CN116521278A

CN116521278A - Reactive scheduling method, device, system and medium

Info

Publication number: CN116521278A
Application number: CN202310400738.6A
Authority: CN
Inventors: 王睿; 李慎国
Original assignee: Hetang Digital Technology Group Hainan Co ltd
Current assignee: Hetang Digital Technology Group Hainan Co ltd
Priority date: 2023-04-14
Filing date: 2023-04-14
Publication date: 2023-08-01

Abstract

The invention discloses a reactive scheduling method, device, system and medium, wherein the method comprises the following steps of initializing a finite state machine; in response to receiving at least one task data, acquiring an initial characteristic value of a finite state machine, and generating a scheduling scheme corresponding to the task data by static scheduling based on the shortest task construction period of the task data as a constraint condition; acquiring a characteristic value of the finite state machine in response to at least one dynamic event occurring and state transition of at least one leaf-level state machine of the finite state machine in the process of executing the scheduling scheme; judging whether the current state of the task data accords with the set step according to the characteristic value of the finite state machine; and triggering dynamic scheduling and updating a scheduling scheme in response to the step that the current state of the task data does not accord with the setting. The method can greatly shorten the data acquisition time and improve the dispatching processing period under the use environment of the task characteristics of high-frequency multidimensional association.

Description

Reactive scheduling method, device, system and medium

Technical Field

The invention belongs to the technical field of scheduling, and particularly relates to a reactive scheduling method, device, system and medium.

Background

With the rapid promotion of computer applications such as intelligent manufacturing, intelligent cities, industrial AI, digital twin and the like, a large business software platform actually forms a complex huge system, is in a multidimensional space of mass data, all business elements (production elements/service elements) in the system feed back own states in real time, and have extremely complex association relations among the states, and meanwhile, the multidimensional association is in a high-frequency dynamic process, such as the high-speed operation of an automatic production line and the real-time supply of water, electricity and gas city infrastructures. In this context, by delaying feedback through bi-directional asynchronous communication, the process processing or scheduling method cannot be integrated with real-time control of the execution units (PLC/AGV/ROBOT) to achieve management and control, which tends to bring about disjointed execution affecting efficiency until the resource allocation is chaotic and uncontrollable. How to ensure orderly and correct operation of each unit/component of the system and to have certain foolproof and error proof capability is a key point for realizing function improvement effect.

Patent document CN112904818A (application number: 202110068572.3) discloses a "complex structural member processing shop prediction-reaction scheduling method", which is mainly aimed at evaluating the influence of a disturbance event on a pre-scheduling scheme based on a designed relative performance deviation index when the disturbance event occurs in the shop processing process, and selecting a corresponding response strategy from immediate rescheduling, delayed rescheduling and neglecting the influence according to different influence levels; in the selection of the rescheduling method, the system state when disturbance occurs is updated, the system state is input into a random forest model to select the rescheduling method, and the scheduling scheme is updated according to the selection result until all tasks are processed.

Although various business process engine technologies based on finite state machines can solve the complexity and flexibility of process processing, the scheduling principle is that occupation of a specific time period of resources is taken as a computing core, when data is a task feature of high-frequency multidimensional association, the data acquisition time is long, the period of task processing is generally tens of seconds to minutes, and the scheduling scheme has low execution efficiency.

Disclosure of Invention

In order to solve the problem that the execution efficiency of a scheduling scheme is low due to long data acquisition time when the existing method processes task features associated with high frequency and multiple dimensions, the invention provides a reactive scheduling method, device, system and medium.

The aim of the invention is achieved by the following technical scheme:

the first aspect of the present invention provides a reactive scheduling method, comprising the steps of:

initializing a finite state machine, wherein the finite state machine comprises 4 layers of hierarchical structures including roots, stems, branches and leaves, and each layer of hierarchical structure comprises at least one level of nested structure;

in response to receiving at least one task data, acquiring an initial characteristic value of a finite state machine, generating a scheduling scheme corresponding to the task data by static scheduling based on the shortest task construction period of the task data as a constraint condition, wherein the task data comprises a plurality of operations;

acquiring a characteristic value of the finite state machine in response to at least one dynamic event occurring and state transition of at least one leaf-level state machine of the finite state machine in the process of executing the scheduling scheme;

judging whether the current state of the task data accords with the set step according to the characteristic value of the finite state machine;

and triggering dynamic scheduling and updating a scheduling scheme in response to the step that the current state of the task data does not accord with the setting.

The second aspect of the present invention discloses a reactive scheduling device, which comprises a memory and a controller which are sequentially in communication connection, wherein the memory is stored with a computer program, and the controller is used for reading the computer program and executing a reactive scheduling method described in the first aspect and any one of possible designs thereof.

A third aspect of the present invention discloses a reactive scheduling system comprising:

a finite state machine comprising a root, stem, branch, leaf total of 4 layers of hierarchies, each layer of hierarchies comprising at least one level of nesting structure,

the reactive scheduling device is the reactive scheduling device in the second aspect, and the reactive scheduling device is in information connection with the finite state machine.

A fourth aspect of the invention discloses a computer readable storage medium having instructions stored thereon which, when run on a computer, perform a reactive scheduling method as described in the first aspect and any of its possible designs.

Compared with the prior art, the invention has at least the following advantages and beneficial effects:

1. the invention is based on a finite scheduler, the finite scheduler is a strategy model based on an object library, the generation of a scheduling scheme is realized by acquiring the output parameters of the finite state machine, the data acquisition speed is greatly improved under the high-frequency multidimensional associated task feature use environment, the task processing period from tens of seconds to minutes is reduced to millisecond level, and the scheduling task execution efficiency is greatly improved.

2. In the scheduling process, the invention realizes the execution efficiency of the scheduling scheme by continuously updating the scheduling scheme.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a reactive scheduling method of the present invention;

FIG. 2 is a dimension diagram of dynamic event definition in the present invention;

fig. 3 is a schematic diagram of a finite state machine implementation process.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, based on the embodiments of the invention, which are apparent to those of ordinary skill in the art without inventive faculty, are intended to be within the scope of the invention.

In addition, the embodiments of the present invention and the features of the embodiments may be combined with each other without collision.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.

In the description of the present invention, it should be noted that, directions or positional relationships indicated by terms such as "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., are directions or positional relationships based on those shown in the drawings, or are directions or positional relationships conventionally put in use of the inventive product, or are directions or positional relationships conventionally understood by those skilled in the art, are merely for convenience of describing the present invention and for simplifying the description, and are not to indicate or imply that the apparatus or element to be referred to must have a specific direction, be constructed and operated in a specific direction, and thus should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and the like, are used merely to distinguish between descriptions and should not be construed as indicating or implying relative importance.

In the description of the present invention, it should also be noted that, unless explicitly specified and limited otherwise, the terms "disposed," "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.

The first aspect of the present invention discloses a reactive scheduling method, which may be performed by, but is not limited to, a reactive scheduling device. The reactive scheduling means may be software or a combination of software and hardware. The reactive scheduling device can be integrated in intelligent mobile terminals, tablet computers, local servers, cloud servers and other intelligent devices. Specifically, the scheduling method includes the following steps S01 to S06. It should be noted that, the steps S01 to S06 referred to herein are not limited to the sequence of the steps, which only plays a role of identifying the sign, and the sequence of the steps is mainly the sequence of the signal in the description.

Step S01, initializing a finite state machine, wherein the finite state machine comprises a root, trunk, branch and leaf 4 layers of hierarchical structures, and each layer of hierarchical structure comprises at least one level of nested structure.

The finite state machine in the scheme is an object library of a complex giant system, and comprises, but is not limited to, a virtual object class, a control object class, a data object class, a service object class and a connection object class which correspond to a physical entity, wherein the finite state machine is obtained from a system model in an automatic mode or is defined by human beings. For example, in MES systems, it may be customized by metadata, and in digital twin systems, it may be extracted by a data model. Meanwhile, a finite state machine is a policy model that decomposes the behavior of an object into manageable states, has a finite number of states, and can operate at any given time through inputs, transitioning from one state to another, or generating an output or an action. A finite state machine can only be in one state at any one time. The finite state machine forms, through message functions, the issuing, processing and reacting to events, i.e. incorporates event driven mechanisms. It has the basic characteristics of object oriented: encapsulation, inheritance, and polymorphism. The object is not limited to a finite state machine, and is hierarchical, and may be a state itself or a state machine. The external service realizes the call through the interface.

Specifically, the finite state machine comprises a root, stem, branch and leaf 4-layer hierarchical structure, and each layer hierarchical structure comprises at least one level of nested structure. Exemplary, operational schematic diagrams are shown in detail in fig. 3.

The root-level state machine is a service model, comprises an object of software in the conventional sense, and the object after the hardware discretization service encapsulation, and comprises the steps of comprehensively modeling the dimensions of multiple threads of geometry/physics/behavior/rules and constraints of a physical object, multiple layers of different granularities, multiple times of pushing evolution/real-time process/external interference and the like, and carrying out optimal configuration on solution items of the physical object.

The dry-stage state machine mainly comprises a scheduling state machine, a control state machine, a monitoring diagnosis state machine and a service management state machine.

The scheduling state machine comprises four main types of dendritic state machines, namely main resources, auxiliary resources, management classes and action objects, wherein the action objects are further divided into dendritic state machine purchasing, outsourcing and processing classes, the main resources are further divided into dendritic state machine processing centers, various industrial types and the like, the management classes are further divided into dendritic state machine process routes, time and the like, and the auxiliary resources are further divided into dendritic state machine design, tools, IEs, energy sources and the like. The control state machine comprises three main kinds of dendritic state machines of perception, transmission and parameters, wherein the perception is further divided into a dendritic state machine state type (in/idle/exiting) and an early warning type (normal/error reporting/dead). The transport class is further subdivided into the class of dendrimers AGV, PLC, ROBOT.

Monitoring the diagnostic state machine and the service management state machine are auxiliary roles and are not listed one by one.

The leaf-level state machine is the object individual at the extreme end of each class, and can also be one state of a certain branch state machine.

The service class task and the management class task are defined according to the characteristics of the task, wherein the granularity division, the hierarchy definition and the state specification of the state machine are defined.

Taking the structure of the intelligent manufacturing task state machine as an example, the structure is shown in table 1 in detail.

TABLE 1 Intelligent manufacturing task state machine

Step S02, in response to receiving at least one task data, obtaining an initial characteristic value of a finite state machine, generating a scheduling scheme corresponding to the task data by static scheduling based on the shortest task construction period of the task data as a constraint condition, wherein the task data comprises a plurality of operations.

The task data may be one task or a task set W composed of a plurality of task data. The task set W is a task sequence { W1, W2, W3 … } containing a plurality of task data, and each task data needs to be called by M pieces of equipment, and is formed by N steps at most, and W is a resource _ij Identification, where i e {1, M }, j e {1, N }, man-hour P _ij The time taken for processing in step j on apparatus i is represented by a natural number of 1 or more in each of M, N. The number pair (i, j) represents the operation of the step j on the equipment i, and the task period is the total execution time for completing all the operations in the task data, namely the task period comprises a plurality of working hours P corresponding to the operations _ij . The task construction period with the control target set as W is shortest, and the task construction period is equivalent to the shortest execution time of each task, namely the execution time C of the task construction period of a single task on the premise of not considering the cooperative optimization of multiple tasks _MAX Shortest.

The task data in the scheme is task processing step data, can be automatically acquired after targeted setting according to a service software platform, and can be defined and given by people. For example, in MES systems, it can be customized by worksheet release, in BPM systems, it can be customized by flow controllers, and in digital twin systems, it can be extracted by data models. The task data includes at least one of production class task data, service class task data, and management class task data. The production task data is process route data, the service task data is service flow data, and the management task data is management flow data.

The step is first scheduling after receiving task data, and the generated scheduling scheme is a pre-scheduling scheme. The first scheduling is static scheduling, defaults that all resource states are available, and the constraint condition is a task construction period C _MAX Minimum. This is a conventional resource-constrained project schedule (Resource Constrained Project Scheduling, RCPS), corresponding to the manufacturing field, namely, job-Shop. Conventional scheduling methods may be employed including the following: dispatch rules, filter bundle searches, local searches, scroll time and heuristic searches, etc., the algorithms used include, but are not limited to, the following: simulated annealing algorithm, particle swarm algorithm, tabu search algorithm, genetic algorithm, neural network, decision tree algorithm, clustering algorithm, etc.

After the scheduling scheme is generated, the scheduling scheme starts to be executed. During execution of the scheduling scheme, there is occurrence of a dynamic event which may operate on the working hour P of the step j on the device i _ij Causing an impact which in turn leads to the optimality of the previous scheduling schemes. Each step is completed according to the set requirement according to the normal sequence execution, and the state transition of the terminal leaf state machine is triggered, or various interferences occur in the execution process, and the method also belongs to the category of dynamic events; in addition, dynamic events can be defined from two dimensions of deterministic/uncertainty, high frequency/low frequency, as shown in FIG. 2.

The uncertainty factor of high frequency is mainly in the communication level of complex giant system, if it happens, the system has a great defect in robustness, and it is generally avoided to eliminate in the design stage.

The deterministic factors of high frequency are mainly represented in the control class, i.e. the part comprised by the control state machine described above.

The low-frequency determining factors mainly comprise preventive maintenance of a machine, shift of equipment, actual working time, increase of shift or time, change of processing priority, addition and deletion of orders, failure of equipment, delay of materials and the like, and the general scheduling method treats the latter six interferences as uncertainty interference due to lack of means for acquiring multi-dimensional environment information of all elements in real time. In the invention, the object library of the complex giant system is defined as a finite state machine, global information can be acquired in real time, and the interference of the classes is transferred in the millisecond-level period of occurrence by the state machine, so the complex giant system is treated according to certainty.

The uncertainty factor of low frequency is mainly represented by the fact that it exists in the traditional system, but is ignored because there is no effective sensing means and analysis method.

In contrast, if a dynamic event occurs during execution of the scheduling scheme, the process proceeds to step S03.

And step S03, responding to at least one dynamic event in the process of executing the scheduling scheme and the state transition of at least one leaf-level state machine of the finite state machine, and acquiring the characteristic value of the finite state machine.

In the process of executing the scheduling scheme, dynamic events occur and the state of a certain leaf level state machine is transferred, all objects are scanned on a current time node in real time, the characteristic values of all state machines are obtained in the millisecond range, and a judgment basis for scheduling scheme optimization is provided for the subsequent steps.

And step S04, judging whether the current state of the task data accords with the set step according to the characteristic value of the finite state machine. If yes, continuing to execute the current scheduling scheme; if the current state of the task data does not conform to the set step, the process proceeds to step S05.

And step S05, triggering dynamic scheduling and updating a scheduling scheme in response to the fact that the current state of the task data does not accord with the set step.

And when the current state of the task data does not accord with the set step, the static scheduling is changed into dynamic scheduling. Dynamic scheduling may be implemented in a variety of ways.

In an optimized scheme, the dynamic scheduling adopts a heuristic search method based on constraint guidance, and specifically comprises the steps of S051 to S052.

Step S051, calculating two time intervals of all operation pairs without sequence constraint, wherein the operation pairs comprise a first operation (i, j) on the device i and a second operation (i, k) on the device i, and the time intervals of the operation pairs comprise sigma (i, j) → (i, k) and sigma (i, k) → (i, j). The time interval σ (i, j) → (i, k) between (i, j) and (i, k) is operated on device i, and the time interval σ (i, k) → (i, j) between (i, k) and (i, j) is operated on device i. For example, if 10 operations are included in a task data, if there is no sequence constraint among 3 operations, 3 operation times constitute 3 operation pairs.

For each operation (i, j), the earliest possible start time and the latest possible completion time of that operation need to be calculated. After all the time windows have been calculated, the time windows on each device are compared with each other. If the time windows of two operations on a given device do not overlap, then a priority relationship between the two operations may be determined. In any one possible schedule, the work corresponding to an earlier time window must be ordered after the operation corresponding to a later time window. In practice, there may be a precedence relationship even if two time windows overlap. Let S _ij ’(S _ij ") indicates the earliest (latest) possible start time of operation (i, j), C _ij ’(C _ij ") represents the earliest (latest) possible completion time of operation (i, j), these assumptions being built under a set of priority constraints. The earliest possible start time (i.e., S') of operation (i, j) may be considered the local commit time of this operation, taken by R _ij A representation; the latest possible completion time (i.e., C ") can be considered as a local time period, using D _ij And (3) representing. The time interval defined between operations (i, j) and (i, k) on device i is:

σ _{(i,j)→(i,k)} ＝S _ik "-C _ij ’

or alternatively

σ _{(i,j)→(i,k)} ＝C _ik "-S _ij ’-p _ij -p _ik

Or alternatively

σ _{(i,j)→(i,k)} ＝d _ik -r _ik -p _ij -p _ik

If it is

σ _{(i,j)→(i,k)} <0

Then there is no viable schedule under the constraint of the current set of priority constraints where operation (i, j) occurs prior to operation (i, k) on device i, so this priority relationship can be exploited which requires operation (i, k) to precede operation (i, j).

In this step of initialization, all time windows are compared to each other and all available precedence relationships are inserted into the extraction graph. Because of these priority constraints, the time window for each operation is adjustable (shrinking, i.e., requiring recalculation of the operation commit time and the time period).

Constraint-guided heuristic searches always attempt to insert one priority constraint (extraction arc) before another in each step. After inserting a priority constraint, all time windows need to be recalculated. For each pair of operations that must be done on the same device, it must be proven that either:

case 1. Sigma (i, j) → (i, k) > 0 and sigma (i, k) → (i, j) <0,

case 2. Sigma (i, k) → (i, j) > 0 and sigma (i, j) → (i, k) <0,

case 3.σ (i, j) → (i, k) <0 and σ (i, k) → (i, j) <0,

case 4.σ (i, j) → (i, k) > 0 and σ (i, k) → (i, j) > 0.

And step S052, performing constraint optimization according to two time intervals until the optimization meets a preset condition, calculating the sequence flexibility of operation pairs except operation without sequence constraint, and preferentially scheduling operation pairs with lower sequence flexibility.

Specifically, if the two time intervals of any one operation pair meet the condition 1 or the condition 2, a new constraint is added, and then the step S051 is performed;

if the two time intervals of any one operation pair meet the condition 3, the previous step is traced back, otherwise, the step P1 is entered.

Step P1, if no operation pair has two time intervals sigma (i, j) → (i, k) not less than 0 and sigma (i, k) → (i, j) > not less than 0, the preset condition is satisfied.

After the preset condition is met, a solution is found, and stopping is carried out; otherwise, the sequential flexibility of operation pairs other than the operation without the sequential constraint is calculated ((i, j) (i, k)), and the pair with the smallest value is selected; if σ (i, j) → (i, k) → (i, j), then operation (i, k) must immediately follow (i, j), otherwise operation (i, j) must immediately follow (i, k), i.e., operation pairs with lower order flexibility schedule preferentially.

That is, if more than one pair of operations satisfies the 4 th case, then the order flexibility φ ((i, j) (i, k)) is judged, and the pair of operations with lower flexibility is preferentially scheduled for processing, wherein:

φ((i,j)(i,k))＝

after all optimization is completed, the following steps are obtained:

C _MAX ＝d _mn 。

step S06, update Process P _ij And updates the scheduling scheme.

For the most main uncertainty factor bringing about the actual man-hour deviation, a man-hour deviation factor (Working Time Deviation Factor, WTDF) is defined, whereby the time deviation brought about is P _ij ' the time offset is a five-tuple:

WTDF＝{L,F,B,H,A}

the meaning of the parameters is shown in table 2, wherein the deviation directions "+", "-" distribution represents increasing and decreasing working time:

TABLE 2 man-hour deviation factor

These factors will in fact be specific to P _ij With superimposed effects, analysis of historical data results in some very typical density function whose values are exponentially related to the system state at which the operation occurs. In a specific state:

P _ij ’＝∑{L,F,B,H,A}

then at any step, P can be calculated by the following formula _ij Make corrections and continue from iteration:

P _ij ＝P _ij +P _ij ’

in the whole method, P is dynamically scheduled for the first time _ij Has been corrected iteratively, P _ij The 'initial value' is non-zero and non-null and is given after training from historical data samples.

The scheme is based on the generation and optimization of a finite state machine, the characteristic state values can be read in a millisecond period to obtain global information of a scheduling class, a control class, a monitoring diagnosis class and a service management class, and the global information is combined with historical data and current data to perform fusion, training and analysis, so that whether the current execution is correct or not is judged, an optimization suggestion of the current resource configuration and a decision strategy of continuous steady-state optimization are given, and an execution instruction is transmitted to a controlled object.

The intelligent scheduling method comprises 1) judging whether the current state of task data accords with the set processing step; 2) Correcting and optimizing the current execution parameters by combining the historical data; 3) Through fusion, training and analysis of historical data and current data, regular change of execution parameters is proposed, and resource allocation optimization suggestions are given; 4) The data of the previous three types are used as sample training, and continuous and stable production/service/management strategy situation is judged.

By fusion analysis based on historical data and current data, continuous steady-state optimization is given by minimizing objective functions related to the construction period.

The second aspect of the present invention discloses a reactive scheduling device, which comprises a memory and a controller which are sequentially in communication connection, wherein the memory is stored with a computer program, and the controller is used for reading the computer program and executing a reactive scheduling method described in the first aspect and any one of possible designs thereof. By way of specific example, the Memory may include, but is not limited to, random-Access Memory (RAM), read-Only Memory (ROM), flash Memory (Flash Memory), first-in first-out Memory (First Input First Output, FIFO), and/or first-out Memory (First Input Last Output, FILO), etc.; the controller may not be limited to use with a microcontroller model STM32F105 series. In addition, the computer device may include, but is not limited to, a power supply unit, a display screen, and other necessary components.

the finite state machine comprises a root, trunk, branch and leaf 4-layer hierarchical structure, and each layer hierarchical structure comprises at least one level of nested structure, wherein the architecture and the operation principle of the finite state machine are shown in fig. 3, fig. 2 and the first aspect, and are not described in detail herein.

The reactive scheduling system obtains multi-dimensional environment information of all elements, and the applicable industrial communication with the execution unit through a certain industrial communication protocol comprises the following types: PROFIBUS, CAN, CIP, P-Net, etc., industrial Ethernet, etherNet/IP, PROFINET, modbus/TCP, HSE, SERCOS, etc., or Wifi, loRa, NB-IoT, 5G, etc. After the open-loop optimization problem of a finite time domain is solved on line, the first element of the control sequence obtained by solving is acted on the controlled object, so that the control closed loop is realized. A fourth aspect of the invention discloses a computer readable storage medium having instructions stored thereon which, when run on a computer, perform a reactive scheduling method as described in the first aspect and any of its possible designs.

Although the present invention has been described with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described, or equivalents may be substituted for elements thereof, and any modifications, equivalents, improvements and changes may be made without departing from the spirit and principles of the present invention.

Claims

1. A reactive scheduling method, comprising the steps of:

2. The reactive scheduling method of claim 1, wherein the dynamic scheduling employs a constraint-based guided heuristic search method.

3. A reactive scheduling method according to claim 1, wherein the dynamic scheduling comprises the steps of:

A. calculating two time intervals of an operation pair without sequential constraint, wherein the operation pair comprises a first operation (i, j) on the device i and a second operation (i, k) on the device, and the two time intervals are respectively a time interval sigma (i, j) → (i, k) between the operations (i, j) and (i, k) and a time interval sigma (i, k) → (i, j) between the operations (i, k) and (i, j);

B. and performing constraint optimization according to the two time intervals until the optimization meets a preset condition, calculating the sequential flexibility of operation pairs except for operation without sequential constraint, and preferentially scheduling operation pairs with lower sequential flexibility.

4. A reactive scheduling method according to claim 3, characterized in that: and performing constraint optimization according to the two time intervals until the optimization meets the preset condition, wherein the constraint optimization comprises the following steps:

if the two time intervals of any operation pair meet the condition 1 or the condition 2, adding a new constraint condition, and then entering a step A, wherein the condition 1 is sigma (i, j) → (i, k) > 0 and sigma (i, k) → (i, j) <0, and the condition 2 is sigma (i, k) → (i, j) > 0 and sigma (i, j) → (i, k) < 0;

if the two time intervals of any one operation pair meet the condition 3, backtracking to the previous step, otherwise, entering the step P1, wherein the condition 3 is sigma (i, j) → (i, k) <0 and sigma (i, k) → (i, j) < 0;

and P1, if no operation pair has two time intervals sigma (i, j) → (i, k) not less than 0 and sigma (i, k) → (i, j) > not less than 0, meeting the preset condition.

5. A reactive scheduling method according to claim 3, characterized in that: the task period of the task data comprises a plurality of working hours P corresponding to the operation _ij ，

Scheduling operation pairs with lower order flexibility preferentially, further comprising:

update man-hour P _ij And updates the scheduling scheme.

6. The reactive scheduling method of claim 5, wherein: the update man-hour P _ij Comprising the following steps:

determining a time deviation as P according to the learning factor, the aging factor, the continuity factor, the preference factor and the load factor _ij ’；

According to the time deviation P _ij ' P time to man hour _ij And carrying out correction and update.

7. A reactive scheduling method according to claim 1, characterized in that: the finite state machine includes:

a root state machine, the root state machine being a service model;

a dry-stage state machine comprising a scheduling state machine, a control state machine, a monitoring diagnostic state machine, and a service management state machine;

the branch state machine comprises an action object type state machine, a resource type state machine, a management type state machine, a perception type state machine, a transmission type state machine and a parameter type state machine, wherein the action object type state machine, the resource type state machine and the management type state machine correspond to the scheduling state machine;

and the leaf level state machine is a state of a certain branch level state machine.

8. A reactive scheduling apparatus comprising a memory and a controller in communication with each other in sequence, the memory having a computer program stored thereon, characterized by: the controller is configured to read the computer program and execute a reactive scheduling method according to any one of claims 1 to 7.

9. A reactive scheduling system, comprising:

a finite state machine comprising a root, stem, branch, leaf total of 4 layers of hierarchical structures, each layer of hierarchical structure comprising at least one level of nested structure;

a reactive scheduling device, the reactive scheduling device being the reactive scheduling device of claim 8, and the reactive scheduling device being in information connection with the finite state machine.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: a reactive scheduling method according to any one of claims 1 to 7 when said instructions are run on a computer.