leisure viewing scenario, then the television can not only be the decision-making device 100 but also the scene device 200. At the same time, home appliances such as air conditioners, lights, and curtains can all be scene devices 200. If the current scenario is a security scenario, then the television is the decision-making device 100, while home appliances such as access control devices, window devices, and speakers can all be scene devices 200.

[0035] In this embodiment of the disclosure, when the current scenario meets the conditions for collaborative operation of devices, the decision-making device 100 can perform collaborative scheduling of two or more scenario devices 200 in the current scenario based on two or more optimization objectives. For example... Figure 2 As shown, the process of equipment collaborative scheduling includes: Step 201: Obtain the current environment information corresponding to the current scene, and the current device status information corresponding to two or more scene devices in the current scene.

[0036] As mentioned above, device collaborative scheduling can be applied to various scenarios, such as: leisure viewing scenarios and security monitoring scenarios in home environments; security monitoring scenarios and pleasant atmosphere scenarios in community environments; and smooth traffic and traffic flow / avoidance scenarios in smart transportation. The decision-making device can communicate with each device in the scenario. Based on received user instructions, acquired environmental information, and device status information, the decision-making device can determine the current scenario and obtain the current environmental information corresponding to the current scenario, as well as the current device status information of two or more devices in the current scenario. For example, in a home environment, if the decision-making device is the television, and the television receives a "enter viewing mode" instruction from the user, it can determine that the current scenario is a leisure viewing scenario. It can then obtain the current temperature, humidity, and lighting information corresponding to the home environment, and the current device status information of the television, speakers, lighting equipment, and curtains. That is, the current environmental information includes: current temperature, current humidity, and current lighting information; while the current device status information includes: current television status, current speaker status, current lighting equipment status, and current curtain status.

[0037] In a community environment, if the decision-making device is the property management central control screen, and if it detects abnormal gatherings of people through cameras or receives a "enter security mode" command from a user, the property management central control screen can determine that the current scenario is a security monitoring scenario. It can then obtain the current temperature, humidity, and light intensity information of the community environment, as well as the current device status information of the image acquisition system, lighting system, and alarm system. In other words, the current environmental information includes: current temperature, humidity, and light intensity; and the current device status information includes: the current status information of each camera, each lighting device, and each alarm device. Step 202: Based on the inference engine, determine the best scheduling strategy that matches the current device state information and the current environment information from the scheduling strategy candidate set. Each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives for device cooperative operation.

[0038] In this embodiment, a multi-objective Pareto optimization decision engine can be used to determine the collaborative operation strategy of the devices. Therefore, before the decision-making device makes a real-time decision, multiple scheduling strategies corresponding to each scenario have been obtained through a multi-objective optimization algorithm. A multi-objective optimization algorithm (MOOA) is a mathematical or heuristic method that searches for a set of non-dominated solutions (Pareto-optimal set) in a decision space containing two or more objectives; its output is a Pareto front that cannot be simultaneously improved by any other solution in the objective space. Various multi-objective optimization algorithms can be used, including mathematical programming algorithms, genetic evolution algorithms, heuristic algorithms, deep learning-based algorithms, and so on.

[0039] In some embodiments, multiple scheduling strategies corresponding to each scenario can be obtained through the NSGA-III algorithm. Specifically, this may include: determining two or more optimization objectives for the collaborative operation of two or more scenario devices based on scenario requirement information, environmental information corresponding to the scenario, and device status information of each scenario device; obtaining and saving a candidate set of scheduling strategies corresponding to the two or more optimization objectives based on the NSGA-III algorithm, wherein each strategy in the candidate set of scheduling strategies is a device collaborative scheduling scheme that achieves a balance between the two or more optimization objectives.

[0040] In some embodiments, determining two or more optimization objectives for the collaborative operation of two or more scenario devices includes translating three types of information—scenario requirements, environmental constraints, and device capabilities—into quantifiable and computable mathematical objective functions. Specifically, based on scenario requirement information, corresponding optimization parameter indicators can be obtained, such as two or more parameters including speed, energy consumption, comfort, and safety. Based on environmental information and the device status information of each scenario device, corresponding software or hardware constraints can be obtained, such as maximum device power consumption, upper limit of ambient light, operating temperature range, etc. Therefore, an objective type can be selected for each optimization indicator, such as minimum value, minimum value, or range, etc., and a corresponding normalization function and weight range can be given, forming two or more objectives f1, f2, ..., f... m .

[0041] For example, a user's voice command "I want to watch a movie" can generate corresponding optimized parameters based on the scenario's needs, including: ① high comfort level ② low energy consumption ③ quick entry into the desired state. Environmental information may include: living room illuminance 800 lx, temperature 28 ℃, humidity 70%, time 20:00. Device status information includes: smart TV: standby power consumption 0.5W, brightness adjustable from 0-100%; surround sound: standby power consumption 0.3 W, volume adjustable from 0-50 levels; smart light strip: current brightness 90%, color temperature 6000 K, adjustable range 10-100%, 2700-6500 K; air conditioner: rated cooling power 800 W, set temperature 16-30 ℃. Therefore, the optimization objectives for the collaborative operation of scene devices are determined as follows: f1 comfort (maximum) = 0.4 × color temperature comfort + 0.3 × volume matching degree + 0.3 × temperature comfort; f2 energy consumption (minimum) = TV power consumption + audio power consumption + light strip power consumption + air conditioner power consumption; f3 response time (minimum) = the number of seconds from the issuance of the command to all devices entering the target state.

[0042] After identifying two or more optimization objectives, a candidate set of scheduling strategies corresponding to the two or more optimization objectives can be obtained and saved based on a multi-objective optimization algorithm. In some embodiments, based on the NSGA-III algorithm, obtaining and saving the candidate set of scheduling strategies corresponding to the two or more optimization objectives includes: determining an initial population, where each individual in the initial population represents a cooperative scheduling strategy, and the cooperative scheduling strategy includes two or more decision variables; calculating the performance value of each individual in the initial population on the two or more optimization objectives; and iteratively evolving the initial population based on the performance values ​​through non-dominated sorting, selection, and improved chromosome crossover and mutation operations to obtain the candidate set of scheduling strategies.

[0043] Among them, N individuals can be randomly generated, each individual x i A corresponding cooperative scheduling strategy is established, which includes two or more decision variables, such as camera startup time, device motion power, device response time, etc. These variables are randomly initialized within a predefined range, thus obtaining the initial population. Two or more optimization objectives have been determined, thus obtaining the f1(x) of the corresponding objective function. i f2(x) i ), ..., f m (x i Each objective function has a specific definition; for example, the security response objective f1(x) i ): Defined as the time from when the camera detects an anomaly to when it starts up. The shorter the time, the higher the performance value; energy consumption control target f2(x) if3(x) is defined as the energy consumption of a device over a certain period of time. Lower energy consumption corresponds to higher performance; of course, other targets also play a role. i ), ..., f m x i ): Define other objectives based on the specific application, such as device lifespan, response latency, etc.

[0044] Thus, based on the initialized population and the corresponding objective function, the performance value of each individual in the population on each objective is determined. For example, if individual x i The camera startup time is 0.8 seconds, then f1(x) i )=0.8, individual x i If the operating power of the device is 10W, then f2(x) i )=10, similarly, calculate the individual's performance values ​​on other objectives based on the individual's decision variables. Thus, for each individual x i Calculate its performance value f1(x) on all targets. i f2(x) i ), ..., f m (x i For example, individual x i The performance values ​​are [0.8, 10, ...]. Therefore, based on performance values, non-dominated sorting and crowding calculation can be performed to select and improve chromosome crossover and mutation operations, iterating to generate the next generation of the population. In other words, the NSGA-III algorithm can effectively evaluate the performance of each individual on multiple objectives, and through non-dominated sorting and crowding calculation, select better individuals to enter the next generation, thereby gradually optimizing the population—that is, performing genetic iteration. When the iteration termination condition is met, the resulting population becomes the candidate set for scheduling strategies. There are several iteration termination conditions, including: reaching a preset maximum number of iterations, or the iteration time reaching a preset time, or the population's performance value changing very little or no longer changing in consecutive iterations, or setting an optimal solution threshold for the objective function, whereby the objective function value of an individual in the population reaches or approaches this threshold.

[0045] In some embodiments, when calculating the performance value of each individual on multiple objectives, a weighted performance value can be obtained, i.e., the performance value of each objective function f1(x). i It will be multiplied by the corresponding weight w i For example, for each individual x i Calculate its performance value on the weighted objective: F(x i =w1×f1(x) i )+ w2×f2(x i )+…+ w m ×f m (xi ) Among them, w1, w2, ..., w m This refers to the weights of each objective. These weights directly affect the overall evaluation of the objective function and can be dynamically adjusted. For example, in a community security scenario, during the daytime when there is more human activity, the weight of the security response objective can be appropriately reduced, while the weight of the energy consumption control objective can be increased; conversely, at night, the weight of the security response objective can be increased, while the weight of the energy consumption control objective can be decreased. This step ensures that the importance of different objectives is reflected in multi-objective optimization.

[0046] After multiple iterations of optimization using the NSGA-III algorithm, a solution set containing 100,000 scheduling strategies can be obtained. Alternatively, there could be 10,000 or even millions of scheduling strategies, meaning the candidate set contains a large number of strategies. To facilitate rapid retrieval by the inference engine when needed, some embodiments of obtaining and saving the candidate set of scheduling strategies corresponding to two or more optimization objectives also include using a hash table or database to store the candidate set.

[0047] There are various multi-objective optimization algorithms. Simulated annealing is a heuristic optimization algorithm. Therefore, in some embodiments, after determining two or more optimization objectives, simulated annealing can be used to obtain multiple scheduling strategies corresponding to the scenario. Specifically, this may include: randomly generating an initial device collaborative scheduling strategy as the current solution, and calculating the objective function value of the current solution under multiple objectives, such as security response and energy consumption control. Then, randomly generating a new solution in the neighborhood of the current solution, and calculating the objective function value of the new solution. If the objective function value of the new solution is better than that of the current solution, the new solution is accepted; otherwise, the new solution is accepted with a certain probability, which gradually decreases as the temperature decreases. The temperature is gradually reduced according to a certain cooling strategy until the termination condition is met. The solution obtained at this point is the approximate optimal solution of the multi-objective optimization, and the corresponding scheduling strategy is stored in the scheduling strategy candidate set. By repeatedly running the algorithm independently, multiple scheduling strategies can be obtained, thus forming the scheduling strategy candidate set.

[0048] Of course, multiple scheduling strategies corresponding to each scenario can also be obtained based on genetic local search algorithms and deep learning algorithms, which will not be listed one by one.

[0049] The decision-making device stores a set of scheduling strategy candidates for each scenario, either locally or in the cloud. Therefore, when making real-time decisions for the current scenario, it only needs to determine the best scheduling strategy that matches the current device state and environment information from the candidate set. The inference engine includes rule-based inference engines, simple machine learning inference engines, lightweight inference engines, etc. The lightweight inference engine, combined with TensorRT optimization, can optimize deep learning models, reduce computation and memory usage, and improve inference speed. Based on the current situation, the lightweight inference engine selects the most suitable scheduling strategy from the candidate set, thus obtaining the best scheduling strategy that matches the current device state and environment information. In some embodiments, determining the optimal scheduling strategy that matches the current device state information and the current environment information includes: inputting the current device state information, the current environment information, and the scheduling strategy candidate set into a lightweight inference engine; determining the fitness value of each strategy in the scheduling strategy candidate set corresponding to the current device state information and the current environment information through forward propagation; and determining the strategy corresponding to the maximum fitness value as the optimal scheduling strategy.

[0050] The lightweight inference engine is based on a deep learning inference framework, combined with TensorRT optimization technology. Given the current device state information and current environment information S, and the candidate scheduling policy set PS corresponding to the current scenario, S is first feature-extracted and transformed into a format (S') suitable for the inference model input. The inference model M takes (S') and PS as input and calculates the fitness value (F) of each scheduling policy under the current device state information and current environment information through forward propagation. i ), where i represents the policy index in the policy solution set. Fitness value (F i This comprehensively considers the degree to which the scheduling strategy meets multiple objectives under the current device status information and the current environment information. For example, for a scheduling strategy (p... i The degree to which security response and energy consumption control objectives are comprehensively met under the current equipment load and ambient lighting conditions is (F). i ).

[0051] TensorRT optimization techniques optimize the inference model M, including model compression, layer fusion, and quantization. Model compression reduces the model size by removing redundant connections and parameters; layer fusion merges multiple adjacent layers into a single computational unit, reducing computational load; quantization converts model parameters and computational data from high-precision to low-precision formats, accelerating computation without sacrificing too much accuracy. Through these optimizations, the inference engine can quickly select the strategy with the highest fitness value as the real-time decision output within 30ms. For example, in a smart community project, current device status information and current environmental information S include the density of personnel activity in the community, ambient light intensity, and current device operating power. The scheduling strategy candidate set PS contains 100,000 strategies. For instance, scheduling strategy (p1) specifies that when personnel activity density is high and light intensity is low, cameras should start quickly and devices should reduce the power of some non-critical functions to control energy consumption. After feature extraction from S, the inference engine inputs it into the inference model M. The inference model optimized by TensorRT quickly calculates the fitness value of each policy. For example, if the fitness value (F1) of the scheduling policy (p1) is high, it indicates that (p1) can better balance the security response and energy consumption control objectives in the current state. Therefore, (p1) is selected as the real-time decision, i.e., the optimal scheduling policy output.

[0052] In some embodiments, determining the fitness value of each policy in the scheduling policy candidate set corresponding to the current device state information and the current environment information includes: determining the fitness value of each policy in the scheduling policy candidate set based on the current device state information and the current environment information, using a lightweight inference engine. i The performance value f1(p) on each optimization objective i f2(p) i ), ..., f m (p i According to formula (1), the fitness of each strategy is obtained. F i =w1×f1(p i )+ w2×f2(p i )+…+ w m ×f m (p i (1) Among them, w1, w2, ..., w m The weights corresponding to each optimization objective. Of course, w1, w2, ..., w... m It can also be dynamically adjusted, that is, dynamically adjusted according to the current device status information and the current environment information.

[0053] As can be seen, by combining a lightweight inference engine with TensorRT optimization technology, the computational load and memory consumption are significantly reduced and the inference speed is improved through the optimization of the inference model. It can select the optimal scheduling strategy from a large set of scheduling strategy candidates in a short time, realize fast real-time decision-making, meet the real-time requirements of device collaborative scheduling, and achieve efficient device collaborative scheduling.

[0054] In some embodiments, a rule-based inference engine can also be used to determine the optimal scheduling strategy that matches the current device state information and the current environment information from the scheduling strategy candidate set. The specific process may include: the rule-based inference engine pre-setting a series of rules; after obtaining the current device state and current environment information, matching and inference are performed according to these rules, and the corresponding scheduling strategy is selected as the optimal scheduling strategy from the scheduling strategy candidate set. For example, in a community security scenario, rules are pre-set according to the scheduling strategy candidate set, including: writing rules and assigning corresponding priorities. The pre-set rules may include: 1. When the illuminance is less than 50 lux and the number of people is greater than 4, the strategy is "high brightness and fast security"; 2. When the illuminance is greater than or equal to 50 lux and the ambient temperature is greater than 28℃, the strategy is "low brightness and wind," etc. Thus, after obtaining the current device state and current environment information, the Rete or Leaps algorithm can be used to quickly find the pre-set rules that meet the conditions, and the highest priority pre-set rule is triggered, with its corresponding strategy determined as the optimal scheduling strategy.

[0055] Of course, a machine learning model, such as a decision tree or neural network, can be trained to obtain an inference engine. This engine, when given the current device state and environmental information, can predict a corresponding optimal scheduling strategy. Therefore, based on multiple inference engines, the optimal scheduling strategy matching the current device state and environmental information can be determined from a candidate set of scheduling strategies. Among these, a lightweight inference engine combined with TensorRT optimization technology significantly improves inference speed by optimizing the inference model, enabling decisions to be made at the millisecond level and achieving real-time decision-making.

[0056] Step 203: Based on the optimal scheduling strategy, perform collaborative scheduling of two or more scene devices.

[0057] Once the decision-making device determines the optimal scheduling strategy, it can coordinate the scheduling of two or more scene devices according to the optimal scheduling strategy and control the operation of the corresponding scene devices.

[0058] The optimal scheduling strategy derived through a lightweight inference engine can be transformed into an executable instruction format and sent to the corresponding scene device. The scene device can then perform the corresponding operation based on the received instruction.

[0059] As can be seen, this embodiment employs a multi-objective optimization algorithm to comprehensively consider multiple objectives in device collaborative scheduling. By simulating the process of biological evolution for global search, it can find a series of strategy solutions that achieve optimal balance among multiple objectives, rather than being limited to the optimal solution for a single objective, significantly improving the overall efficiency of device collaborative scheduling. Specifically, the lightweight inference engine combined with TensorRT optimization technology enables the system to complete the process of selecting the optimal strategy from the strategy solution set in an extremely short time, within milliseconds, making rapid decisions. This ensures that in emergency situations such as sudden security incidents, the devices can respond quickly and adopt appropriate collaborative scheduling strategies.

[0060] Of course, in this embodiment of the disclosure, after coordinating the scheduling of two or more scene devices, the method further includes: adjusting the optimization target or updating the candidate set of scheduling strategies based on the obtained scene device operating status information and target achievement result information; wherein, adjusting the optimization target includes adjusting the weight of each optimization target based on the operating status information and target achievement result information.

[0061] The operational status information of the scene devices includes: CPU / GPU temperature, power consumption, fan speed, task queue length, sensor sampling values, etc., while the target achievement result information can include the relationship with the target f1(x) i f2(x) i ), f m (x i The corresponding measured values, such as: measured security delay, electricity meter reading, user satisfaction rating, etc.; thus, performance evaluation can be performed, including: obtaining the relative deviation, for example: relative deviation Δm = (measured value) The system compares the target value (or user satisfaction score) with a set threshold. If the performance evaluation result is poor (e.g., the relative deviation is greater than a first set difference, or the user satisfaction score is less than a set threshold), the optimization objective can be adjusted or the candidate set of scheduling strategies can be updated. If the relative deviation is greater than the first set difference but less than a second set difference, the weight of each optimization objective can be adjusted. If the relative deviation is greater than or equal to the second set difference, it may be necessary to obtain a new candidate set of scheduling strategies based on NSGA-III and replace the old one. This allows the system to adjust the optimization objective or update the solution set based on the equipment's operating status and the achievement of the target, enabling the system to have adaptive capabilities. It can dynamically optimize the scheduling strategy based on actual operating results, continuously improve the overall efficiency of equipment collaborative scheduling, and ensure that equipment collaborative operation continuously meets optimization needs and adapts to changes in the environment and requirements during long-term operation.

[0062] The following describes the operational process in a specific embodiment, illustrating the system device collaborative scheduling process provided by the embodiments of the present invention.

[0063] In this embodiment of the disclosure, a smart community is provided, in which multiple cameras are installed for security, and multiple smart devices are also provided, such as lighting system devices, alarm system devices, etc. The central control screen device can be a decision-making device and can communicate with the cameras, lighting system devices, and alarm system devices.

[0064] Among them, the central control screen device can obtain and save a set of scheduling strategy candidates that match the security scenario of the community. Figure 3 This is a schematic diagram illustrating a process for obtaining a candidate set of scheduling strategies according to an embodiment of this disclosure. Figure 3 As shown, the process by which the central control screen device obtains a candidate set of scheduling strategies that match the community security scenario includes: Step 301: Based on the community security scenario requirements, the corresponding environmental information and the device status information of each scenario device, the central control screen device determines the security response target and energy consumption control target for the coordinated operation of two or more scenario devices.

[0065] Based on security requirements and energy consumption budget, the security response target is set as follows: cameras must activate within one second of detecting an anomaly. The energy consumption control target is that the monthly power consumption of the entire community will not exceed 15 kWh. Furthermore, the weight of each target can be determined based on environmental information and the device status information of each scene's equipment. For example, when security needs are higher at night, the weight of the security response target can be appropriately increased.

[0066] Step 302: The central control screen device determines the initial population based on the security response target and energy consumption control target, and uses the initial population as the current population.

[0067] In the initialization population, each individual represents a cooperative scheduling strategy, which includes response time and energy consumption.

[0068] Step 303: The central control screen device calculates the performance value of each individual in the current population on the security response target and energy consumption control target.

[0069] For example, in each iteration, individuals in the current population are evaluated, and their performance values ​​on the two optimization objectives of security response and energy consumption control are calculated.

[0070] Step 304: Based on the performance values, determine whether the security response target and energy consumption control target have been achieved. If yes, proceed to step 307; otherwise, proceed to step 305.

[0071] Step 305: Based on the performance values, the central control screen device iterates to generate the next generation population through non-dominated sorting, selection, and improved chromosome crossover and mutation operations.

[0072] Step 306: The central control screen device identifies the next generation population as the current population and returns to step 303.

[0073] Step 307: The central control screen device obtains a set of candidate scheduling strategies based on the iterated population, and stores the set of candidate scheduling strategies using a hash table.

[0074] After multiple iterations, a candidate set of scheduling strategies containing 100,000 strategies was generated. These scheduling strategies achieve different degrees of balance between security response and energy consumption control. For example, some scheduling strategies focus on fast security response and have relatively high energy consumption; others focus on low energy consumption and have slightly slower security response speeds.

[0075] In this way, the central control screen device saves a set of candidate scheduling strategies that match the community security scenario. Of course, through the above process, it can also save a set of candidate scheduling strategies that match multiple scenarios in the smart community. Thus, the central control screen device can make real-time security decisions and coordinate device scheduling based on the saved set of candidate scheduling strategies.

[0076] Figure 4 This is a flowchart illustrating a device collaborative scheduling method provided in an embodiment of this disclosure. (In conjunction with...) Figure 4 The process for coordinated scheduling of equipment includes: Step 401: Determine if the current scenario is a community security scenario? If yes, proceed to step 402; otherwise, return to step 401.

[0077] For example, if the current time is 23:00, the current scenario can be determined to be a community security scenario. Alternatively, if the number of people gathered in a certain area of ​​the community is greater than the set number, the current scenario can be determined to be a community security scenario.

[0078] Step 402: The central control screen device obtains the current environmental information corresponding to the community security scenario, as well as the current device status information corresponding to two or more scenario devices in the community security scenario.

[0079] Step 403: Based on the current device status information and current environment information, the central control screen device, using a lightweight inference engine, determines the performance value f1(p) of each strategy in the candidate scheduling strategy set for both security response objectives and energy consumption control objectives. i f2(p) i ).

[0080] Step 404: The central control screen device determines the performance value f1(p) based on the performance value f1(p). i f2(p)i The fitness of each strategy is obtained by formula (1), along with the corresponding weight values.

[0081] Step 405: The central control screen device determines the strategy corresponding to the maximum fitness value as the optimal scheduling strategy.

[0082] Step 406: The central control screen device coordinates the scheduling of devices in the community security scenario according to the optimal scheduling strategy.

[0083] Step 407: The central control screen device obtains the operating status information and target achievement result information of the scene devices, and receives feedback performance.

[0084] After cameras, lighting equipment, and other devices operate in coordination, they can send operational status information such as camera startup time and energy consumption to the central control screen. They can also obtain information on target achievement results, thereby receiving performance feedback, such as: relative deviation Δm = (measured value) Target value / target value, or user satisfaction rating, etc.

[0085] Step 408: When the feedback performance matches the preset weight adjustment performance conditions, the central control screen device adjusts the weights corresponding to the security response target and the energy consumption control target based on the operating status information and the target achievement result information.

[0086] Step 409: When the feedback performance matches the preset solution set adjustment performance conditions, the central control device obtains a new set of scheduling strategy candidates corresponding to the security response target and the energy consumption control target based on the NSGA-III algorithm, and saves the new set of scheduling strategy candidates after replacing the old set of scheduling strategy candidates.

[0087] As can be seen, in this embodiment, the NSGA-III algorithm is used to simulate the biological evolution process for global search, finding a set of strategy solutions that achieve an optimal balance between security response goals and energy consumption control goals. This significantly improves the overall efficiency of device collaborative scheduling, effectively reducing energy consumption while ensuring security effectiveness. Furthermore, the lightweight inference engine combined with optimization techniques can quickly make collaborative scheduling decisions, avoiding security vulnerabilities or energy waste caused by decision delays, greatly improving the response speed and practicality of device collaborative operation.

[0088] Based on the above-mentioned method for equipment collaborative scheduling, a device for equipment collaborative scheduling can be constructed, such as... Figure 5 As shown, it includes: an information acquisition module 510, a strategy determination module 520, and a collaborative scheduling module 530.

[0089] The information acquisition module 510 is configured to acquire the current environment information corresponding to the current scene, and the current device status information corresponding to two or more scene devices in the current scene.

[0090] The strategy determination module 520 is configured to determine the best scheduling strategy that matches the current device state information and the current environment information from the scheduling strategy candidate set based on the inference engine. Each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives for device cooperative operation.

[0091] The collaborative scheduling module 530 is configured to perform collaborative scheduling of two or more scene devices according to the optimal scheduling strategy.

[0092] In some embodiments, it also includes: The target determination module is configured to determine two or more optimization targets for the collaborative operation of two or more scenario devices based on scenario requirement information, environmental information corresponding to the scenario, and device status information of each scenario device.

[0093] The genetic iteration module is configured to be based on the NSGA-III algorithm to obtain and save a set of scheduling strategy candidates corresponding to two or more optimization objectives. Each strategy in the set of scheduling strategy candidates is a device cooperative scheduling scheme that achieves a balance between two or more optimization objectives.

[0094] In some embodiments, the genetic iteration module is specifically configured to determine an initial population, where each individual in the initial population represents a cooperative scheduling strategy, the cooperative scheduling strategy including: two or more decision variables; calculate the performance value of each individual in the initial population on two or more optimization objectives; and iteratively evolve the initial population based on the performance value through non-dominated sorting, selection, and improved chromosome crossover and mutation operations to obtain a candidate set of scheduling strategies.

[0095] In some embodiments, the genetic iteration module is also configured to store a set of candidate scheduling strategies using a hash table or database.

[0096] In some embodiments, the strategy determination module 520 includes: The first determining unit is configured to input the current device state information, the current environment information, and the scheduling policy candidate set into the lightweight inference engine, and determine the fitness value of each policy in the scheduling policy candidate set corresponding to the current device state information and the current environment information through forward propagation.

[0097] The second determining unit is configured to determine the strategy corresponding to the maximum fitness value as the optimal scheduling strategy.

[0098] In some embodiments, the first determining unit is specifically configured to determine each policy (p) in the scheduling policy candidate set based on the current device state information and the current environment information, using a lightweight inference engine.i The performance value f1(p) on each optimization objective i f2(p) i ), ..., f m (p i According to formula (1), the fitness of each strategy is obtained. F i =w1×f1(p i )+ w2×f2(p i )+…+ w m ×f m (p i (1) Among them, w1, w2, ..., w m The weights corresponding to each optimization objective.

[0099] In some embodiments, it also includes: The feedback adjustment module is configured to adjust the optimization targets or update the candidate set of scheduling strategies based on the obtained operating status information and target achievement results of the scene devices. The adjustment of optimization targets includes adjusting the weight of each optimization target based on the operating status information and target achievement results.

[0100] The following specific embodiments illustrate the device collaborative scheduling process of the device collaborative scheduling device provided by the present invention.

[0101] In this embodiment of the disclosure, in a smart home, the television can be either a decision-making device or a scene device, meaning that the television is equipped with a device for coordinated scheduling of devices. Scene devices may include: television, lights, air conditioners, etc.

[0102] Figure 6 This is a schematic diagram of a device for collaborative scheduling of equipment provided in an embodiment of this disclosure. Figure 6 As shown, the device for equipment collaborative scheduling includes: an information acquisition module 510, a strategy determination module 520, a collaborative scheduling module 530, a target determination module 540, a genetic iteration module 550, and a feedback adjustment module 560. The strategy determination module 520 includes: a first determination unit 521 and a second determination unit 522.

[0103] Based on the needs of home movie-watching scenarios, as well as corresponding environmental information (living room illuminance, temperature, humidity) and device status information for each scenario device (TV standby power consumption, SoundBar standby power consumption, smart light brightness, air conditioner cooling power), the target determination module 540 determines the "movie-watching immersion target" and "energy consumption control target" for the coordinated operation of two or more scenario devices. For example, the movie-watching immersion target is set as "light brightness ≤ 10% and color temperature 2700 K, sound delay ≤ 100ms, air conditioner silent mode", and the energy consumption control target is "total power consumption ≤ 0.3 kWh during the entire movie-watching period".

[0104] The genetic iteration module 550 determines the initial population based on the immersion goal and energy consumption control goal. Each individual contains decision variables: light brightness, color temperature, sound volume, TV backlight, air conditioner temperature and wind speed. It calculates the performance values ​​of each individual in the current population on the immersion goal and energy consumption control goal, including: immersion score, estimated power consumption, and iterates through non-dominated sorting, selection, crossover and mutation operations to obtain a set of 80,000 scheduling policy candidates, which are stored in a hash table. The policies achieve different balances between immersion and power consumption, such as high immersion and high energy consumption, low energy consumption and medium immersion.

[0105] Thus, when the user says "watch a movie" via voice or triggers the function with a single button press on the remote control, the current scenario is determined to be a home movie-watching scenario. Consequently, the information acquisition module 510 can acquire current environmental information and current device status information. The current environmental information may include: current illuminance and current temperature. The current device status information may include: TV power-on status, current brightness of lights, air conditioner operating status, etc.

[0106] Therefore, in the strategy determination module 520, the first determination unit 521, based on the TensorRT optimization inference engine, calculates the performance value of each scheduling strategy in the candidate scheduling strategy set on the immersion target and energy consumption target, and obtains the corresponding fitness according to formula (1). The second determination unit 522 can determine the scheduling strategy corresponding to the maximum fitness as the optimal scheduling strategy. Thus, the collaborative scheduling module 530 can issue collaborative instructions according to the optimal scheduling strategy, for example, adjusting the lights to 10% 2700 K, the SoundBar volume to 40%, the TV backlight to 80%, and the air conditioner to silent mode at 26℃.

[0107] After the scene devices operate according to the collaborative instructions, the feedback adjustment module 560 can obtain the scene devices' operating status information, including: measured brightness of the illuminance meter, power socket reading, actual audio delay, etc. It can also obtain target achievement result information, including: user satisfaction button rating, and obtain the corresponding feedback performance. If the feedback performance matches the preset weight adjustment performance conditions, the weights corresponding to the immersion target and energy consumption target can be adjusted. Alternatively, if the feedback performance matches the preset solution set adjustment performance conditions, a new set of scheduling strategy candidates corresponding to the immersion target and energy consumption target can be obtained based on the NSGA-III algorithm, and the old set of scheduling strategy candidates can be replaced and saved.

[0108] As can be seen, in this embodiment, the device for collaborative device scheduling uses the NSGA-III algorithm to simulate the biological evolution process for global search, finding a set of strategy solutions that achieve an optimal balance between immersion goals and energy consumption goals. This significantly improves the overall efficiency of collaborative device scheduling, effectively reducing energy consumption while ensuring the viewing experience. Furthermore, the lightweight inference engine combined with optimization techniques enables rapid collaborative scheduling decisions, greatly improving the response speed and practicality of collaborative device operation.

[0109] This disclosure provides an apparatus 700 for collaborative scheduling of devices, the structure of which is as follows: Figure 7 As shown, it includes: The processor 1000 and memory 1001 may further include a communication interface 1002 and a bus 1003. The processor 1000, communication interface 1002, and memory 1001 can communicate with each other via the bus 1003. The communication interface 1002 can be used for information transmission. The processor 1000 can invoke logical instructions stored in the memory 1001 to execute the device collaborative scheduling method described in the above embodiment.

[0110] Furthermore, the logic instructions in the aforementioned memory 1001 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0111] The memory 1001, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this disclosure. The processor 1000 executes functional applications and data processing by running the program instructions / modules stored in the memory 1001, that is, it implements the method for device collaborative scheduling in the above method embodiments.

[0112] The memory 1001 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 1001 may include high-speed random access memory and may also include non-volatile memory.

[0113] This disclosure provides a device for collaborative scheduling, including a processor and a memory storing program instructions, wherein the processor is configured to execute a method for collaborative scheduling of devices when executing the program instructions.

[0114] This disclosure provides a system for collaborative scheduling of devices, including: a decision-making device, and two or more scene devices; wherein, The decision-making device is configured to acquire the current environment information corresponding to the current scene, and the current device status information corresponding to two or more scene devices in the current scene; based on the inference engine, it determines the best scheduling strategy that matches the current device status information and the current environment information from the scheduling strategy candidate set, wherein each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives of device collaborative operation; and performs collaborative scheduling of two or more scene devices according to the best scheduling strategy.

[0115] This disclosure provides a storage medium storing program instructions that, when executed, perform the method for device collaborative scheduling as described above.

[0116] This disclosure provides a computer program product, which includes a computer program stored on a storage medium. The computer program includes program instructions that, when executed by a computer, cause the computer to perform the data retrieval method described above.

[0117] The aforementioned storage medium can be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0118] The technical solutions of this disclosure can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in this disclosure. The aforementioned storage medium can be a non-transitory storage medium, including: a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and other media capable of storing program code; it can also be a transient storage medium.

[0119] The foregoing description and accompanying drawings fully illustrate embodiments of the present disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, procedural, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included or replace parts and features of other embodiments. The scope of the embodiments of this disclosure includes the entire scope of the claims and all available equivalents of the claims. While the terms “first,” “second,” etc., may be used in this application to describe elements, these elements should not be limited by these terms. These terms are used only to distinguish one element from another. For example, a first element may be called a second element without changing the meaning of the description, and similarly, a second element may be called a first element, provided that all occurrences of “first element” are consistently renamed and all occurrences of “second element” are consistently renamed. First and second elements are both elements, but may not be the same element. Moreover, the terminology used in this application is only for describing embodiments and is not intended to limit the claims. As used in the description of the embodiments and claims, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” are intended to also include the plural forms. Similarly, the term “and / or” as used herein means including one or more of the associated listed any and all possible combinations. Additionally, when used herein, the terms “comprise” and its variations “comprises” and / or “comprising” refer to the presence of stated features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof. Without further limitations, an element defined by the phrase “comprising an…” does not exclude the presence of additional identical elements in the process, method, or apparatus that includes said element. In this document, each embodiment may focus on the differences from other embodiments, and similar or identical parts between embodiments can be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section disclosed in the embodiments, the relevant parts can be referred to the description of the method section.

[0120] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this disclosure. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0121] The methods and products disclosed in the embodiments herein (including but not limited to devices and equipment) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units may be merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to implement this embodiment according to actual needs. In addition, the functional units in the embodiments of this disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0122] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

Claims

1. A method for collaborative scheduling of equipment, characterized in that, include: Obtain the current environment information corresponding to the current scene, as well as the current device status information corresponding to two or more scene devices in the current scene; Based on the inference engine, the best scheduling strategy that matches the current device state information and the current environment information is determined from the scheduling strategy candidate set. Each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives for device cooperative operation. Based on the optimal scheduling strategy, coordinate the scheduling of two or more scenario devices.

2. The method according to claim 1, characterized in that, Also includes: Based on the scenario requirements, the corresponding environmental information, and the device status information of each scenario device, determine two or more optimization objectives for the collaborative operation of two or more scenario devices. Based on the third-generation non-dominated sorting genetic NSGA-III algorithm, a set of scheduling strategy candidates corresponding to two or more optimization objectives is obtained and stored. Each strategy in the set of scheduling strategy candidates is a device cooperative scheduling scheme that achieves a balance between two or more optimization objectives.

3. The method according to claim 2, characterized in that, The step of obtaining and storing the candidate set of scheduling strategies corresponding to two or more optimization objectives includes: Determine the initial population, where each individual in the initial population represents a cooperative scheduling strategy, which includes two or more decision variables; Calculate the performance value of each individual in the initial population on two or more optimization objectives; Based on the performance values, the initial population is iteratively evolved through non-dominated sorting, selection, and improved chromosome crossover and mutation operations to obtain a candidate set of scheduling strategies.

4. The method according to claim 2, characterized in that, The step of obtaining and storing the candidate set of scheduling strategies corresponding to two or more optimization objectives also includes: The candidate set of scheduling strategies is stored using a hash table or database.

5. The method according to claim 1, characterized in that, The determination of the optimal scheduling strategy that matches the current device status information and the current environment information includes: The current device status information, current environment information, and the scheduling policy candidate set are input into the lightweight inference engine. Through forward propagation, the fitness value of each policy in the scheduling policy candidate set corresponding to the current device status information and current environment information is determined. The strategy corresponding to the maximum fitness value is determined as the optimal scheduling strategy.

6. The method according to claim 5, characterized in that, The fitness value corresponding to each strategy in the candidate scheduling strategy set and the current device state information and current environment information includes: Based on the current device status information and the current environment information, and using a lightweight inference engine, each policy (p) in the candidate scheduling policy set is determined. i The performance value f1(p) on each optimization objective i f2(p) i ), ..., f m (p i ); According to formula (1), the fitness of each strategy is obtained; F i =w1×f1(p i )+ w2×f2(p i )+…+ w m ×f m (p i )(1) Among them, w1, w2, ..., w m The weights corresponding to each optimization objective.

7. The method according to any one of claims 1-6, characterized in that, After performing collaborative scheduling of two or more scene devices, the method further includes: Based on the obtained operating status information and target achievement information of the scene devices, adjust and optimize the targets or update the candidate set of scheduling strategies; among them, adjusting the optimization targets includes adjusting the weight of each optimization target based on the operating status information and target achievement information.

8. An apparatus for coordinated scheduling of devices, the apparatus comprising a processor and a memory storing program instructions, characterized in that, The processor is configured to perform the method for device cooperative scheduling as described in any one of claims 1-7 when executing the program instructions.

9. A system for coordinated scheduling of equipment, characterized in that, include: Decision-making equipment, and two or more scene devices; among which, The decision-making device is configured to acquire the current environment information corresponding to the current scene, and the current device status information corresponding to two or more scene devices in the current scene; based on the inference engine, it determines the best scheduling strategy that matches the current device status information and the current environment information from the scheduling strategy candidate set, wherein each strategy in the scheduling strategy candidate set is obtained by a multi-objective optimization algorithm based on two or more optimization objectives of device collaborative operation; and performs collaborative scheduling of two or more scene devices according to the best scheduling strategy.

10. A storage medium storing program instructions, characterized in that, When the program instructions are executed, they perform the method for device collaborative scheduling as described in any one of claims 1-7.

Images