DE10048747B4 - Process and processing of process signals - Google Patents

Abstract

Method for processing process signals (P), in particular of a plant process (2), in which a first processing unit (4) detects the process signals (P) in a processing cycle, generates these individual messages (E _X ) and sends the individual messages (E _X ) for forwarding wherein a maximum number (N _max ) of the individual messages (E _X ) that can be generated during a processing cycle is predetermined and independent of the quantity of the incoming process signals (P).

Description

The The invention relates to a method for processing process signals; in particular for processing process signals of a plant process.

at a plant, for example, in an automation system, are a variety of individual components provided, in particular via a Bus system communicate with each other. By means of the example formed as a fieldbus bus system, the process signals of individual components transmitted to processing units. For complex systems As a rule, a large number of such processing units are provided, which are usually hierarchically structured. In particular, in addition to one Variety of distributed processing units a central processing unit provided, for example, within the framework of a centralized or control room. In the decentralized processing units takes place a first processing of the process signals by preparing them into individual messages become. For this purpose, the individual process signals, for example an identifier is assigned and further processing steps are performed. In The central processing unit then become the different ones Single messages evaluated.

One essential point for This evaluation is in a time-accurate evaluation of the individual process signals to see. This is particularly important when the individual messages different processing units compared and the chronological order of the process signals for the evaluation is essential. This is for example in case of a fault of high interest when coming from multiple processing units in the shortest time Time error messages are captured to decide where the error is first occurred. As accuracy for such a timely or time-synchronous evaluation of the process signals is typically An accuracy of less than 10 milliseconds is sought.

One essential criterion for the time accuracy is the time duration of a processing cycle respective decentralized processing unit. The processing cycle includes detecting the process signals, generating the process signals associated with Single messages and their transmission to the central processing unit. The processing cycle thus comprises a detection period, a Processing period and a transmission period. A process signal, the for example during the processing period at the decentralized processing unit is received, only with the next Processing cycle processed. The time accuracy is therefore decisively influenced by the duration of the processing cycle. A temporal resolution the individual process signals within a processing cycle is not given.

One The problem now is that with fluctuating durations for the individual Processing cycles the time accuracy varies, making a statement about the actual Time accuracy can not be made. In particular, it will the resulting newspaper inaccuracy in a comparison of individual reports increased from several decentralized processing units. A waver the duration of a processing cycle becomes particular caused by a fluctuating process signal quantity. Because the The duration of a processing period depends on the number of process signals from.

The DE 195 29 718 C2 discloses a method for processing process signals of a technical process in which a processing unit detects process signals in a processing cycle, generates them from these individual messages and makes them available for forwarding via a bus interface to a field bus in memory areas. The storage areas have a variable length, which can be adapted to the actually required needs. This is done depending on the amount of incoming messages.

Of the Invention is based on the object, a processing of process signals to enable with high time accuracy.

The Task is according to the invention solved by a method for processing process signals in particular a plant process, in which a first, in particular decentralized processing unit in a processing cycle, the process signals detected from these Single messages generated and the individual messages for forwarding providing a maximum number of during a processing cycle specifiable single messages and regardless of the amount of incoming Process signals is.

The limitation of the number of process signals, which are processed into single messages during a processing cycle, ensures a constant processing period and thus a substantially constant, equally long processing cycle, independent of the quantity of incoming process signals. With this method, therefore, advantageously a variable or fluctuating input load at a signal input of the first processing unit to a constant Verarbei load or constant output load.

Preferably become primary signals, those from the process signals or from these derived signals formed and processed into individual reports, cached in a first memory. This will ensure that even with a fluctuating process signal quantity all process signals recorded and no information is lost. The use of the first memory allows the editing of during a processing cycle incoming process signals to several Split processing cycles. This embodiment is the Idea at a high load, so a high number of Process signals, these intermediate and then successively to process into single messages during low load periods and thereby reduce. Under light load periods are here Processing cycles with a small amount of process signal Understood.

to Achieve as much as possible high time accuracy, the processing period is as short as possible held and in particular to that at the respective first processing unit adapted to expected process signal volume. This is the maximum Number of during a single processing cycle generated in the field of time average of the number of incoming processing cycles Process signals. This ensures that that in averaged over all process signals be processed and no information is lost.

Around the amount of data to be processed and forwarded per Processing cycle possible To keep low, preferably only changes in the process states of the plant process recorded and evaluated. For this purpose, it is expediently provided the process signals to changes to monitor and in case of a change a specific process signal associated with this process signal change signal as primary signal to generate and only this process to a single message.

Conveniently, becomes the primary signals imprinted a time information in the form of a time stamp. Thereby will also be independent from the storage duration of a process signal in the first memory a exact time assignment of the primary signal to the occurrence of the Process signal allows.

advantageously, become the primary signals assigned to the same time stamp, which at the same time belong to the same processing cycle, together stored in a memory block to which the timestamp is assigned becomes. It is therefore not necessary to give each primary signal its own time stamp but the timestamp is preferred only once, and although the memory block, assigned. In this memory block are then only the primary signals filed, which are assigned to this timestamp. This will processing capacity saved and thus kept the processing period as short as possible.

Conveniently, different process signals are assigned to different channels, and the memory block has a number of individual memory locations, each of which is assigned a specific channel. this has the advantage of having the association between the channel and the storage space is fixed and not first defined during each processing cycle must become. This is the processing burden in terms of Storing the process signals or their associated primary signals largely independent from the actual Signal traffic.

Furthermore this fixed allocation between storage space and channel is advantageously used, to generate the single message from the position of the memory location derive a signal identifier assigned to the single message becomes. About that In addition, the single message is additionally the time stamp as well one in the primary signal assigned process information assigned. The single message contains so the complete and evaluable information on what, when, where happened.

Conveniently, the individual messages are forwarded to a second one before being forwarded Memory cached. Because the forwarding usually has a Bus takes place, which have a fluctuating transmission capacity can, is thus an independence of the possible Data throughput achieved on the bus line. So while the first memory buffering a fluctuating input load is the second memory for one Buffering or balancing different transmission capacities the bus line provided. Overall, this procedure will help with the two storage or Buffer systems a constant processing cycle and thus a achieves constant and high time accuracy.

Conveniently, the single messages in the second memory are combined to form message blocks, As these look forward to the forwarding and transmission in the bus advantageous are.

Around in case of a fault to ensure the first value acquisition, that is the detection of the first incoming process signals, remain at the exhausted storage capacity on The process unit ignoring the first processing unit is disregarded.

For transmission the single messages and preferably the message blocks to a second, preferably central processing unit is to this of the first processing unit transmits a message signal as soon as the individual messages are provided for forwarding. This will allow that the second processing unit merges the respective message blocks into one Time at which they have processing capacity free Has.

alternative For this purpose, it is provided that a substantially constant number the individual messages cyclically to the second processing unit is forwarded.

One embodiment The invention will be explained in more detail below with reference to FIGS. Show it each in a schematic representation:

1 a diagram of a method for processing process signals and

2 the time course of the frequency of incoming process signals in relation to the time course of the frequency of processed to individual messages processed process signals.

According to 1 become from a plant process 2 Process signals P to a first, decentralized processing unit 4 transmitted. In this, the process signals P are processed into individual messages E _X , which processing unit to a second central processing 6 to get redirected. The index "X" stands for a multitude of discrete single messages, which in 1 are numbered with numbers.

The plant process 2 For example, an automation process, in which different system components are provided, each process signals P to the first processing unit 4 to transfer. Each of these individual components, ie each type of process signal P, is assigned a channel K _n . The index "n" stands for a variety of discrete, in the 1 numbered channels K _n . These can be realized by different input channels in the first processing unit or by software technology by an identification information in the process signal P hardware. The different channels K _{n of} the plant process 2 are schematic by a diagram 8th represented on the ordinate the channels K _n (indicated by horizontal lines) and the frequency P _{H of} the occurring process signals P (process signal frequency) are plotted against the time t. The process signal frequency P _H is by a curve 10 shown.

That through the bend 10 represented fluctuating process signal volume defines a fluctuating input load for the second processing unit 4 , According to the invention, it is provided to map this fluctuating input load to a largely constant processing load or output load.

This basic principle is in 2 illustrates in which the process signal frequency P _H and the frequency of the individual messages E _H (frequency of messages) are plotted against the time t.

The course of the process signal frequency P _H is characterized by cyclically occurring load peaks, between which only a small base load occurs. This process is typical, for example, for a manufacturing process in which a workpiece is processed by a machine (system component) and then forwarded to the next system component. During processing of the workpiece, comparatively little process signals P are emitted. When machining is finished, the workpiece may be ejected, the machine may need to be reset, and a conveyor belt may need to be moved further, requiring a variety of actions at this time, resulting in a large amount of process signals.

This periodic load will - like the 2 can be seen - imaged on a constant continuous load, which is represented by the message frequency E _H. The constant value of the message frequency E _H corresponds to a predetermined maximum number N _{max of} the individual messages E _X to be generated per processing cycle. For the shortest possible processing cycle, it is provided that the maximum number N _{max of} the individual messages E _X generated per processing cycle depends on the time average M of the process signals P per processing cycle. This measure is a largely constant continuous load for the decentralized processing unit 4 achieved. In order to reserve a certain reserve capacity for unforeseen events, the maximum number N _{max of} the producible individual messages E _{X can be set} slightly above the time average M.

Due to this constant continuous load, the processing load is for the first processing unit 4 essentially constant. The width of the individual bars in the message frequency E _H corresponds to the duration of a processing cycle of the first processing unit 4 , Thus, a constant number of individual messages E _X from the individual process signals P is processed per processing cycle. As a result, the load per processing cycle is independent of the amount of incoming process signals P and largely constant. Due to the uniform load per processing cycle, resulting in a constant period of individual processing precipitates a high time accuracy. A high time accuracy is particularly necessary to that of different decentralized processing units 4 to the central processing unit 6 transmitted single messages E _X temporally analyze. Prerequisite for a time differentiation of the individual single messages E _X is a high time accuracy.

As already explained, the basic idea for achieving the time accuracy is to map a fluctuating input load to a substantially constant processing load. In order to lose no information, it is in the first processing unit 4 a buffer system 12 with a first memory 14 and a second memory 16 intended. This is in 1 illustrated, by which the preparation of the process signals P to the individual messages E _{X is} explained: The incoming process signals P are in a comparator 18 monitored for changes, ie the state of the respective process signal P of a channel K _n in the previous processing cycle is compared with the current state. If there is a change, it is calculated by the comparator 18 generates a change signal A. The information regarding the individual channels K _n is thereby from the comparator 18 maintained in order to ensure an assignment of the respective process signal P to the respective system component.

The change signals A are in the first memory 14 cached. Namely, the change signals A detected during a processing cycle are stored in a memory block B _{m associated with} the respective processing cycle. Each of these memory blocks B _m is assigned a time information in the form of a time stamp T _m . The time stamp T _m is in this case from an internal clock, not shown, of the first processing unit 4 is generated, and that each processing cycle, a time stamp T _{m is assigned} . The internal clock is synchronized in a conventional manner with other clocks in the individual processing units.

The index "m" is also here for a plurality of individual memory blocks B _m , which are each assigned to a time stamp T _m associated processing cycle with change signals A.

During a processing cycle, only one memory block B _m is therefore described in each case. According to 1 this is the memory block B1. Each of the memory blocks B _m in this case has a K _n to the channels corresponding number of storage locations _n to S, wherein a respective space is S _n a particular channel assigned to K _n. If a change occurs in a specific channel K _n, a change signal A is assigned to this channel K _n and this is assigned in an allocated memory location S _n in the memory block B _m . According to the exemplary embodiment, changes are made, for example, on the channels K1, K2 and K4, which are stored in the allocated memory locations S1, S2, S4 in the memory block B1. This is indicated by crosses. In the first store 14 the memory blocks B _{m are represented} by rows and the memory locations S _{n are represented} by columns.

The memory blocks B _m thus described are stored in the first memory 14 cached. For each processing cycle, a constant number of the change signals A stored in the individual memory blocks B _m is now processed into individual messages E _X.

These single messages E _X are in message blocks 22 summarized, with each message block 22 a fixed number of individual messages E _{X is} assigned. In this case, an order of the individual messages can be made either after the time stamp T _m or after the respective channel K _n . Ie in a special message block 22 Only single messages E _X with the same time stamp T _m or only individual messages E _{X of} a specific channel K _{n are} combined.

To generate the individual messages E _X , a signal identifier J and the time stamp T _m assigned to the respective memory block B _{m are} assigned from the position of the memory location S _{n of} the respective change signal A. In addition, the single message E _X still includes the process information contained in the change signal A übe the change made to the process state.

In the method described here is by means of the buffer system 12 one per processing cycle of the first processing unit 4 essentially constant load and thus achieved high time accuracy. Due to the bipartite buffer system 12 with the first memory 14 and the second memory 16 Both a variable, fluctuating input load is buffered, as well as a fluctuating transmission capacity on the bus line to the second, central processing unit 6 considered. Overall, this achieves that regardless of the amount of incoming process signals P and regardless of the transmission capacity on the bus line a high time resolution of all incoming information is enabled.

In the case of a serious disruption in the process flow, it will usually come to a large amount of process signals P, which may lead to the storage capacity of the first or second memory 12 . 14 überschrit will be. In this case, the later arriving process signals P are no longer detected. This ensures that the first incoming process signals later from the central processing unit 6 can be evaluated (initial value acquisition).

For the transmission of the individual message blocks 22 to the second processing unit 6 standard mechanisms are used. In particular, the first processing unit 4 in the presence of a delivery-ready message block 22 an indication signal MS to the second processing unit 6 transmitted. The message block 22 is then from the second processing unit 6 as soon as it has enough capacity for this. As an alternative to this transmission mechanism, a predetermined amount of individual messages E _X or message blocks can also be per processing cycle 22 periodically and automatically to the second processing unit 6 be delivered. This results in a periodic and constant output load, which corresponds to the single message frequency E _H , as in 2 is shown.

Claims (13)

Method for processing process signals (P), in particular a plant process ( 2 ), in which a first processing unit ( 4 ) detects the process signals (P) in a processing cycle, generates from these individual messages (E _X ) and provides the individual messages (E _X ) for forwarding, wherein a maximum number (N _max ) of the individual messages (E _X ) that can be generated during a processing cycle are specified and regardless of the amount of incoming process signals (P). Method according to Claim 1, in which primary signals (P, A) to be processed for individual messages (E _X ), which are formed from the process signals (P) or from signals (A) derived therefrom, are stored in a first memory ( 12 ) are cached. Method according to Claim 1 or 2, in which the maximum number (N _max ) of the individual messages (E _X ) which can be generated during a processing cycle is in the region of the time average (M) of the number of process signals (P) received per processing cycle. Method according to one of the preceding claims, in which the process signals (P) are monitored for changes and when a process signal (P) changes, a change signal (A) is generated as a primary signal and this is further processed into a single message (E _X ). Method according to one of Claims 2 to 4, in which the primary signals (P, A) are provided with a time stamp (T _m ). Method according to one of Claims 2 to 5, in which primary signals (PA) assigned to the same time stamp (T _m ) are stored together in a memory block (B _m ) to which the time stamp (T _m ) is assigned. Method according to Claim 6, in which different process signals (P) are allocated to different channels (K _n ), and the memory block (B _m ) has a number of individual memory locations (S _n ) to which a specific channel (K _n ) is permanently assigned becomes. Method according to Claim 7, in which the respective individual message (E _X ) is formed from the assigned primary signal (P, A) by the individual message (E _X ) of the time stamp (T _m ), one from the position of the memory location (S _n ) derived signal identifier (J) and a process information contained in the primary signal (P, A) is assigned. Method according to one of the preceding claims, in which the individual messages (E _X ) are sent to a second memory ( 16 ) are cached. Method according to Claim 9, in which a plurality of the individual messages (E _X ) are stored in a message block ( 22 ) and in the second memory ( 16 ) are cached Method according to one of Claims 2 to 10, in which, when the storage capacity is exhausted, the incoming ones Process signals (P) are disregarded. Method according to one of the preceding claims, in which the first processing unit ( 4 ) to a second processing unit ( 6 ) a message signal (MS) is transmitted as soon as the individual messages (E _X ) are provided for forwarding. Method according to one of Claims 1 to 11, in which a substantially constant number of individual messages (E _X ) is sent cyclically to a second processing unit ( 6 ) is forwarded.