JP2009025849A - Distributed control system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distributed control system for reducing the number of processors in the system. <P>SOLUTION: This control system 20 includes a first processor 21a, a second processor 21b, first control algorithm 54 running on the first processor, and second control algorithm 54 running on the second processor. The control system also includes third control algorithm 57 including a plurality of sequential steps sequentially running on the first processor and the second processor. Thus, the number of the processors in the system can be reduced. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a control system including a first processor, a second processor, a first control algorithm operating on the first processor, and a second control algorithm operating on the second processor. About.

  This type of control is typically classified as a “distributed control system”. Distributed control is used in systems where control functions can be distributed among different processors. Distributed control is sometimes referred to as “decentralized control” in contrast to “centralized control”.

  In most commonly available distributed control systems, the architecture is based on a central controller that controls the entire process and multiple sub-controllers that control multiple sub-tasks throughout the process. For example, in a robot manipulator, a central controller adjusts the entire operation of the robot manipulator, and a joint controller controls the operation of individual joints (joints). The central controller calculates the desired motion of each joint and sends this reference value to another joint controller that aims to follow the reference trajectory as accurately as possible.

  Another known application of a distributed control system is a system used to control the operation of a large diesel engine on a ship. In this example, the central controller controls the entire engine process, such as, for example, engine rotational speed and engine power output. Another “cylinder control unit” controls the operation of each individual cylinder, for example, fuel injection or exhaust valve opening. The central controller calculates how each individual cylinder should function to achieve the overall desired result. Therefore, the central controller sends a reference command to each cylinder control unit. The individual cylinder control unit then controls the detailed operation of the cylinder in order to achieve the reference command.

  One disadvantage of a typical system is that at least one processor needs to be used for central control functions, and that one processor must be used for each sub-task. There is a point. Thus, if a large diesel engine with 12 cylinders needs to be controlled, typically 12 + 1 processors are required. In some cases, a plurality of cylinders can be controlled by a single processor, thereby reducing the number of processors.

  Another disadvantage of a typical system is that if the central processor fails, the system itself is very vulnerable. For dangerous applications such as motor control, it is necessary to design the system to have some amount of redundancy. In most cases, the central control algorithm is implemented on two separate processors. The monitor routine monitors these two processors and can switch between the two processors if one processor fails. However, using two redundantly configured processors and knowing when to switch can be very complex. Furthermore, the need for redundantly configured processors increases the number of required processors and further increases the complexity of communication paths in the system.

  Accordingly, a first feature of the present invention is to provide a control system that reduces the number of processors required.

  Another feature of the present invention is to provide a control system that is resistant to the failure of one or more processors.

  Another feature of the present invention is to provide a control system in which redundancy is handled in a concise manner.

  The above-described feature is the control system described in the first paragraph of this specification, further comprising a third control algorithm including a plurality of sequential steps that operate sequentially on the first and second processors. Further provided by a control system that further includes. In this way, dividing the third control algorithm between the two processors allows the two processors to share the load of the third control algorithm.

  In the context of this specification, the phrase “sequential steps that operate sequentially on the first and second processors” refers to a plurality of steps that are operated sequentially, that is, sequentially. It should be understood as multiple steps. The first number of steps operates on the first processor, and the second number of steps operates on the second processor when the first number of steps finishes on the first processor. To do.

  The third control algorithm may include a control algorithm that operates as an iterative loop. A PI controller is a good example of this type of control algorithm. Dividing the execution of the third control algorithm, the execution of the first example of the loop is made on the first processor, and the execution of the second example of the loop is made on the second processor. can do. The execution of the third example of the loop can again be done in the first processor. This process can be repeated continuously.

  In one embodiment of this control system, the first control algorithm can control the first actuator and the second algorithm (55) can control the second actuator (3b).

  In some embodiments of the control system, the third control algorithm can provide a reference value for the first and second control algorithms. In this way, the third control algorithm can be viewed as an algorithm for some kind of adjustment that adjusts the operation of the first and second control algorithms.

  In order to allow multiple processors to communicate with each other, the control system may further include a communication system disposed between the first processor and the second processor. This communication system can, for example, transfer data used by the third control algorithm.

  The control system also provides a data vector transferred from the first processor to the second processor by the communication system when a third control algorithm switches from the first processor to the second processor. It may also contain. In this way, the control data used by the third control algorithm can be moved from one processor to the next when the third control algorithm is moved from one processor to the next. it can.

  The data vector described above may include a time stamp or sequence number that is set when the data vector is updated. In this way, the control algorithm determines when the data vector was last updated and / or which processor last updated the data vector.

  In one embodiment, the third control algorithm includes an integrator, and the integral value obtained by the integrator is transferred from the first processor to the second processor. In this way, the PI controller can be realized in the third control algorithm even when it is shared by a plurality of processors. In one embodiment, the above integral values can be transferred via a data vector.

  In some cases, the gain of the controller gain of the third control algorithm can be specified. This can be used, for example, even when the sampling time is not fixed and depends on external parameters. In this case, the gain designation may depend on the time interval between the time stamp in the data vector and the current time.

  In certain applications of this control system, this control system can be used to control a piston engine. The first control algorithm may control the operation of the first engine cylinder and the second control algorithm may control the operation of the second engine cylinder. The third control algorithm can further control the rotational speed and / or power output of the engine.

  In the embodiment described above, the control system can be synchronized with engine rotation. In this case, the gain designation may depend on the engine speed.

  In the embodiment described above, the execution of the third control algorithm can be switched between the first processor and the second processor based on the timing of the piston engine. In other words, the third control algorithm can move from one processor to the next in response to engine rotation. In this way, the sampling time of the third control algorithm depends on the engine speed.

  In this application, the term “include / include” is used to identify the presence of the described feature, number, step or component, but one or more other features, numbers, It does not exclude the presence or addition of steps, components or combinations thereof.

  For example, in the present claim 1, the control system includes three control algorithms and two processors. However, as will be appreciated by those skilled in the art, this current claim 1 includes any number of control algorithms greater than 3 and any number of processors greater than 2. You should understand that also covers.

  For the purposes of this specification, in this application, the term “control system” should be understood as a system that includes a plurality of interconnected physical components that operate in a coordinated manner in a predetermined manner. is there. These physical components may all be hardware based or may be programmed as software. In the latter case, the software programmed in the physical component is assumed to be a technical component of the physical component.

  The prior art control system 1 shown in FIG. 1 is typical of a distributed control system. In this example, the system is applied to a large diesel engine 2 on a ship (not shown). In this system, the engine 2 includes a plurality of cylinders 3a, 3b, and 3c, and the cylinders are controlled by cylinder control units (CCU) 4a, 4b, and 4c, respectively. The cylinder control unit 4 controls fuel injection, exhaust, supply of lubricant, and the like of the connected cylinder. Each CCU is connected to a communication system which is a communication bus 5 in this example. The plurality of CCUs thus communicate with each other to exchange sensor information and the like, as is well known in the prior art.

  Individual CCUs are coordinated by a central controller called an engine control unit (ECU) 6. The ECU calculates reference values for the individual CCUs and sends these values to the individual CCUs via the bus 5. For example, one of the typical functions of the ECU is to control the engine speed. The ECU receives a speed request signal from a steering person in the bridge of the ship 7. The ECU then compares the speed request signal with the actual speed of the engine. The actual speed of the engine is measured by a crankshaft rotation sensor 8 mounted on the engine. The crankshaft rotation sensor sends the measured speed to the ECU via the bus. Based on the difference between the actual speed and the desired speed, the ECU determines how much power each individual cylinder should provide. The ECU then sends a reference signal to each CCU. Each CCU then determines how these cylinders should be controlled.

  In one example, the speed control algorithm in the ECU can be based on a PI controller. The greater the error between the desired speed and the actual speed, the more the ECU commands the CCU to inject more fuel. Real fuel injection and fuel injection formation are controlled by individual CCUs. The ECU can also control other functions of the engine, such as emission control and vibration control.

  In order to prevent the engine from being stopped when the ECU 6 fails, a redundant backup ECU 9 is provided in the system. The backup ECU monitors the main ECU 6 and stops the main ECU 6 when a failure is detected.

  However, this redundancy can be handled in a number of ways well known to those skilled in the art. For example, another redundant controller (not shown) can be used separately from the main ECU 6 and the redundant ECU 9. In other cases, both ECUs send their control values over the bus and individual CCUs can choose to use the correct signal.

  However, the above-described system requires two ECUs in addition to the CCU. Furthermore, redundancy introduces introductory complexity in the system.

  FIG. 2 is a schematic diagram of an embodiment of a control system 20 according to the present invention. The control system 20 is applied to the large diesel engine 2 in the same ship as in the above-mentioned example relating to the prior art. Accordingly, the same reference numerals are used for the same components.

  As in the prior art system, the engine 2 includes a plurality of cylinders 3a, 3b, 3c, which are controlled by cylinder control units (CCU) 21a, 21b, 21c, respectively. As in the above example, each CCU controls the operation of one cylinder. However, as those skilled in the art will appreciate, if it is desired to use a smaller number of processors and there is an excess of processing power available in the CCU processor, each CCU will have multiple cylinders. Can be controlled.

  However, unlike the prior art, the control system 20 according to the present invention does not include a separate engine control unit having its own processor. The control algorithm operating in the ECU in the prior art is now divided into a plurality of sequential steps that operate sequentially in the CCU processor. This is possible because the ECU control algorithm can be operated during idle time available in the CCU.

  The ECU algorithm in this particular example is synchronized with engine rotation, and the ECU algorithm is sent from one CCU to the next in the same order that multiple cylinders ignite.

  When the engine is started, the first cylinder CCU starts an ECU algorithm that calculates the error between the desired rotational speed and the actual rotational speed, and a value relating to how much fuel should be injected into the first cylinder. Calculate This control value is based on the output of the PI controller. This value is sent to the first cylinder's CCU cylinder algorithm. The first cylinder CCU algorithm then injects the requested light fuel and sends the ECU algorithm to the control CPU of the second cylinder.

  When the speed control function is sent to the second cylinder CCU, the first cylinder CCU further sends a state vector or data vector to the second cylinder CCU. In this particular embodiment, the state vector includes the value of an integral part of the ECU control algorithm. In this way, the PI controller is implemented in a distributed manner and each cylinder CCU participates sequentially in the ECU control algorithm. The ECU control algorithm and the CCU control algorithm are synchronized with the crankshaft rotation of the engine.

  The state vector can also be used to send additional data between multiple CCUs. For example, a sensor connected to a CCU can be read when the CCU has control of the state vector, and the CCU can be updated before sending the state vector to the next CCU.

  As one of the advantages of the distributed control method according to the present invention, even if one CPU fails, the state vector from the CPU before the CPU in which the next CPU in the chain fails is used. Sometimes. This requires all CCUs to read the state vector from the bus each time it is sent.

  For example, consider the case where the CCU of cylinder x has control of an ECU control algorithm. The CCU of cylinder x calculates the error between the desired speed and the actual speed, reads the current value of the integrator from the state vector, updates the state vector with the new value of the integrator, Calculate the control value. This control value is sent to the CCU control algorithm for cylinder x, and the state vector is sent to the bus 5. All other CCUs read the state vector sent by the cylinder x CCU and store it in its own internal memory. Control is then transferred to the CCU of cylinder x + 1 that repeats this process.

  However, consider the case where the CCU in cylinder x + 1 fails. In this case, no control action occurs and cylinder x + 1 does not function. Furthermore, the state vector is not updated. However, when the crankshaft reaches a predetermined position, the ECU control algorithm is transferred to the CCU of the cylinder x + 2. Then, the CCU of the cylinder x + 2 continues the control based on the state vector sent on the bus by the CCU of the cylinder x.

  In this way, no extra programming is required even if the CPU fails. The cylinder connected to the failed CPU simply fails, but the other cylinders continue to function without any problems. As a result, a very simple system can be obtained and its function can be easily determined.

  Note that in the above example, if a failed CCU is present in the cycle, the sample time between two successive ECU control algorithm steps increases. This results in stability problems. Thus, the CCU is provided with a routine that can identify whether a previous CCU has failed. In this case, the CCU can adjust its own control algorithm to allow for a larger sample time.

  This can be realized, for example, when the state vector includes an identification stamp or a sequence number representing the CCU that last updated the state vector. For example, in the above example, when the CCU in cylinder x + 2 receives the state vector, the identification stamp or sequence number identifies cylinder x. The CCU of cylinder x + 2 will therefore know that cylinder x + 1 has failed.

  This is also achieved by adding a time stamp to the state vector. Thus, the gain of the ECU algorithm is specified according to the time difference between the time stamp in the state vector and the current time. For example, in the above example, the CCU in cylinder x can add a time stamp to the state vector immediately before sending the state vector onto the bus. The CCU in cylinder x + 2 reads this time stamp and determines the sample time between the last control action and the current time. Next, the CCU of cylinder x + 2 adjusts its gain if the time interval is greater than expected. This approach is useful in this particular example because the ECU control algorithm is synchronized with the crankshaft. Thus, the time interval between the steps of the ECU control algorithm is variable and depends on the crankshaft speed. Therefore, the above-described approach is also useful when a gain is specified for the crankshaft rotational speed.

  It should be noted that the selected gain specification can take on many shapes that are both linear and non-linear. For example, at very low speeds, the gain margin can be reduced so that it will not have too low a gain that could otherwise cause problems with engine operation. Therefore, if a gain designation is used, this gain designation should be adjusted to match the system being controlled.

  The state vector or data vector described above can also be used to transfer data other than control data. For example, sensor data can be transferred from one CCU to the next through a state vector. For example, one of the tasks of a CCU is to request the current cylinder temperature from a cylinder temperature sensor located in the cylinder to which the CCU is connected. The CCU can then update the corresponding value in the state vector before sending the state vector to the next CCU. In this way, the state vector will always have relevant data at all times. The effect of using the moving state vector as described here is that data movement is well defined. Thus, it becomes easy for the system programmer to know where the data is at any point in time. In a conventional distributed control system, data travels through many paths, so the data flow is less deterministic than in the present invention.

  Further, the above-described mobile state vector in the presently described example is effective because it contains all the data required by the ECU control algorithm. Because the ECU control algorithm and state vector are transferred together from one CCU to the next, the ECU will always have current data to use in operation.

  However, it is also possible to combine mobile state vectors with conventional distributed control communications with respect to time data. For example, if there is a problem with one of the cylinders, this one cylinder can send a signal regarding the problem to the rest of the system over the bus. Then, the ECU control algorithm can cope with the problem even when processing is currently performed in the CCU of another cylinder.

  FIG. 3 shows a timing diagram 30 of a typical conventional distributed control system. This system has four processors P1, P2, P3, P4. The system further includes four control algorithms 31, 32, 33, 34. The first three control algorithms 31, 32, 33 control process sub-elements (sub-elements), and the last control algorithm coordinates these sub-elements.

  At the end of the first sample 35, the control algorithm 34 performing the adjustment sends the new reference value to the sub-control algorithms 31, 32, 33. However, when a failure occurs in the processor P4 in which the control algorithm 34 that performs this adjustment exists, the entire process stops.

  FIG. 4 shows a timing diagram of a control system according to the present invention. The system includes three processors P1, P2, P3 and the same four control algorithms 31, 32, 33, 34 as in the prior art. However, the control algorithm 34 that performs the adjustment is divided into three sequential steps 34a, 34b, and 34c. These sequential steps are distributed sequentially among the three processors P1, P2, P3 during the processor idle time. In this way, the entire processor is saved.

  Note that the timing diagram in FIG. 4 is schematic. As will be appreciated by those skilled in the art, the time interval between tasks used for task designation in individual processors is short.

  FIGS. 5 and 6 show timing diagrams for three processors P1, P2, P3 when the control system according to the invention is applied to a large diesel engine having three cylinders. P1 is a part of the cylinder control unit (CCU) of the cylinder 1, P2 is a part of the CCU of the cylinder 2, and P3 is a part of the CCU of the cylinder 3. Thus, an engine with three cylinders has been simplified for illustration purposes, and it will be apparent to those skilled in the art that the method can be applied to engines having any number of cylinders. It is.

  FIG. 5 shows a timing diagram 50 when the motor is operating at 80 RPM. FIG. 6 shows a timing diagram 60 when the motor is operating at 200 RPM. In addition, these drawings are schematic drawings for illustrating the concept by omitting complicated points.

  When the motor operates at 80 RPM (FIG. 5), it makes one complete rotation every 750 ms. Stars 51, 52 and 53 indicate the time when each cylinder is at its top dead center (TDC) position. As can be seen in the figure, one of the cylinders is in a TDC position every 250 ms. In this example, the CCU control algorithms 54, 55, 56 require 25ms to complete and must be completed 25ms before the cylinder reaches the TDC. The ECU control algorithm 57 requires 25 ms to complete and must be completed 70 ms before the cylinder reaches the TDC. When the ECU control algorithm ends the calculation, it gives the calculated control value to each CCU control algorithm and sends it to the next processor.

  In FIG. 6, the motor operates at a rotational speed of 200 RPM. The motor therefore makes a complete revolution every 300 ms. The TDC points of different cylinders are therefore separated by 100 ms intervals rather than 250 ms. This reduces the amount of time available for processing. However, the control algorithm requires the same amount of time and must finish the same time before the TDC.

  As can be appreciated, there is more idle time in the processor when the motor is operating at 80 RPM, and less idle time in the processor when the motor is operating at 200 RPM.

  It should be noted that the ECU control algorithm can start the calculation before the previous CCU control algorithm finishes the calculation. In this way, the ECU control algorithm can start calculating before the TDC position of the previous cylinder. For example, a six cylinder engine operating at 200 RPM will have one cylinder located at TDC every 50 ms. Thus, for example, at some point in the process, while the CCU control algorithm for cylinder 1 is operating at P1, the ECU control algorithm operates at P2 and calculates a control value for the CCU control algorithm for cylinder 2 Means.

  Note that the order in which the ECU control algorithm moves from one processor to the next does not need to follow the firing order of the cylinders if such is not desired. Further, the ECU control algorithm can operate on one processor of one CCU and send a calculated signal to the processor of another CCU. For example, the ECU control algorithm can operate in the CCU processor of cylinder 1 and send the calculated reference value to the CCU of cylinder 3. Further, the ECU control algorithm can operate in the CCU processor of the cylinder 2 and send the calculated reference value to the CCU of the cylinder 1.

  The foregoing description and drawings are shown in simplified and schematic form to illustrate exemplary embodiments. Internal electronic and mechanical details are not shown, but this is known to those skilled in the art, and it is necessary to complicate the individual description unnecessarily. Is from within.

  A person skilled in the art would include a number of different inventions in this application, and could be used as material for a divisional application if necessary. For example, an invention relating to a mobile state vector in a network can be used independently of an invention relating to distributed control.

  The above description focuses on the case where the present invention is applied to engine control. However, those skilled in the art should appreciate that the present invention can be used in many other applications where distributed control can be applied. Therefore, the protection scope of the present invention is not limited to application examples related to engine control.

1 is a schematic diagram of a prior art control system typically used to control a large diesel engine. FIG. 1 is an I schematic diagram of a first embodiment of a control system according to the present invention applied to a large diesel engine. Fig. 4 shows an example of a timing diagram for a prior art control system including four processors and four control algorithms. Fig. 4 shows an example of a timing diagram for a system according to the present invention comprising three processors and the same four control algorithms shown in Fig. 3; 2 shows an example of a timing diagram at a first low rotational speed of a control system according to the invention applied to a large diesel engine. Fig. 6 shows a timing diagram at a second high rotational speed of the embodiment of Fig. 5;

Claims (17)

A first control algorithm (54) operating on the first processor (21a), a second processor (21b), the first processor and a second control algorithm operating on the second processor A control system (20) comprising:
The control system further comprising a third control algorithm (57) including a plurality of sequential steps that operate sequentially on the first processor and the second processor.   The control system (20) of claim 1, wherein the sequential steps of the third control algorithm (57) include a first number of steps and a second number of steps, wherein the first number. Of the second number of steps are executed on the first processor (21a), and the second number of steps is performed when the first processor finishes the first number of steps. A control system characterized by operating on   The control system (20) of claim 1, wherein the third control algorithm (57) includes an iterative control loop, and one execution of an example of the control loop is performed on the first processor (21a). The control system is characterized in that the next example of the control loop is performed on the second processor (21b).   The control system (20) according to any one of claims 1 to 3, wherein the first control algorithm (54) controls a first actuator (3a) and the second algorithm. (55) is a control system for controlling the second actuator (3b).   The control system (20) according to any one of claims 1 to 4, wherein the third control algorithm (57) is based on the first and second control algorithms (54, 55). A control system characterized by providing a value.   The control system (20) according to any one of claims 1 to 5, wherein the third system is arranged between the first processor (21a) and the second processor (21b). A control system further comprising a communication system (5) for transferring data used by the control algorithm (57).   The control system (20) according to any one of claims 1 to 6, wherein the third control algorithm (57) switches from the first processor to the second processor. The control system further comprising a data vector transferred between the first processor (21a) and the second processor (21b) by the communication system (5).   8. The control system (20) of claim 7, wherein the data vector includes a time stamp or sequence number that is set when the data vector is updated.   The control system (20) according to any one of claims 1 to 8, wherein the third control algorithm (57) includes an integrator, and the integration value of the integrator is the first control value. The control system is transferred from the processor (21a) to the second processor (21b).   10. Control system (20) according to claim 9, characterized in that the integral value is transferred via the data vector.   11. Control system (20) according to any one of claims 1 to 10, wherein the third control algorithm (57) and / or the first and second control algorithms (54, 55). The control system is characterized in that the controller gain is specified.   12. The control system (20) according to claim 11, wherein the gain designation depends on a time interval between a time stamp in the data vector and a current time.   13. A control system (20) according to any one of the preceding claims, wherein the first control algorithm (54) is used for controlling a piston engine (2). A control system for controlling the operation of the cylinder (3a), wherein the second control algorithm (55) controls the operation of the second engine cylinder (3b).   14. The control system (20) according to claim 13, wherein the third control algorithm (57) controls the rotational speed and / or power output of the engine (2).   15. Control system (20) according to claim 13 or 14, characterized in that it is synchronized with the rotation of the engine (2).   16. Control system (20) according to claim 15, characterized in that the gain designation depends on the rotational speed of the engine (2).   17. The control system (20) according to claim 15 or claim 16, wherein execution of the third control algorithm (57) is performed with the first processor (21a) based on the timing of the piston engine (2). A control system which is switched between the second processor (21b).