CN106773711A - The hybrid tasks scheduling method and model of a kind of railway locomotive operation steerable system - Google Patents

Abstract

The invention provides a mixed task scheduling method and model of a railway locomotive operation and control system. The scheduling method of the present invention takes a frame period as the basic scheduling unit, including preprocessing; for periodic real-time tasks, the scheduling sequence is sorted based on the table-driven secondary priority rules; for non-periodic real-time tasks, the heuristic search strategy and The method of fuzzy control is arranged; then the recovery of time slices and the collection of scheduling results and feedback data are carried out, and the remaining execution time is judged; if the cycle time is not used up, non-real-time tasks are executed. The hybrid task scheduling method and model of the present invention can greatly reduce system overhead, and can more flexibly deal with various changes in the system execution process, and can reduce the influence of system uncertainty on scheduling.

Description

A hybrid task scheduling method and model for railway locomotive operation and control system

technical field

The invention relates to the field of railway locomotive operation and control systems, in particular to a mixed task scheduling method and model of the railway locomotive operation and control system.

Background technique

The railway locomotive operation control system is based on the train energy-saving control mechanism and operation model, and integrates LKJ communication, locomotive traction calculation and kinematic equations, optimization calculation, fuel consumption calculation, adaptive control, display and other auxiliary functions. system. It can simulate the dynamic process of train operation, perceive the changes of train, line and environment according to the input parameters of LKJ, and calculate the speed-distance curve and gear-distance optimal energy-saving maneuvering strategy suitable for the locomotive, and in a very short time Optimal control is given within the time interval. Because the system runs in the environment of real-time vehicle control, the system tasks must be completed in real time and accurately. For the reasonable scheduling of tasks, reducing system latency plays a vital role in ensuring the normal and stable operation of the system.

The railway locomotive operation and control system is a typical complex optimization control system, and its task model is a mixed task set composed of periodic real-time tasks, non-periodic real-time tasks and non-real-time tasks. Although the computer and embedded development technology are developing rapidly at present, and the processing ability of hardware equipment is getting stronger and stronger, computing resources are still very precious. The problem of optimizing real-time task scheduling of control systems has always been a hot and difficult point in the field of computer research.

Existing scheduling methods for mixed task sets can be classified into table-driven, priority-driven, search-driven, etc. according to different driving modes.

The table-driven real-time scheduling scheme is a scheduling scheme based on the static scheduling table algorithm, which is widely used in high-precision systems such as space shuttle software systems and missile navigation systems. This scheduling scheme has low scheduling overhead and high operation precision, and has a strong guarantee for the scheduling of hard real-time tasks. But its disadvantage is that it is not flexible enough, and it can only complete the scheduling of periodic tasks, and it needs to accurately locate the arrival time, ready time and worst execution time of each task in a cycle. In the task scheduling model of railway locomotive operation and control system, there are both periodic and non-periodic real-time tasks. The non-periodic tasks arrive at random, and the periodic tasks will change dynamically with the system state. Therefore, this scheduling scheme is not suitable for this uncertain scheduling environment.

The priority-driven real-time scheduling scheme can be subdivided into static priority scheduling and dynamic priority scheduling. It is a very common scheduling scheme, which is applied in the classic real-time operating system VxWorks, μC/OS, and the scheduling scheme has strong flexibility. But its flexibility is based on frequent task switching, which consumes a lot of system resources. In the railway locomotive operation and control system, the priority of the control task is obviously higher than that of the optimization task, and the periodic completion of the control process is the most fundamental requirement of the system. If the priority preemption scheduling scheme is adopted, it will cause: first, periodic control tasks cannot be completed on time due to priority inversion and other reasons; second, frequent context switching leads to low resource utilization and system instability under high load conditions . Therefore, the priority-driven scheduling scheme cannot solve the scheduling problem of the railway locomotive operation and control system.

Common search-driven real-time scheduling schemes include myopic algorithms and thrifty algorithms. This scheduling scheme has strong flexibility, but the disadvantage is that the search process has a large overhead. For the railway locomotive operation and control system, using the search method to simultaneously schedule periodic tasks and non-periodic tasks will greatly reduce the periodic task scheduling. success rate, which seriously affects system performance.

Therefore, the various scheduling schemes currently available cannot meet the scheduling requirements in the complex optimization control system. It is very important to design a scheduling scheme that can satisfy the real-time control and optimization stability at the same time and can give full play to the maximum performance of the processor.

Contents of the invention

The purpose of the present invention is to provide a mixed task scheduling method and model of a railway locomotive operation and control system. The task set of the railway locomotive system is a mixed task set composed of periodic tasks, non-periodic tasks and non-real-time tasks. Among them, periodic tasks are the core of the entire control process, with the strictest time limit constraints, diversified execution cycles, strong timing, and many resource sharing and mutual exclusion situations; non-periodic tasks are to maintain system stability, improve system performance and system optimization An important part of the system, it has the characteristics of random arrival, rich types and forms of tasks, and a wide range of priority changes; non-real-time tasks are the guarantee for system maintenance and debugging, and do not have real-time processing requirements, but the calculation takes the most time. Considering the scheduling requirements of different types of tasks, the invention can clearly describe the time constraints of the tasks themselves, the timing constraints and resource constraints between tasks, accurately reflect the characteristics of task execution in actual scenarios, and realize the reasonable scheduling of mixed tasks.

The present invention realizes through following technical scheme:

A mixed task scheduling method for a railway locomotive operation and control system, characterized in that the mixed task includes: periodic real-time tasks, non-periodic real-time tasks and non-real-time tasks; the mixed task scheduling method takes one frame period as a basic Scheduling unit, the hybrid task scheduling method includes the following steps:

(1) When the frame period starts, the hybrid task scheduling method performs scheduling preparation, and the scheduling preparation includes at least task switching and resource checking;

(2) For periodic real-time tasks, apply the table-driven two-level priority rule to sort the scheduling sequence, and schedule periodic real-time tasks according to the sequence;

(3) For non-periodic real-time tasks, apply the method based on heuristic search strategy and fuzzy control to arrange non-periodic real-time tasks, and schedule non-periodic real-time tasks according to the sequence;

(4) Reclaim unused time slices in the current cycle and collect scheduling results and feedback data to determine the remaining execution time;

(5) If the current cycle time is not used up, execute the non-real-time task;

(6) At the end of the frame period, if the execution of the non-periodic real-time task or non-real-time task has not ended, the periodic real-time task will be executed at the beginning of the next frame period, and the unfinished non-periodic real-time task or non-real-time task will be sealed. task, to ensure the real-time performance of periodic real-time tasks.

Further, the table-driven secondary priority rule is a non-preemptive scheduling method proposed for the predictability of periodic real-time tasks in the locomotive operation and control system, and the secondary priority is divided into groups Priorities and instance priorities are first sorted by group priorities in the step (2), and then by instance priorities to finally generate a scheduling table.

Further, the described heuristic-based search strategy introduces a method of fuzzy control, and proposes a new valuation function:

H(n,t)=h(n,t)*n _st

n _st =(n _d -t)-(n _c -e(n,t))

where H(n,t) is the valuation function, h(n,t) is the dynamic threshold coefficient of aperiodic real-time task n at time t, n _st is the remaining idle time of aperiodic real-time task n at time t, n _d is The deadline of task n, n _c is the worst execution time, e(n,t) is the time that task n has been executed at time t.

Further, in the table-based two-level priority rule, the design rule of the group priority is: first set several task groups according to the order of deadlines of periodic real-time tasks in the task set from small to large , and then according to the timing constraints between tasks, determine the predecessor tasks of each periodic task and adjust the corresponding task group priority; the instance priority is the sorting rule of the internal tasks of the task group, and the sorting adopts the minimum and worst execution time rule .

Further, in the evaluation function, the completion rate of non-periodic real-time tasks is also used as a fuzzy input parameter, and the corresponding relationship between input parameters and threshold coefficients is shown in the following table:

The task completion rate is defined as follows: n _cr (t)=e(n,t)/n _c .

Further, in the step (5), for non-real-time tasks, the scheduling sequence is scheduled according to FIFO rules.

On the other hand, the present invention provides a mixed task dispatching model of a railway locomotive operation and control system, which is characterized in that the dispatching model is divided into two parts, which are respectively a dispatching unit and an execution unit, and the dispatching unit adopts claim 1 The method described in any one of -6 schedules the mixed tasks.

Further, the scheduling unit includes a task collector, a real-time scheduler and a task integration unit. During the operation of the locomotive operation and control system, real-time tasks are continuously generated, and the task collector is used to centralize the generated tasks Processing, when each clock signal arrives, it is submitted to the real-time scheduler; the real-time scheduler calls the periodic real-time task scheduling algorithm and the non-periodic real-time task scheduling algorithm respectively for different types of tasks before the arrival of the next cycle, and generates corresponding A periodic real-time task scheduling sequence and an aperiodic real-time task scheduling sequence; finally, the task integration unit combines the two sequences to generate a unified scheduling sequence, which is submitted to the task execution unit; the execution unit executes the task according to the scheduling sequence Scheduling, and feeding back the scheduling results and intermediate state information to the real-time scheduler, which collects and saves them.

The dynamic threshold coefficient mentioned in the present invention refers to that in the present invention, through the fuzzy control method, the estimated value is multiplied by a parameter. This parameter is determined by the fuzzy control method and has different values in different states. It is called dynamic threshold coefficient here.

The beneficial effect of adopting the above-mentioned technical scheme is:

(1) Compared with the priority scheduling algorithm, the algorithm designed by the present invention for periodic real-time tasks can greatly reduce the system overhead, including the overhead of task switching and the overhead of the processor executing the scheduling algorithm; compared to the static table-driven Scheduling algorithm, which can more flexibly deal with various changes in the system execution process. Moreover, after the system runs stably, the modification overhead required by the scheduling table generated by the method of the present invention is very small.

(2) The present invention has adopted a kind of evaluation function of brand-new design, and application fuzzy control method is just fuzzy to the parameter of evaluation function, on the basis of guaranteeing to minimize aperiodic real-time task response time, mainly has three advantages: the first , the fuzzy application can reduce the impact of system uncertainty on scheduling. In actual operation, the state of the system cannot be precisely predicted. Different from the task parameters specified in the scheduling model, the time parameters of the task will vary slightly in different states of the system. difference, and fuzzy technology can weaken the impact of this difference; second, fuzzy theory can synthesize the cumulative impact of multiple factors on the current state of the task, such as task idle time and completion rate in the design of the present invention; third, fuzzy threshold It can reduce the jitter of the scheduling algorithm under marginal conditions. If the judgment is a fixed value, the system is prone to thrashing under boundary conditions.

Description of drawings

Fig. 1 is the flow chart of the hybrid task dispatching method of the railway locomotive operation control system of the present invention;

Fig. 2 is a task scheduling sequence diagram within a frame period of the scheduling method of the present invention;

Fig. 3A and Fig. 3B are the relationship diagrams of idle time and completion rate membership function when the present invention is performing aperiodic real-time tasks;

Fig. 4 is a schematic diagram of a mixed task scheduling model of the railway locomotive operation and control system of the present invention.

detailed description

In order to make the present invention clearer, the present invention will be described in detail below in conjunction with the accompanying drawings.

First of all, in order to clearly describe various task sets in the operation and control system of railway locomotives, the present invention defines task models as follows:

(1) Periodic real-time task: Periodic real-time task is a real-time task with fixed period and cyclic execution. Periodic real-time tasks can be expressed as a five-tuple:

p=(a,r,t,d,c)

The elements of the tuple are: the arrival time of the periodic real-time task, the ready time of the periodic real-time task, the execution period of the periodic real-time task, the deadline for the execution of the periodic real-time task, and the time period of the periodic real-time task on a single processor Worst execution time.

(2) Aperiodic real-time task: Aperiodic real-time task is a real-time task with random arrival time, and its execution time is uncertain. Aperiodic real-time task can be expressed as a quadruple:

n=(a,r,d,c)

The elements of the tuple are: the arrival time of the aperiodic real-time task, the ready time of the aperiodic real-time task, the deadline for the execution of the aperiodic real-time task, and the worst execution time of the aperiodic real-time task on a single processor .

(3) Non-real-time tasks: Non-real-time tasks are tasks other than real-time tasks in the system, represented by a quadruple:

o=(a,r,s,l)

The elements of the tuple are: the arrival time of the non-real-time task, the ready time of the non-real-time task, the start execution time of the non-real-time task, and the delay time of the non-real-time task.

This embodiment provides a mixed task scheduling method for a railway locomotive operation and control system. The specific implementation process is shown in Figure 1, which includes:

Step S101, at the beginning of the frame period, the scheduling method performs preparatory work, including task switching, resource checking and so on.

The scheduling method of the present invention takes a frame cycle as a basic scheduling unit, wherein the frame cycle is a reference cycle of system operation, and describes the minimum unit of system cycle division. The task period in the periodic real-time task set is described by the multiple relationship of the frame period.

As shown in Figure 2, at the beginning of a frame period, the scheduling method needs to perform task switching at the end of the previous period. Usually, the tasks that may be executing at the end of the last frame period are aperiodic real-time tasks or non-real-time tasks. The above two types of tasks have a lower priority than periodic real-time tasks in the railway locomotive operation and control system of the present invention. At the same time, the algorithm needs to check resources to ensure that the tasks in the previous cycle have released related resources.

Step S102, for the periodic real-time tasks, apply the table-driven two-level priority rule to sort the scheduling sequence, and schedule the periodic real-time tasks according to the sequence.

The periodic real-time task is the most important part of the mixed tasks of the locomotive operation and control system, and it usually undertakes the function of real-time control, so it should also be assigned a higher priority. Although the locomotive operation and control system is a complex optimal control system, once the system parameters are determined, the time parameters of periodic real-time tasks are also predictable. At the same time, in different application scenarios and environments, the operating status of the system is different, and the periodic real-time tasks will also change dynamically.

In order to make full use of the predictability of periodic real-time tasks in the control process, and at the same time deal with the uncertainty in the execution process in a targeted manner, the present invention designs a non-preemptive scheduling based on table-driven two-level priority rules algorithm. The second-level scheduling rule examines the deadline of each task and the relationship of timing constraints between tasks. During specific implementation, each periodic real-time task has two priorities, which are group priority and instance priority. When arranging the scheduling sequence of periodic real-time tasks, it is first sorted by group priority, then by instance priority, and finally generates a scheduling sequence.

The specific generation rules are as follows: (1) In order to make full use of the utilization rate of the processor, first set up several task groups according to the order of deadlines in the task set from small to large. According to the period of the task, assign each periodic real-time task to the task group, and the periodic tasks in the same group have the same group priority, that is, have the same degree of importance; (2) According to the timing constraints between tasks, determine each The predecessor task of a periodic task. If the group priority of the predecessor task of the task is greater than the task, then keep the group priorities of the current two tasks unchanged; otherwise, reassign the predecessor task of the task to the priority group of the task; (3) For the priority order of tasks in the group, the smallest worst execution time is used as the criterion, that is, the task with the smaller worst execution time has a higher instance priority. After setting the instance priorities of all tasks, adjust the priority order of task instances according to the relationship of timing constraints, so that the timing constraints can be satisfied; (4) Determine the execution order of tasks, first sort according to the group priority, and then for Tasks with the same group priority are sorted according to the instance priority from high to low, and finally a periodic task schedule is generated.

The two-level priority rules take into account the dynamic characteristics of computational overhead and task deadlines. By utilizing the predictability of periodic real-time tasks, some preprocessing can be done on the parameters of the schedule during the system initialization phase, such as the setting of the priority group and the pre-generation of the schedule. In the actual system design process, because periodic real-time tasks have certain local time characteristics, that is, the set of periodic real-time tasks is stable within a period of time. Therefore, after the schedule table is generated for the first time, each subsequent generation can be modified on the basis of the previous generation, and the required modification cost is very small.

Step S103, for the non-periodic real-time tasks, apply the method based on heuristic search strategy and fuzzy control, arrange the non-periodic real-time tasks, and schedule the tasks according to the sequence.

Aperiodic real-time tasks are characterized by random arrival, and the system cannot predict the arrival time of aperiodic real-time tasks. The scheduling goal of aperiodic real-time tasks is to maximize the completion success rate of aperiodic real-time tasks and minimize the response time of aperiodic real-time tasks on the premise of ensuring that the periodic real-time task scheduling meets the deadline.

Integrating the characteristics of system scheduling objectives and mixed task sets, the aperiodic real-time task scheduling problem can be regarded as a constrained path search problem, that is, to reasonably arrange aperiodic real-time tasks in the remaining time of periodic real-time tasks, so that The success rate of aperiodic real-time task scheduling and the utilization rate of the processor both reach ideal values. The invention designs an aperiodic real-time task scheduling algorithm based on a heuristic search strategy, and uses the idea of fuzzy control to reduce the instability of the algorithm.

The heuristic search strategy process is: for a state space E composed of ordered n-tuples, given a constraint set D and the optimal objective function C(x) about the components in the n-tuples, by performing Evaluation, quickly get an n-tuple in E that satisfies all the constraints in constraint D, and this n-tuple is the optimal solution to the heuristic search problem. In the present invention, the n-tuple corresponds to the scheduling sequence of non-periodic real-time tasks, and the tuple that meets the requirements of constraint combination D belongs to the schedulable sequence. If the tuple makes the optimal objective function C(x) reach the extreme value at the same time, Then the tuple is the optimal schedulable sequence. In the specific implementation, the algorithm adopts the greedy method, and selects the task with the highest estimated value from the task set through the evaluation function for sequence sorting. It can be proved that if the evaluation function is designed reasonably, satisfies the requirement of monotonicity, and the task set is schedulable, the schedulable nodes obtained by this greedy method must be the optimal schedulable nodes.

The key of the heuristic search strategy is the design of the evaluation function. This invention introduces the idea of fuzzy control and proposes a new evaluation function design method. The specific design process is as follows:

(1) Define the idle time of the aperiodic real-time task as:

n _st =(n _d -t)-(n _c -e(n,t))

Define the completion rate of an aperiodic real-time task:

n _cr (t)=e(n,t)/n _c

Among them, n _st is the remaining idle time of aperiodic real-time task n at time t, n _d is the deadline of task n, n _c is the worst execution time, e(n,t) is the task n has been executed at time t time.

If the idle time of an aperiodic real-time task is smaller, it is more urgent and should be scheduled first; if the completion rate of an aperiodic real-time task is higher, then it needs to stay in the processor to complete the rest The part that needs to be scheduled first.

(2) The present invention designs a fuzzy threshold rule, which can determine the importance of non-periodic real-time tasks based on their idle time and completion rate. The membership functions of idle time and completion rate of aperiodic real-time tasks are shown in Figure 3A and Figure 3B.

Three language fuzzy sets are used to correspond to different inputs, idle time: short, medium, long, and completion rate: low, medium, high. In order to avoid computing complexity in the process of real-time control, the present invention adopts the fuzzy threshold coefficient of specific output representative task, and in the process of real-time calculation, the time complexity of obtaining the fuzzy threshold of aperiodic real-time task by look-up table is O( 1). The corresponding relationship table of the threshold coefficient of the fuzzy input is shown in the following table:

The first row in the above relationship table indicates the idle time level of the aperiodic real-time task, that is, the urgency of the current task, and the first column indicates the completion rate of the aperiodic real-time task, that is, the urgency of the current task to continue to be completed. During specific implementation, the idle time and completion rate of all currently unscheduled aperiodic real-time tasks at time t are calculated. Calculate the corresponding membership degree according to the idle time and completion rate, and then calculate the dynamic threshold coefficient proportionally according to the corresponding relationship in the table.

(3) The present invention defines the valuation function as follows:

H(n,t)=h(n,t)*n _st

Among them, H(n,t) is the evaluation function of non-periodic real-time task at time t, h(n,t) is the dynamic threshold of non-periodic real-time task n at time t, the calculation method of this value is according to the evaluation function design process ( 2) The expression in the step is calculated.

Through the above steps, the scheduling sequence of non-periodic real-time tasks can be realized, and the scheduling process of tasks can be performed according to the sequence.

Step S104, the post-processing is to recover time slices, collect scheduling results and feedback data, and determine the remaining execution time.

The task scheduling sequence of the scheduling algorithm of the present invention within a frame period is shown in FIG. 2 . Steps S102 and S103 can complete the scheduling sequence sorting of periodic real-time tasks and non-periodic real-time tasks in the mixed task set, and complete the scheduling process. Next, the post-processing of the algorithm is carried out to recover the time slices in the frame period and collect the scheduling results and feedback data. The above work is to make full use of the time resources in the frame period and make preparations for further scheduling of non-real-time tasks.

Step S105, if the cycle time is not used up, then execute the non-real-time tasks, and the scheduling order of them is scheduled according to the FIFO rule.

Considering the post-processing result of step S104, if there are still time slices left in the current frame period, select a non-real-time task from the mixed task set for execution, and the selected rule is the FIFO rule. In this way, time slice resources within the frame period are fully utilized.

If there is no non-real-time task in the current mixed task set, it will enter the idle state.

Step S106, at the end of the frame period, if the aperiodic real-time task or the non-real-time task has not finished executing, it will be preempted by the periodic real-time task of the next frame period, so as to ensure the real-time performance of the periodic real-time task.

Through steps S101 to S105, all scheduled executions of mixed task sets in the system can be realized. At the end of the period, there may be non-periodic real-time tasks or non-real-time tasks that are still executing. At this time, the priority of these two types of tasks is lower than that of periodic real-time tasks, and they will be replaced by the next frame period in the next frame period. The real-time tasks within the frame period are preempted, thereby ensuring the real-time performance of the periodic real-time tasks.

The above S101 to S106 are the execution steps of the task scheduling algorithm in one frame period, and the scheduling of the mixed task set is still executed in accordance with the order of the algorithm in the next frame period.

The hybrid task scheduling model of the railway locomotive operating system proposed by the present invention is shown in Figure 4, including a task scheduling unit and an execution unit. The scheduling unit includes task collector, real-time scheduler and task integration unit. During the operation of the system, real-time tasks are continuously generated, and their sources include human-computer interaction, external signals, system feedback, etc. After the tasks are initialized, they are centrally processed by the task collector, and submitted to the real-time scheduler when each clock signal arrives; the real-time scheduler integrates these tasks before the arrival of the next cycle, and calls the periodic real-time task scheduling algorithm for different types of tasks respectively And aperiodic real-time task scheduling algorithm, correspondingly generate periodic real-time task scheduling sequence and aperiodic real-time task scheduling sequence; finally, combine the two sequences through the task integration module to generate a unified scheduling sequence, and submit it to the task execution unit. The execution unit schedules the tasks according to the instructions of the scheduling sequence, and feeds back the scheduling results and some intermediate states to the real-time scheduler, which collects and saves them.

It can be seen from the technical solution of the above invention that the present invention designs a non-preemptive scheduling algorithm based on table-driven two-level priority rules for periodic real-time tasks, which has low system overhead and takes into account the advantages of priority scheduling. Flexibility: A non-periodic real-time task scheduling algorithm based on a heuristic search strategy is designed for non-periodic real-time tasks, using feasibility analysis for task sets, limiting the recursion depth of the heuristic search algorithm, and adjusting the parameters of the heuristic evaluation function Fuzzification reduces the jitter generated during scheduling, and the algorithm can better handle non-periodic real-time task scheduling; for non-real-time tasks, the classic FIFO rule is used to make full use of the system's time resources and computing resources, ensuring maximum Non-real-time tasks are scheduled and executed in a timely manner. The method and model designed by the invention can effectively solve the mixed task scheduling requirements of the complex railway locomotive operation and control system.

Although the principle of the present invention has been described in detail above in conjunction with the preferred embodiments of the present invention, those skilled in the art should understand that the above embodiments are only explanations for the exemplary implementation of the present invention, and are not intended to limit the scope of the present invention. The details in the embodiments do not constitute a limitation to the scope of the present invention. Without departing from the spirit and scope of the present invention, any obvious changes such as equivalent transformations and simple replacements based on the technical solutions of the present invention fall within the scope of the present invention. within the scope of protection.

Claims (8)

1. A mixed task scheduling method of a railway locomotive operation control system, characterized in that, said mixed task comprises: periodic real-time tasks, non-periodic real-time tasks and non-real-time tasks; said mixed task scheduling method uses a frame period As a basic scheduling unit, the hybrid task scheduling method includes the following steps: (1) When the frame period starts, the hybrid task scheduling method performs scheduling preparation, and the scheduling preparation includes at least task switching and resource checking; (2) For periodic real-time tasks, apply the table-driven two-level priority rule to sort the scheduling sequence, and schedule periodic real-time tasks according to the sequence; (3) For non-periodic real-time tasks, apply the method based on heuristic search strategy and fuzzy control to arrange non-periodic real-time tasks, and schedule non-periodic real-time tasks according to the sequence; (4) Reclaim unused time slices in the current cycle and collect scheduling results and feedback data to determine the remaining execution time; (5) If the current cycle time is not used up, execute the non-real-time task; (6) At the end of the frame period, if the execution of the non-periodic real-time task or non-real-time task has not ended, the periodic real-time task will be executed at the beginning of the next frame period, and the unfinished non-periodic real-time task or non-real-time task will be sealed. task, to ensure the real-time performance of periodic real-time tasks. 2. The hybrid task scheduling method of a kind of railway locomotive operation and control system according to claim 1, characterized in that, said table-driven secondary priority rule is for periodic real-time tasks in the locomotive operation and control system A kind of non-preemptive scheduling method proposed by the predictability, described secondary priority is divided into group priority and instance priority, in described step (2) first sorts by group priority, then by instance priority Sort, and finally generate a schedule. 3. the hybrid task scheduling method of a kind of railway locomotive operation and control system according to claim 1, is characterized in that, described based on heuristic search strategy has introduced the method for fuzzy control, has proposed a kind of new valuation function: H(n,t)=h(n,t)*n _st n _st =(n _d -t)-(n _c -e(n,t)) where H(n,t) is the valuation function, h(n,t) is the dynamic threshold coefficient of aperiodic real-time task n at time t, n _st is the remaining idle time of aperiodic real-time task n at time t, n _d is The deadline of task n, n _c is the worst execution time, e(n,t) is the time that task n has been executed at time t. 4. the hybrid task scheduling method of a kind of railway locomotive operation and control system according to claim 2, is characterized in that, in the described two-level priority rule based on table-driven, the design rule of described group priority is: First, set up several task groups according to the deadlines of periodic real-time tasks in the task set from small to large, and then determine the predecessor tasks of each periodic task and adjust the corresponding task group priority according to the timing constraints between tasks; The instance priority mentioned is the sorting rule of the internal tasks of the task group, and the sorting adopts the minimum and worst execution time rule. 5. the hybrid task scheduling method of a kind of railway locomotive operation and control system according to claim 3, it is also characterized in that, in the evaluation function, the completion rate of the aperiodic real-time task is also used as a fuzzy input parameter, and the input parameter The corresponding relationship with the threshold coefficient is shown in the following table: The task completion rate is defined as follows: n _cr (t)=e(n,t)/n _c . 6. The hybrid task scheduling method of a railway locomotive operation and control system according to claim 1, further characterized in that, in the step (5), for non-real-time tasks, the scheduling sequence is scheduled according to FIFO rules. 7. A mixed task dispatching model of a railway locomotive operation control system, characterized in that, the dispatching model is divided into two parts, respectively a dispatching unit and an execution unit, and the dispatching unit adopts any one of claims 1-6 Mixed tasks are scheduled using the method described in this item. 8. The mixed task scheduling model of a kind of railway locomotive operation and control system according to claim 7, characterized in that, said dispatch unit comprises a task collector, a real-time scheduler and a task integration unit, in the locomotive operation and control system During the running process, real-time tasks are continuously generated, and the task collector is used to centrally process the generated tasks, and submit them to the real-time scheduler when each clock signal arrives; the real-time scheduler, before the arrival of the next cycle, For different types of tasks, the periodic real-time task scheduling algorithm and the non-periodic real-time task scheduling algorithm are called respectively, and the periodic real-time task scheduling sequence and the non-periodic real-time task scheduling sequence are correspondingly generated; finally, the two sequences are merged by the task integration unit Generate a unified scheduling sequence and submit it to the task execution unit; the execution unit executes task scheduling according to the scheduling sequence, and feeds back the scheduling result and intermediate state information to the real-time scheduler, and the real-time scheduler collects and save.