Edge computing system and scheduling method for heterogeneous multi-source sensors

Abstract

An edge computing system for heterogeneous multi-source sensors includes an event buffer for storing output data of the heterogeneous multi-source sensors; a scheduling module, which starts sorting events in the event buffer when an event is added to the event buffer; a task storage area for storing tasks executed by the edge computing system; and an execution module, which obtains events in the event buffer, selects corresponding tasks in the task storage area and executes them. In addition, the present invention also discloses a scheduling method for an edge computing system for heterogeneous multi-source sensors. The invention is particularly suitable for many application scenarios of heterogeneous multi-source sensors, and can be better compatible with various protocols of connected sensors while reducing system costs, and effectively improve the real-time performance of input signal processing.

Description

Technical Field

[0001] The present invention relates to the field of computer systems, and in particular to an edge computing system and a scheduling method for heterogeneous multi-source sensors. Background Art

[0002] With the advent of the Internet of Things era, in the intelligent transformation of traditional power systems, tens of thousands of traditional power parameter detection sensors are facing an urgent need to access the Internet and cloud computing. The cloud server computing model based on improving the computing speed of cloud processors and increasing the number of concurrent processors to improve cloud computing power cannot meet the real-time requirements of sensor signal processing in many power industries; at the same time, as the number of cloud server access nodes increases, the cost of cloud computing hardware and the complexity of cloud software will soar rapidly, which cannot meet the goal of building an intelligent power system with high cost performance.

[0003] In addition, there are many types of detection sensors in the actual power system, and the protocol standards vary with the version iteration. The communication buses include RS-485, RS-232, Ethernet, narrowband wireless, etc. If tens of thousands of traditional sensors are connected to the cloud wirelessly or wired, the difficulty and labor cost of protocol compatibility, network wiring and on-site construction will be greatly increased; secondly, it will lead to the extension of the data transmission path from the sensor to the cloud, and the real-time, stability and reliability of the system operation process cannot be guaranteed. Summary of the invention

[0004] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the edge computing system is particularly suitable for power parameter detection sensor scenarios with small data transmission volume, low data communication frequency, and high real-time requirements.

[0005] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the edge computing system can divide application scenarios into areas and aggregate and process heterogeneous multi-source sensors in the same area to achieve real-time data processing.

[0006] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the edge computing system can effectively shorten the wiring path of the heterogeneous multi-source sensors, effectively reduce the mutual interference and construction costs during communication between the sensors, and improve the accuracy of sensor data communication.

[0007] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the edge computing system can communicate with a cloud server to reduce the computing power requirements and software complexity of the cloud server and save manpower and material resources.

[0008] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the signal preprocessing module in the edge computing system can quickly extract events and data from the output of the heterogeneous multi-source sensors, effectively improving the efficiency of the scheduling algorithm.

[0009] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the scheduling module in the edge computing system can prioritize events in the event buffer, so that the present invention is more efficient and flexible than the scheduling algorithm based on task priority sorting, and is more suitable for the single-board computer design scheme for microprocessor applications.

[0010] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the execution module in the edge computing system can find and execute tasks corresponding to events from a task storage area, and optimize the event scheduling algorithm by obtaining the task execution time to make the edge computing system more intelligent.

[0011] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein a task storage area of ​​the edge computing system is communicatively connected to a cloud server, and tasks in the task storage area that exceed a preset threshold can be deployed on the cloud server for execution, so as to realize the real-time performance of the edge computing system.

[0012] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the edge computing system is connected to a cloud server, and can schedule exceptions, results and task execution time during task execution through an event buffer to improve the real-time performance of the edge computing system.

[0013] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the edge computing system is connected to a cloud server, and can schedule exceptions, results and task execution time during task execution through an execution module, thereby enhancing the response speed of cloud events.

[0014] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the event buffer, scheduling module, event priority table and task execution schedule in the scheduling method establish a self-regulating processing model for event input, sorting, execution and feedback, thereby improving the robustness of the scheduling method.

[0015] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the cloud server in the scheduling method effectively shares the task execution work in the edge computing system without increasing the complexity of edge computing system development, thereby improving the real-time performance of the edge computing system in processing sensor events.

[0016] One object of the present invention is to provide an edge computing system and scheduling method for heterogeneous multi-source sensors, wherein the cloud server in the scheduling method can transmit the task execution status to the edge computing system in the form of events, thereby realizing an easily scalable data communication method from the cloud to the edge without affecting the hard real-time event processing.

[0017] In one aspect of the present invention, the present invention provides an edge computing system for heterogeneous multi-source sensors, comprising:

[0018] an event buffer, storing output data of the heterogeneous multi-source sensor, parsing the output data to obtain an event, and adding the event to the event buffer;

[0019] a scheduling module, which starts sorting the events in the event buffer when an event is added to the event buffer;

[0020] a task storage area for storing tasks executed by the edge computing system; and

[0021] An execution module obtains events in the event buffer, selects corresponding tasks in the task storage area and executes them.

[0022] According to one embodiment of the present invention, the edge computing system further includes an event priority list, which records the priorities of the heterogeneous multi-source sensor output events, and the scheduling module sorts the events in the event buffer according to the event priority list.

[0023] According to one embodiment of the present invention, the edge computing system further includes a task execution schedule, which records the execution time of the tasks in the task storage area, and the task execution schedule participates in the operation of the scheduling module.

[0024] According to one embodiment of the present invention, the edge computing system is also communicatively connected to a cloud server, and the cloud server is used to execute the task corresponding to the event, wherein the task execution time is greater than a first preset threshold.

[0025] According to one embodiment of the present invention, events generated when the cloud server executes tasks in the task storage area are fed back to the edge computing system through the event buffer.

[0026] According to one embodiment of the present invention, the edge computing system further includes a task list for identifying tasks corresponding to events in the event buffer to be executed in the edge computing system or the cloud server.

[0027] In another aspect of the present invention, the present invention provides an edge computing system scheduling method for heterogeneous multi-source sensors, comprising the following steps:

[0028] (a) acquiring output data of the heterogeneous multi-source sensor, parsing the output data to obtain an event, and adding the event to an event buffer;

[0029] (b) triggering a scheduling module to sort the events in the event buffer according to an event priority list, wherein the event priority list records the priorities of the events output by the heterogeneous multi-source sensors;

[0030] (c) an execution module obtains events in the event buffer and calls a corresponding task in a task storage area;

[0031] (d) the execution module records the task execution time and writes it into a task execution schedule; and

[0032] (e) The task execution schedule participates in the calculation of the scheduling module.

[0033] According to one embodiment of the present invention, the method in step (a) further comprises the steps of:

[0034] (a.1) when parsing the output data of the heterogeneous multi-source sensor, determining whether an event included in the output data has data;

[0035] (a.2) If the event has data, storage information of the data is associated with the event.

[0036] According to one embodiment of the present invention, the step (c) further comprises the following steps:

[0037] (c.1) before calling the corresponding task in the task storage area, first obtain the execution time of the corresponding task from the task execution schedule;

[0038] (c.2) determining whether the execution time is greater than a first preset threshold;

[0039] (c.3) if the execution time is greater than the first preset threshold, the corresponding task is transferred to a cloud server for execution;

[0040] (c.4) The execution time of the corresponding task is written into the task execution schedule.

[0041] According to one embodiment of the present invention, the step (c.3) further includes a task list, wherein the task list records the execution of the corresponding tasks in the edge computing system or the cloud server.

[0042] According to one embodiment of the present invention, the step (c.4) further includes the following steps:

[0043] The cloud server directly transmits the task execution time to the execution module, or the cloud server adds the task execution time to the event buffer in the form of an event; then the execution module selects the corresponding task for processing according to the event type; finally, the execution module notifies the cloud server of the time when the event is obtained as the time when the cloud server executes the task, and writes it into the task execution schedule. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic block diagram of an application of an edge computing system for heterogeneous multi-source sensors according to the present invention.

[0045] Figure 2 It is a schematic diagram of executing the edge computing system according to a preferred embodiment of the present invention.

[0046] Figure 3 It is a schematic diagram of the improved execution of the edge computing system according to the above-mentioned preferred embodiment of the present invention.

[0047] FIG. 4A to FIG. 4C It is a schematic diagram of the execution of the edge computing system and the cloud server according to the above preferred embodiment of the present invention.

[0048] Figure 5 It is another improved execution diagram of the edge computing system and the cloud server according to the above-mentioned preferred embodiment of the present invention.

[0049] in,

[0050] 1. Event buffer, 2. Task storage area. DETAILED DESCRIPTION

[0051] Before describing in detail any embodiment of the present invention, it should be understood that the present invention is not limited in its application to the construction and arrangement details of the components described below or illustrated in the following figures. The present invention can have other embodiments and can be practiced or carried out in various ways. In addition, it should be understood that the words and terms used here are for descriptive purposes and should not be considered restrictive. The use of "including", "comprising" or "having" and its variations herein is intended to cover the items and their equivalents and additional items displayed below. Unless otherwise specified or limited, the terms "install", "connect", "support" and "couple" and their variations are widely used and cover direct installation and indirect installation, connection, support and connection. In addition, "connect" and "couple" are not limited to physical or mechanical connections or connections.

[0052] Furthermore, on the first aspect, in the disclosure of the present invention, the orientation or positional relationship indicated by terms such as "longitudinal", "transverse", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" are based on the orientation or positional relationship shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore the above terms cannot be understood as limitations on the present invention; on the second aspect, the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" cannot be understood as a limitation on the quantity.

[0053] Refer to the attached specification of the present invention Figure 1 , discloses a schematic connection method of a heterogeneous multi-source sensor and an edge computing system in a preferred embodiment. Figure 1 In the example, the output signals of a temperature sensor, a humidity sensor, a gas monitoring sensor, an electrical parameter monitoring sensor, and a smoke detection sensor are communicatively connected to the input of the edge computing system. Optionally, the output of the edge computing system is communicatively connected to an emergency control switch to implement a scenario where the sensor, device, or system connected to the edge computing system encounters an emergency and requires emergency treatment. It is worth noting that Figure 1 The temperature sensor, humidity sensor, etc. disclosed in the disclosure are all specific implementations of the heterogeneous multi-source sensor. The heterogeneous in the heterogeneous multi-source sensor refers to the different types or structures of sensors, and the different data communication interfaces and protocols. Multi-source refers to at least three sensors, not just one or two sensors interacting with the edge computing system. Figure 1The communication method between the sensor and the edge computing system can be but not limited to RS-232, RS-485, CAN, Zigbee, WiFi, Bluetooth, Ethernet, etc. In addition, Figure 1 The system also includes a cloud server for receiving the data parsed by the edge computing system and Figure 1 The sensor data is stored and managed in a database in the cloud. The edge computing system can obtain sensor data uploaded by itself or other edge computing systems from the cloud server. The cloud server is also communicatively connected to a control room, a monitoring room, a regulatory department, a management client and an operation and maintenance client to enable the data of the cloud server to interact with the user.

[0054] Figure 2 The internal processing modules and data flow of the edge computing system are schematically disclosed, including an event buffer 1 for storing events, a scheduling module, an event priority list, an execution module and a task storage area 2. Figure 1 , the edge computing system receives the input signal of the sensor and pushes it into the event buffer 1 in the form of an event. At this time, the scheduling module is triggered and called; then the scheduling module queries the event priority list, obtains the priority of the event pushed into the event buffer 1, and adjusts the pop-up order of the event according to the priority; the execution module obtains an event in the event buffer 1, queries the task in the task storage area 2 according to the event, and executes it. Among them, the number of events that can be stored in the event buffer 1 is one of 1, 2, 3, ..., N, and the number of tasks that can be stored in the task storage area 2 is one of 1, 2, 3, ..., M, and N and M are natural numbers.

[0055] It is worth mentioning that the event buffer 1 can be, but is not limited to, a first-in, first-out buffer. In addition, after the edge computing system obtains the sensor input signal and before pushing the event into the event buffer 1, it also includes a signal preprocessing module. The signal preprocessing module parses and preprocesses the original sensor signal obtained by the edge computing system, extracts the signal of each sensor, and converts the signal into an event. The signal preprocessing module can be, but is not limited to, the following two working modes:

[0056] In one way, the signal preprocessing module numbers each sensor connected to the edge computing system, uses the sensor number as the event type, and uses the data obtained from the input signal of the preprocessed sensor as the event data. The event type and event data together constitute an event. The advantage of doing so is that when the amount of sensor data connected to the edge computing system is large and is implemented by different construction units, it is helpful for clarifying the responsibilities of each unit. In addition, when designing the tasks in the task storage area 2, they can be divided according to the sensor number. Only one task is assigned to one sensor number, and the task divides different event data into subtasks. This solution is well-organized and has a clear division of labor. It is more conducive to the cooperative development of hardware and software systems in the edge computing system by multiple departments and units, and is not restricted by the geographical distance of the construction unit.

[0057] In another way, the signal preprocessing module divides and specifies the data of each sensor of the edge computing system one by one, and divides the sensor data into multiple events according to their importance. The number of events owned by each sensor can be different, and the event priority of the same sensor can also be different. In other words, first of all, it is necessary to clearly count the number, category, output data type, output data range, etc. of sensors connected to the edge computing system before design, and they can be listed one by one in the form of a table. Secondly, according to the sensor and the event type owned by the sensor, a priority is assigned to each event according to the time limit of the event to the system response, and it is stored in the event priority list. In this way, the data content output by some sensors can be directly converted into a defined event, and the event can no longer be accompanied by data content, which will better reduce the event length and increase the storage space and sorting time of the event in the event buffer 1. Since each event is clearly defined in the event priority list, the working efficiency of the scheduling module is further improved. This implementation is particularly suitable for small and medium-sized systems where sensors connected to the edge computing system are concentrated, the implementation distance is short, and simultaneous construction is convenient. Its advantages are high efficiency in event scheduling and task execution, and fast processing speed of sensor output data, which can meet application scenarios with high requirements for real-time processing.

[0058] The working principle of the edge computing system is described by taking three events as examples. It is assumed that the event priority is a positive integer from 0 to 9, where 0 has the highest priority and 9 has the lowest priority. The event priority list structure is described exemplarily, and the priority values ​​of each event are assumed, as shown in Table 1.

[0059] Table 1 Event priority list structure

[0060]

[0061] Figure 3 is Figure 2A task execution schedule is added on the basis of . There are two ways to obtain the initial value of the task execution schedule. One is to obtain it by measuring the length of the task execution time; the other is to obtain it by timing when the edge computing system performs software testing after the task program design is completed. The execution time value of each task corresponding to the task execution schedule is updated by the execution module according to the actual execution time of the task each time the task is executed, and is used as one of the parameters in the event priority scheduling by the scheduling module.

[0062] It is worth mentioning that each event in the present invention corresponds to only one task. This often causes the task code in the task storage area 2 to be repeated, and the use of the task storage area 2 occupies most of the processing and storage resources. Therefore, further, the same code in all tasks can be extracted to form a common function component area. In the common function component area, the same program code in the execution of more than two tasks is stored, and when the task corresponding to the event is executed, only the program code in the common function component area is called.

[0063] For example, in combination with the event priority table, the task execution time list is shown in Table 2.

[0064] Table 2 Task execution time list

[0065]

[0066] The event priority determination algorithm of the scheduling module in the edge computing system can use the product of the pre-set priority of the event minus the inverse of the task execution time and a preset coefficient, and round the calculation result to an integer, so that the scheduling module does not need to use floating-point data operations when calculating, and can directly use integer operators to realize event priority calculation, effectively improving the calculation and sorting speed of event priorities, and better realizing the real-time nature of event processing. Assuming that the preset coefficient is 10, according to the values ​​in Table 1 and Table 2, it can be concluded that the event priorities participating in the scheduling module are as follows:

[0067] Event 1 = 2-10÷12 = 2–0 = 2

[0068] Event 2 = 3-10÷4 = 3–2 = 1

[0069] Event 3 = 7-10÷30 = 7–0 = 7

[0070] Therefore, the scheduling module prioritizes the three events in the event buffer 1 according to the priority of event 1 being 2, the priority of event 2 being 1, and the priority of event 3 being 7. This is only a simple exemplary description of the event priority calculation in this embodiment.

[0071] Furthermore, the preset coefficient can also be obtained by the following method:

[0072] (1) Calculate the average of the most recent execution time of all tasks to obtain a first average value;

[0073] (2) eliminating all tasks whose execution times, the difference between the most recent execution time of all tasks and the first average value, exceeds a first preset threshold;

[0074] (3) The remaining task execution time after elimination is averaged again to obtain a second average value, where the second average value is the preset coefficient.

[0075] Among them, the first preset threshold is preferably 10% to 20% of the average value, which can dynamically adjust the priority of the events in the event buffer 1 in a timely manner, better coordinate the processing efficiency between events, and improve the real-time performance of event processing. When the first preset threshold is 10%, the adjustment range of the task execution time list for the event priority is relatively small; when the first preset threshold is 20%, the adjustment range of the task execution time list for the event priority is relatively large; when the first preset threshold is more than 20%, due to the excessive adjustment range of the event priority, the priorities of some high-priority events and low-priority events will be reversed, affecting the real-time performance of the preset high-priority event processing. Through creative work, the inventors have found that for power systems, especially substation systems, the first preset threshold in the range of 10%-20% can better improve the real-time performance of event processing.

[0076] Now turn to the attached Figure 1 In the embodiment, the cloud server is communicatively connected to the edge computing system and pushes data from the cloud server to the client. Similarly, the cloud server can also push the user's operation to the edge computing system through the cloud server to enable the user to control the emergency switch. In addition, Figure 4A The invention also illustrates a schematic method of communicating directly between the cloud server and the execution module in the edge computing system. Figure 4A During the communication between the cloud server and the execution module, if the execution module finds that the execution time of the task code in the task storage area 2 is greater than a second preset threshold, when the execution module encounters the event again, the task code is placed on the cloud server for execution, and the execution result of the cloud server is fed back to the execution module, thereby increasing the speed at which the execution module reads other events. The selection of the second preset threshold needs to meet two conditions at the same time:

[0077] (1) The second preset threshold is greater than ten times the time taken by the execution module from calling the cloud server to the cloud server starting to execute the task code;

[0078] (2) The task execution time is less than four times the average execution time of all tasks.

[0079] Furthermore, Figure 4B The edge computing system further includes a task list, which is used to record events and their corresponding tasks to be executed on the cloud server or the edge computing system.

[0080] exist Figure 4B Going a step further, we get Figure 4C The cloud server sends the return result and exception information of its executed task to the edge computing system in the form of events, and the edge computing system pushes the events into the event buffer 1. Figure 4A and Figure 4B , Figure 4C The approach is more suitable for situations where there are more tasks to be performed on the cloud server. The advantage of this solution is that it clarifies the full life cycle of event issuance, circulation, and execution, and treats the feedback signal of the cloud server as the output signal of the external sensor, which is more conducive to improving the real-time response of high-priority events.

[0081] Figure 5 is Figure 4C Going one step further, a communication channel is established between the task storage area 2 of the edge computing system and the cloud server. The communication can migrate the tasks of the task storage area 2 between the edge computing system and the cloud server, thereby saving the task storage space of the edge computing system.

[0082] It is worth mentioning that the present invention Figure 5 In addition to the implementation method, all task codes of the edge computing system can also be placed on the cloud server at the same time. The execution module determines whether the task is executed in the edge computing system or the cloud server through the task list, thereby further improving the execution efficiency of the event.

[0083] Reference Figure 3 The present invention also discloses an edge computing system scheduling method for heterogeneous multi-source sensors, and the specific steps are as follows.

[0084] First, the edge computing system obtains the output signal of the heterogeneous multi-source sensor, parses the output signal to obtain an event, and adds the event to an event buffer 1;

[0085] Secondly, the event triggering a scheduling module is added to the event buffer 1, and the scheduling module sorts the events in the event buffer 1 according to an event priority list, and the event priority list records the priority of the output events of the heterogeneous multi-source sensors;

[0086] Next, an execution module obtains the event in the event buffer 1 and calls a corresponding task in a task storage area 2;

[0087] Then, the execution module records the task execution time and writes it into a task execution schedule; and

[0088] Finally, the task execution schedule participates in the calculation of the scheduling module.

[0089] It is worth mentioning that the edge computing system obtains the event by passing the signal output by the sensor through a signal preprocessing module. The working process of the signal preprocessing module includes the following steps:

[0090] First, the edge computing system receives a signal sent by a sensor through interruption or polling, and distinguishes events and data information contained in the signal; and

[0091] Next, if the signal contains data, the storage information of the data is associated with the event, so that the data can be passed to the task as a parameter when the execution module calls the task, or can be used during the execution of the corresponding task of the event.

[0092] It is worth mentioning that the reference Figure 4A-4B The execution module can also allocate tasks to the cloud server or the edge computing system for execution according to the task execution time, and its allocation strategy is as follows:

[0093] First, before calling the corresponding task in the task storage area 2, the execution module obtains the execution time of the corresponding task from the task execution schedule;

[0094] Secondly, determining whether the execution time is greater than a first preset threshold;

[0095] Then, if the execution time is greater than the first preset threshold, the corresponding task is transferred to a cloud server for execution;

[0096] Finally, the execution module writes the execution time of the corresponding task into the task execution schedule.

[0097] It is worth mentioning that the reference Figure 4C There are two methods for calculating the time it takes for the cloud server to execute a task:

[0098] First, the cloud server directly transmits the task execution time to the execution module;

[0099] Secondly, the cloud server adds the task execution time to the event buffer 1 in the form of an event; then the execution module selects the corresponding task for processing according to the event type; finally, the execution module notifies the cloud server of the time when the event is obtained as the time when the cloud server executes the task, and writes it into the task execution schedule.

[0100] Generally speaking, in the above two task execution timing methods, the cloud server execution time is different, and each has its own applicable scenarios. The first method is applicable to the scenario where the tasks executed on the cloud server are relatively small and the processor resource occupancy rate of the edge computing system is less than 30%; the second method is applicable to the situation where one of the following conditions is met.

[0101] (1) The processing occupancy rate of the edge computing system is above 40%;

[0102] (2) The cloud server execution task data accounts for more than 10% of the total number of executed tasks;

[0103] (3) During the execution of the task by the cloud server, multiple signals need to interact with the execution module;

[0104] (4) The sensor places strict real-time requirements on the edge computing system to process the signals it sends.

[0105] Reference Figure Attached Figure 5 In the edge computing system scheduling method for heterogeneous multi-source sensors, after the execution module obtains the name of the task corresponding to the event, it notifies the edge computing system or the cloud server to execute the corresponding task according to the record in the task list.

[0106] It is worth mentioning that the task list may further include an event name, and the event name is used to query its corresponding task, thereby further improving the execution efficiency of the execution module.

[0107] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. Edge computing system for heterogeneous multi-source sensors, characterized by: include: an event buffer, storing output data of the heterogeneous multi-source sensor, parsing the output data to obtain an event, and adding the event to the event buffer; a scheduling module, which starts sorting the events in the event buffer when an event is added to the event buffer; a task storage area for storing tasks executed by the edge computing system; An execution module, which obtains events in the event buffer, selects corresponding tasks in the task storage area and executes them; an event priority list recording the priorities of the events output by the heterogeneous multi-source sensors, and the scheduling module sorting the events in the event buffer according to the event priority list; and A task execution schedule records the execution time of the tasks in the task storage area, and the task execution schedule participates in the calculation of the scheduling module; wherein, The scheduling module's event priority determination algorithm uses the product of the event's preset priority minus the inverse of the task execution time and a preset coefficient, and rounds the result; wherein, The preset coefficient is obtained by the following method: (1) Calculate the average of the most recent execution time of all tasks to obtain a first average value; (2) eliminating all tasks whose execution times, the difference between the most recent execution time of all tasks and the first average value, exceeds a first preset threshold; (3) The remaining task execution time after elimination is averaged again to obtain a second average value, where the second average value is the preset coefficient.

2. According to the edge computing system of claim 1, the edge computing system is also communicatively connected to a cloud server, and the cloud server is used to execute the task corresponding to the event, wherein the task execution time is greater than a first preset threshold.

3. An edge computing system according to claim 2, wherein events generated by the cloud server executing tasks in the task storage area are fed back to the edge computing system through the event buffer.

4. The edge computing system according to claim 2, wherein the edge computing system further includes a task list, identifying tasks corresponding to events in the event buffer to be executed in the edge computing system or the cloud server.

5. The edge computing system scheduling method for heterogeneous multi-source sensors is characterized by: The scheduling method comprises the following steps: (a) acquiring output data of the heterogeneous multi-source sensor, parsing the output data to obtain an event, and adding the event to an event buffer; (b) triggering a scheduling module to sort the events in the event buffer according to an event priority list, wherein the event priority list records the priorities of the events output by the heterogeneous multi-source sensors; (c) an execution module obtains events in the event buffer and calls a corresponding task in a task storage area; (d) the execution module records the task execution time and writes it into a task execution schedule; and (e) The task execution schedule participates in the calculation of the scheduling module; wherein, The scheduling module's event priority determination algorithm uses the product of the event's preset priority minus the inverse of the task execution time and a preset coefficient, and rounds the result; wherein, The preset coefficient is obtained by the following method: (1) Calculate the average of the most recent execution time of all tasks to obtain a first average value; (2) eliminating all tasks whose execution times, the difference between the most recent execution time of all tasks and the first average value, exceeds a first preset threshold; (3) The remaining task execution time after elimination is averaged again to obtain a second average value, where the second average value is the preset coefficient.

6. The scheduling method according to claim 5, further comprising the steps of: (a.1) when parsing the output data of the heterogeneous multi-source sensor, determining whether an event included in the output data has data; (a.2) If the event has data, storage information of the data is associated with the event.

7. The scheduling method according to claim 5, wherein the edge computing system is also communicatively connected to a cloud server, and the step (c) further comprises the following steps: (c.1) before calling the corresponding task in the task storage area, first obtain the execution time of the corresponding task from the task execution schedule; (c.2) determining whether the execution time is greater than a first preset threshold; (c.3) if the execution time is greater than the first preset threshold, the corresponding task is transferred to the cloud server for execution; (c.4) The execution time of the corresponding task is written into the task execution schedule.

8. The scheduling method according to claim 7, wherein the step (c.3) further comprises a task list, wherein the task list records the corresponding tasks running in the edge computing system or the cloud server.

9. The scheduling method according to claim 7, wherein the step (c.4) further comprises the following steps: The cloud server directly transmits the task execution time to the execution module, or the cloud server adds the task execution time to the event buffer in the form of an event; then the execution module selects the corresponding task for processing according to the event type; finally, the execution module notifies the cloud server of the time when the event is obtained as the time when the cloud server executes the task, and writes it into the task execution schedule.

Images