AUTOMATION SYSTEM FOR MONITORING A SAFETY-CRITICAL PROCESS

DE502019012442D1Active Publication Date: 2024-11-28PILZ GMBH & CO KG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DE502019012442
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2018-08-21
Filing Date
2019-06-19
Publication Date
2024-11-28
Estimated Expiration
2039-06-19

AI Technical Summary

Technical Problem

Current automation systems for safety-critical processes face challenges in achieving high safety integrity levels due to processor-dependent fault control measures, which become impractical for complex processors, and Software Coded Processing methods are time-consuming and complex, especially when implementing signature methods.

Method used

An automation system that uses a non-safe platform with a fail-safe peripheral module and a monitoring device to ensure the execution of safety functions, where the monitoring device can operate independently to detect errors and initiate a safe shutdown, utilizing a secure runtime environment and external components to simplify the implementation of safety functions on standard hardware.

Benefits of technology

This approach allows for the implementation of complex safety functions on standard hardware with reduced complexity and cost, achieving high safety integrity levels by distributing safety-related tasks across optimized components, simplifying certification, and ensuring reliable error detection and response.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The present invention relates to an automation system for monitoring a safety-critical process, a corresponding method and a monitoring device for ensuring the execution of a safety function implemented from user programs on a platform and a fail-safe peripheral module for coupling a platform, which executes user programs to implement a safety function, to a safety-critical process.

[0002] Automation systems for monitoring safety-critical processes are used to reduce the risk posed by technical systems to people and the environment. To this end, safety functions are implemented that transfer the technical system or safety-critical process to a safe state in the event of danger. According to DIN EN ISO 13849-1 and DIN EN ISO 12100, the term "safety function" refers to a safety-related control function of a machine that reduces the risk posed by the machine to an acceptable level. An example of a safety function is the shutdown of a machine after an emergency stop button is pressed.

[0003] Originally, safety functions were implemented through individual safe modules, for example, in the form of safety switching devices with relay technology, which operated independently of the control system of the system being monitored. In subsequent development, safety switching devices were subsequently logically linked to implement more complex safety functions.

[0004] Safety controllers are now used for even more complex tasks. Safety controllers emerged primarily from the desire to be able to program safety functions similar to a programmable logic controller (PLC). In their actual function, safety controllers therefore differ only slightly from PLC controllers. Essentially, a safety controller is two individual PLC controllers that execute a user program in parallel, use the same process image of inputs and outputs, and constantly synchronize with each other.

[0005] Internally, however, the structure of a safety controller differs significantly from that of a PLC controller in terms of implementing the special safety-related features. A safety controller therefore generally differs from a PLC controller in that it has two separate channels, a diverse structure with different hardware, continuous testing of inputs and outputs, continuous comparison of user data, voltage and time monitoring, and safe shutdown in the event of an error or danger. In addition, the components involved in the safety functions - especially the CPUs (Central Processing Unit) - must be fail-safe. To implement standard-compliant safety, in particular to achieve a Safety Integrity Level (SIL 3) according to IEC 61508, redundant CPUs were originally used, i.e. at least two CPUs that monitor each other.

[0006] EP 1 043 641 A2 discloses a fail-safe automation system based on a standard CPU. The error control measures are largely integrated into the user program and include: safety protocols, temporal program execution control, logical program execution and data flow control, data protection through information redundancy, diverse processing, and self-tests within the process fault tolerance time. Instructions that cannot be implemented diversely are tested within the process fault tolerance time. In addition, background tests are performed by the CPU's operating system to detect multiple errors within the multiple error occurrence time.

[0007] A disadvantage of the safety concept according to EP 1 043 641 A2 is that the error control measures mentioned are processor-dependent and therefore proof of the fault tolerance of the processor must be provided for SIL 3 according to IEC 61508. For complex processors, however, a diagnostic coverage (DC) of at least 99% based on a failure mode effect analysis (FMEA) of the processor is no longer practically feasible. The processor used in EP 1 043 641 A2, on the other hand, is a special ASIC of low complexity that directly executes the provided code. Proof of fault tolerance can still be provided here with reasonable effort. When using other standard CPUs, however, the corresponding code cannot be executed directly, so the compiler for implementing the corresponding code on the machine code must also be analyzed.Furthermore, hardware-related background tests are required to detect multiple errors, which are not implemented in standard CPUs or are not implemented with sufficient effectiveness and therefore have to be implemented separately when using a standard CPU.

[0008] So-called Software Coded Processing (SCP) pursues a processor-independent approach. A system extended with SCP is capable of revealing transient, permanent, and systematic execution errors during runtime. Furthermore, interference between different programs (functional programs, middleware, operating system) can be detected and displayed. The best-known variant of SCP is so-called AN coding, i.e., an arithmetic coding in which each value of a program is multiplied by a constant A, in particular a prime number. All values ​​and states in the system that are not multiples of A are subsequently considered invalid states.

[0009] Forin, P.: "Vital coded microprocessor principles and application for various transit systems" in Perrin, JP: Control, Computers, Communications in Transportation. Selected Papers from the IFAC / IFIP / IFORS Symposium, Pergamon, Oxford, UK, 1990, pp. 79-84 describes a first approach to a fail-safe system based on coded processing. The system described comprises a fail-safe input module that encodes the input signals and a standard CPU (coded monoprocessor) that uses coded operations to calculate coded state and output signals from the coded input signals and creates a signature for them. A fail-safe dynamic controller checks the signatures calculated by the standard CPU and, in the event of a fault, deactivates a safe output module.The method thus combines arithmetic coding (multiplication with prime number A) and a signature method (static signature, dynamic signature) and enables a computational determination of a diagnostic coverage level without having to perform a complex FMEA of the CPU.

[0010] Süßkraut, Martin, and Kaienburg, Jörg: "Safety critical smart systems with software-coded processing" further discloses a practical implementation of a fail-safe system based on SCP. The implementation is based on two different (diverse) executions (native and coded) of the same safety function and, due to its special design, supports industry standards such as IEC 61508. Native execution corresponds to the execution of the original source code of the safety function (native program) and processes the native input values ​​and states. Thus, only the native states are processed, and the result of the native execution is the native outputs. Coded execution executes a coded form of the original safety function (coded program). This requires that the original source code of the safety function has been previously transformed and coded.The coded program then operates on coded input values, coded states, and generates a coded output. The native and coded programs both operate on the same output values, with the coded input values ​​being the coded variant of the native input values. The same applies to the state and output values. The native and coded programs are executed and managed by a so-called diversity framework. The diversity framework is generated from the source code of the original safety function by enriching the source code with appropriate annotations. The diversity framework coordinates the parallel execution of the two programs (native and coded) and verifies error-free execution using checksums it creates over the native and coded output values.Furthermore, the diversity framework can monitor the control flow of the safety function by integrating the control flow check data flow in such a way that any control flow error changes the output checksum.

[0011] A disadvantage of the aforementioned SCP-based methods, however, is that the coded operations require a lot of runtime, and code generation is time-consuming and complex, especially when signature methods are also used. For example, with SPC, the coded program requires more bits for storing and processing values ​​than the native variant. If, for example, the native program uses 32-bit data types, the coded program must use at least data types that can store A times the largest possible value of a 32-bit data type. Therefore, the coded program is usually implemented with 64-bit data types if the native program uses 32-bit data types. Likewise, all native operations must be replaced by corresponding coded operations, the execution of which takes more runtime than the native operations and requires a correspondingly complex tool chain to create.

[0012] DE 10 2006 046 286 A1 shows further state of the art.

[0013] Against this background, it is an object of the present invention to provide a safe automation system that avoids the above disadvantages, is based on standard components and at the same time enables the implementation of a standard-compliant safety function of a high safety category.

[0014] According to one aspect of the present invention, this object is achieved by an automation system for monitoring a safety-critical process, comprising a non-safe platform for executing user programs that implement a safety function, a fail-safe peripheral module via which the safety-critical process can be coupled to the user programs, and a monitoring device that is configured to ensure the execution of the safety function on the non-safe platform, wherein the monitoring device can be coupled to the non-safe platform via a first communication interface and the fail-safe peripheral module can be coupled to the non-safe platform via a second communication interface, and wherein the monitoring device is configured to execute at least one fail-safe service independently of the non-safe platform,to exchange data cyclically generated by the fail-safe service with the fail-safe peripheral module according to a defined scheme, and wherein the fail-safe peripheral module is configured to signal a safe state depending on the implemented safety function and based on the data exchange that takes place via the non-safe platform.

[0015] This task is also solved by a corresponding procedure.

[0016] The present invention therefore proposes to secure the implementation of a safety function implemented primarily in software on any non-secure platform using platform-external components. The goal is to design the external components that are crucial for safety as simply as possible. In other words, the idea is to implement the complex design of a safety function primarily in software on a low-cost platform and to outsource the required inherent fault tolerance to components that are easy to implement and thus certifiable with less effort.

[0017] The external components, preferably implemented as dedicated hardware, interact in such a way that in the event of faulty execution of the user programs on the platform, faulty behavior of the platform itself, or errors in communication with it, the process enters a safe state. To this end, a monitoring device ensures the execution of the user programs, preferably by collaborating with a safe runtime environment on the platform. In the event of a fault, the monitoring device can initiate the execution of the safety function independently of the platform via the fail-safe peripheral module using a separate service executed independently of the platform, based on the "watchdog" principle, and in particular, can shut down the process.

[0018] The platform can be a software, hardware, or virtual platform that serves as the basis for developing and executing application programs. The platform is a non-secure platform, for example, a single-channel system like a standard PC. Alternatively, the platform can also be provided in the form of cloud computing, particularly as Infrastructure as a Service (IaaS) or Plattform as a Service (PaaS) In a preferred embodiment, the platform may already include a secure runtime environment. In this context, "non-secure" means that the non-secure platform does not meet the relevant safety requirements to independently execute a safety function with the required degree of inherent fault tolerance.

[0019] The fail-safe service can implement simple data-related functions or operations that can be implemented reliably and fail-safely in hardware or software. Furthermore, these can be certified with minimal effort according to the relevant safety standards, even for high safety categories. The monitoring device is therefore limited in scope to the bare minimum required for the exclusive execution of at least one fail-safe service.

[0020] The monitoring device is connected to the fail-safe peripheral module via a communication interface, which is also non-safe, i.e., a single-channel communication interface, in order to cyclically exchange data generated by the fail-safe service with the peripheral module according to a defined pattern. The data exchange, which takes place via the platform, can ensure proper operation of the monitoring device and, associated with this, monitoring of the platform or a secure runtime environment based on it, whereby malfunction of the monitoring device also leads to the triggering of the safety function. In addition to the fail-safe service, the monitoring device can have further devices which interact with the platform and, in particular, with a secure runtime environment operating on the platform independently of the user programs.

[0021] The advantage of the present invention is that any complex safety function can be implemented on standard hardware or standard modules, with the final safety being supplemented by a simple, separate component. The invention is therefore based on the principle of making the individual components of the automation system involved in the safety function only as safe as necessary, with a high level of safety being achieved through the interaction of the individual parts. In this way, for example, computationally intensive tasks can be implemented on fast and cost-effective hardware, and the necessary safety features are supplemented by dedicated hardware. The dedicated hardware is not capable of executing the complex safety function on its own, but has just enough computing power to safeguard the execution of the safety function by executing simple services.

[0022] The monitoring device can therefore be designed to be less complex than the platform executing the safety function, which simplifies safety acceptance by the responsible authorities. A particularly advantageous feature is the fact that safety protection is independent of the safety function to be implemented. This allows for safety acceptance to be carried out essentially independently of the safety function implemented in the individual case.

[0023] The triggering of the safety function itself can be initiated by the monitoring device, while its execution is left to a fail-safe peripheral module, which is already present in an automation system based on standard components. One focus of the invention is therefore on distributing individual aspects of a safe control system across different components, with each component setting its own focus and being optimized for this purpose. In this way, a complex safety function can be implemented cost-effectively and flexibly in a high safety category.

[0024] In a preferred embodiment, the monitoring device is configured to receive a random value from the platform, in particular a random value generated by the fail-safe peripheral module. The random values ​​can be used to easily determine whether a remote station is operating properly. A random generator can also be easily implemented and safety-tested. This embodiment thus advantageously contributes to a simplified design of the fail-safe service.

[0025] In a further embodiment, the fail-safe service comprises a counter unit configured to generate a continuous counter value, and a coding unit configured to generate a key value. A fail-safe counter unit and a fail-safe coding unit can together offer a fail-safe service that can be used to adequately secure the executing platform. At the same time, a fail-safe counter unit and a fail-safe coding unit are easy to implement as fail-safe components and are regularly available as ready-made, fail-safe modules. The monitoring device can thus be implemented particularly cost-effectively. Preferably, the counter unit and the coding unit are implemented in a fail-safe manner, i.e., the design ensures that an error in one of these units is detected.For example, the counter and encoder can be designed with multi-channel redundancy and implemented on multi-channel redundancy hardware. Since both the counter and encoder are simple units, their inherent fault tolerance can be reliably ensured and verified with minimal effort. This design thus contributes to implementing the safety of the automation system at a particularly cost-effective level.

[0026] In a further embodiment, the coding unit can generate the key value from a random value and the continuous counter value. The random value is provided by or via the platform, i.e., supplied to the monitoring device from outside. The monitoring device checks whether the random values, which are preferably received cyclically, change over time. At the same time, coding with the continuous counter value can easily ensure that all numbers are cycled through. The monitoring device therefore does not have to separately check whether the random number generator, which provides the random value, actually cycles through all values, which would be difficult due to the monitoring device's intended, weak performance. This embodiment thus further contributes to a cost-effective implementation of the monitoring device.

[0027] In a further embodiment, the first communication interface is further configured to forward the continuous counter value and the key value to the platform, and the fail-safe peripheral module is configured to signal the safe state depending on the implemented safety function, the key value, and the continuous counter value. In this embodiment, a data exchange that can ensure the protection of the safety function executed on the platform can be easily implemented.

[0028] In a further embodiment, the monitoring device is a standalone hardware component, particularly a hardware dongle, that can be removed from the platform by the user. In this embodiment, the monitoring device is also implemented independently of the platform on the hardware side and can be connected to it, for example, as a simple USB dongle. The standalone hardware and the encapsulation of the security function executed on the platform within this hardware enable independent certification of the monitoring device, which can be carried out particularly easily and cost-effectively. This design thus contributes to a particularly cost-effective implementation of the monitoring device.

[0029] In an alternative embodiment, the monitoring device is integrated into the fail-safe I / O module. "Integrated" in this context means that the individual functional groups of the monitoring device are implemented on the hardware of the fail-safe I / O module. In particular, components can be used jointly by the monitoring device and the fail-safe I / O module at low cost, for example, by the fail-safe I / O module and the monitoring device accessing common safe hardware. This embodiment thus contributes to the simple implementation of the automation system, since, in addition to the platform, only one separate component is required, on which the monitoring device and the fail-safe I / O module are integrated.

[0030] In a further embodiment, the first and / or second communication interface is a non-safe communication interface. In this embodiment, the communication interfaces do not represent safe units per se that could ensure safety-related communication, but are, for example, commercially available USB interfaces. In this case, safety is ensured independently of the communication, in particular through the interaction of the monitoring device with the fail-safe peripheral module. This embodiment has the advantage that a cost-effective standard communication interface can be used, which advantageously further reduces the costs of the overall system.

[0031] In a further embodiment, the monitoring device is configured to send an incorrect value, which does not correspond to the key value, as the key value at a defined time. By intentionally sending incorrect values ​​and reading back the outputs of the fail-safe peripheral module, the safe interaction of the monitoring device with the fail-safe peripheral module can be easily tested.

[0032] In a preferred embodiment, the monitoring device is configured to invert a bit of the key value, preferably in a rolling manner, to generate the incorrect value. The monitoring device thus cyclically checks whether incorrect numerical values ​​as the key value lead to an expected failure, preferably by inverting a bit of the correct key value in a rolling manner. For an 8-bit number, a stuck error can be detected in this way after eight test cycles at the latest. This embodiment thus advantageously contributes to improving security.

[0033] In a further embodiment, the fail-safe peripheral module comprises a decoding unit for generating an intermediate value from the continuous counter value and a random value, as well as a comparator unit for comparing the intermediate value with the key value. The fail-safe peripheral module can thus have complementary devices to the monitoring device in order to interact with it. In particular, the decoding unit and the comparator unit are implemented in a fail-safe manner. This embodiment contributes to the monitoring device being designed correspondingly simply, since the fail-safe peripheral module is adapted to the monitoring device. This division enables a cost-effective automation system for a high safety category.

[0034] In a further embodiment, the fail-safe peripheral module comprises a first and a second output circuit that can be activated and deactivated and on which the comparator unit acts alternately, so that the first or second output circuit is alternately activated, and wherein the fail-safe peripheral module signals a safe state when the first or second output circuit is activated. The fail-safe peripheral module has two output circuits that can be activated and deactivated like switches. Preferably, the output circuits are configured to deactivate themselves after a defined period of time following activation. For this purpose, the output circuits can each have, for example, a monostable multivibrator (also called a monoflop), on which the comparator unit acts alternately, so that at least one monostable multivibrator is triggered.The fail-safe peripheral module signals a safe state when at least one of the two monostable flip-flops is triggered. By using two output circuits that can be easily activated and deactivated, e.g., using monostable flip-flops, the fail-safe peripheral module can be advantageously simplified, as the otherwise conventional multi-channel redundant evaluation unit with two redundant processors can be omitted. Since monoflops or other switch-like components are many times cheaper than processors, this design contributes to a cost-effective implementation of the fail-safe module.

[0035] In a further embodiment, the fail-safe peripheral module is configured to generate a timeout signal via one of the two output circuits when an incorrect value is transmitted as a key value and simultaneously signal the safe state via the other output circuit. By controlling the output circuits so that one holds while the other is tested, the proper functioning of the monitoring device and the fail-safe peripheral module can be verified without actually having to transfer the process to the safe state. Such a test therefore advantageously does not limit the availability of the automation system.

[0036] In a further embodiment, the fail-safe peripheral module has a secure random number generator to generate a random value (S), or the fail-safe peripheral module is configured to receive a random value (S). The random value, which is added to the monitoring device externally, can thus be generated in particular by the fail-safe peripheral module, so that the random number generator can access the secure devices of the fail-safe peripheral module for its implementation. The implementation of a secure random number generator can thus be realized particularly easily.

[0037] In a further embodiment, the fail-safe I / O module is a standalone hardware component. This allows the fail-safe I / O module to be implemented independently of the platform and designed to be fail-safe. The fail-safe I / O module can thus be flexibly deployed across different platforms, while the inherent fault tolerance of the I / O module can be independently ensured. This design thus further contributes to a flexible design of the automation system.

[0038] In a further embodiment, the monitoring device and / or the fail-safe peripheral module are designed with multi-channel redundancy. A multi-channel redundancy design is a well-known approach in safety technology for implementing a fail-safe system. Since the monitoring device and the fail-safe peripheral module are designed as simply as possible, a multi-channel redundancy design can also be implemented particularly cost-effectively.

[0039] In a further embodiment, the fail-safe I / O module includes a fail-safe clock generator. A safe clock is regularly required for the implementation of various safety functions to ensure the fail-safe nature of various components. Because the safe clock is provided by the fail-safe I / O module, no additional safe component needs to be installed. This design thus also contributes to a flexible design of the automation system.

[0040] Embodiments of the invention are illustrated in the drawings and explained in more detail in the following description. They show: Fig. 1 is a schematic representation of an automation system according to a first embodiment of the present invention, Fig. 2 is a schematic representation of a safety model of an automation system according to a further embodiment of the present invention, Fig. 3 is a schematic representation of an architectural model of an automation system, Fig. 4 is a state diagram of an error routine of an automation system according to a preferred embodiment of the present invention, Fig. 5 is a schematic representation of an automation system according to a second embodiment of the present invention, Fig. 6 is two embodiment variants according to the second embodiment of the present invention, and Fig. 7 is a circuit diagram of a preferred embodiment of the automation system according to the invention according to the second embodiment.

[0041] In the Fig. 1 A first embodiment of the novel automation system is designated in its entirety by reference numeral 10. The automation system 10 here comprises hardware and software components that together form a safe automation system. Safe in this context means that a Safety Integrity Level 3 according to IEC 61508 can be achieved by the automation system according to the invention for an implemented safety function.

[0042] The automation system comprises a platform 12, which serves as a data processing device and computing unit. Platform 12 can be dedicated hardware 14, real or virtual computing systems 16, or infrastructure 18 located in a cloud. Platform 12 can also be composed of a combination of the aforementioned devices.

[0043] Platform 12 is a non-secure platform or is at least partially composed of non-secure components. "Non-secure" here means that the components of platform 12 do not have or cannot demonstrate sufficient inherent fault tolerance within the meaning of the relevant safety standard to implement a secure system on their own. Platform 12 can thus be composed, for example, of commercially available PC hardware and, in particular, include standard processors. Likewise, in another embodiment, the platform or components of the platform can also be provided in the form of cloud computing.

[0044] Platform 12 can consist of hardware and software components. For example, platform 12 can be a non-secure device, such as a PC or a Raspberry Pi, running a non-secure real-time operating system (RTOS), such as a Linux derivative.

[0045] The automation system 10 further comprises at least one fail-safe peripheral module 20 for fail-safe input from the process and fail-safe output to the process. The fail-safe peripheral module 20 enables the connection of the automation system to the process. The fail-safe nature of the peripheral module 20 and safety-related communication between the peripheral module 20 and the automation system 10 are achieved using principles known from safety technology and are not the subject of this disclosure.

[0046] The platform 12 and the fail-safe peripheral module 20 together form the substructure and the interfaces of the safe automation system 10, while a safe runtime environment 22 located on the platform and user programs 24, 26 implement the actual safety function 28.

[0047] The secure runtime environment 22 is a software layer located between the platform 12 and the application programs 24, 26, providing the application programs 24, 26 with secure resources 30 that are independent of the platform 12. The secure runtime environment 22 loads the application programs 24, 26, executes them on the platform, and coordinates their mutual links. The secure runtime environment 22 thus represents a small platform itself, with the aid of which the application programs 24, 26 are executed.

[0048] Preferably, the user programs 24, 26 are specifically designed for the platform 12, i.e., they are written in a programming language that can be converted into native machine code of the platform 12 for execution. The secure runtime environment 22 is therefore not an "operating system" for the execution of the user programs 24, 26, but merely encapsulates secure resources 30 for the user programs 24, 26, so that these can be integrated by the user programs 24, 26 without the user programs 24, 26 having to ensure secure execution of the secure resources 30. The secure runtime environment 22 can, in turn, be divided into platform-specific components and platform-independent components, with the secure resources 30, in particular, being implemented by platform-independent components, so that the secure runtime environment 22 can be easily ported to different platforms.

[0049] The safety function 28 can be any safety function, such as an emergency stop, two-hand control, or decentralized I / O. The safety function 28 is thus a safety-related control function implemented by the automation system. According to the invention, the safety-related control function is implemented by two user programs 24, 26 that are executed in parallel. In addition to their redundant execution, the user programs are diverse from one another, meaning that one and the same result is achieved by the user programs 24, 26 in different ways.

[0050] A computation is considered fully diverse if the mutual freedom from repercussions of the parts of a processor used for the computation, or the mutual freedom from repercussions of the different uses of the same parts of a processor, is verified. Such verification typically requires hardware tests, which can be omitted in coded processing within the meaning of SCP and may also be omitted in the automation system according to the invention if these hardware tests can be compensated for by additional tests and the secure resources 30 of the secure runtime environment and made externally verifiable.

[0051] In a preferred embodiment, the diversity between the first and second application programs 24, 26 is achieved by the second application program 26 working with diverse data from the first application program 24 and, as a result, using different instructions for the same calculation steps. The use of inverse data does not provide complete diversity, but it does provide sufficient diversity to enable adequate fault tolerance in conjunction with the secure runtime environment 22 and the tests 31 provided by the runtime environment 22.

[0052] A program version that works with inverse data (hereinafter referred to as an inverse user program) can be easily generated automatically from the native program version according to a fixed schema. The inverse user program can thus be executed with comparable effort to the native user program. Likewise, in contrast to a coded user program in the sense of SCP, the inverse user program can work with the same data types and use the same instruction set, since calculations with inverse data are based on the same arithmetic. An inverse user program is therefore less complex than a coded user program based on arithmetic coding, for example, AN coding. Furthermore, a tool chain for creating the inverse program version can be simplified from the native program version.The simplified tool chain enables rapid porting to other systems or the use of a different programming language. Overall, this increases the flexibility of the entire approach.

[0053] The invention is therefore based, among other things, on the assumption that, although execution errors in an application program can be revealed more clearly and reliably by coded processing according to SCP than by the diverse processing proposed here, the effort involved is not justified, since the application must always be integrated into a security model in order to achieve a high security rating.

[0054] It is therefore an idea, instead of increasing the intrinsic fault safety of individual components, such as the intrinsic fault safety of the user program through SCP, to take an overall approach and make the individual components of the automation system only as safe as necessary, so that overall a safe automation system is created that can execute the safety function according to the highest possible safety level of the relevant standard. One focus of the present invention is therefore on designing the individual components of the automation system, i.e. in particular the user programs 24, 26, the safe runtime environment 22 and the platform 12, as simple and flexible as possible, so that a balanced compromise between safety, complexity and portability is achieved at all levels of the automation system.

[0055] Fig. 2 shows a schematic representation of a safety model of an automation system according to a preferred embodiment of the present invention.

[0056] The secure runtime environment 22 is based on a non-secure device 32 with a non-secure real-time operating system 34. The non-secure device 32 and the non-secure real-time operating system together form the platform 12. The secure runtime environment 22 is a software layer that, on the one hand, provides secure resources 30, independent of the platform 12, for the user programs 24, 26 in a resource layer 36 and, on the other hand, forms a local security instance 38 to monitor the user programs 24, 26 and the security function implemented by them. The user programs 24, 26 are connected to the process to be monitored via a secure peripheral module 20.

[0057] The safe runtime environment 22 can, in turn, be monitored by a safe device. The safe device (hereinafter also referred to as Safety Pacemaker (SPM) 40) can be a simple hardware dongle, e.g., in the form of a commercially available USB stick, with one or more simple microcontrollers, such as a PIC and / or AVR. It is understood that a hardware dongle is not limited to a USB interface; the SPM 40 can also be connected via Ethernet, RS232, or another communication interface.

[0058] The SPM 40 implements a second, preferably external, security instance 41. The second security instance 41 tests the secure runtime environment 22, in particular the first local security instance 38, through targeted error injection. The error injection occurs via services and manipulators of the secure runtime environment 22, which are activated by the external security instance 41. In this way, data comparators, a system clock, and timer tests of the secure runtime environment 22 can be checked, and errors during the execution of the user programs 24, 26 can be revealed without having to consider and / or implement corresponding tests during the development of the user programs 24, 26. The first user program 24 can thus be written in a single channel, without the application programmer having to consider the security of the user program.A second application program 26, which is diverse from the first application program, can be generated automatically using a tool chain. The application developer can thus focus exclusively on implementing the actual safety function.

[0059] In a preferred embodiment, the SPM 40 can additionally provide a safety clock 42, in particular to implement a second shutdown path on the safe peripheral module 20.

[0060] Fig. 3 shows an architectural model in which an embodiment of the automation system 10 is embedded.

[0061] The architectural model essentially comprises three components: the automation system 10, an external safety unit, such as the aforementioned SPM 40, and a communication interface 44 between these two components.

[0062] All three components can have a hardware-neutral layer 46 and a hardware-specific layer 48. In the automation system 10, the secure runtime environment 22 decouples the hardware-specific layer 48 from the hardware-neutral layer 46 with a hardware-specific component, a so-called System Abstraction Layer (SAL). The SAL 50 abstracts the secure runtime environment 22 from an operating system 52 and the associated hardware 54. Special hardware 56, which is addressed by operating system drivers 58, can be abstracted by independent SAL drivers 60.

[0063] On the hardware-neutral layer 46, the automation system 10 implements a resource layer 36. Resource layer 36 provides the secure resources and, as shown here, is preferably duplicated and diverse. Resource layer 36 can provide timers and the process data for each channel. Furthermore, resource layer 36 can provide safety-related functionalities, such as secure CRC calculation.

[0064] An application manager 62a, 62b is located on the resource layer 36, preferably separately for each user program 24, 26. This manager checks the consistency of the two user programs 24, 26 when they start up and checks their CRC. The application manager 62a, 62b is also preferably responsible for cyclical calling of the user programs, their execution monitoring, and for providing an API. The API can contain functions for setting outputs, reading inputs, and starting and reading application timers. The application manager 62a, 62b is also preferably duplicated and diverse, as shown here.

[0065] The secure runtime environment 22 of the automation system 10 further implements a local safety instance 38. The local safety instance 38 performs the necessary safety tasks. This includes RAM and ROM tests, cycle and sequence monitoring, and timer tests. Furthermore, the local safety instance 38 performs the consistency check of the process data of the user programs 24, 26. For this purpose, the local safety instance 38 preferably runs only once on the system and can be checked externally via the SPM 40.

[0066] The SPM 40 is a fail-safe device with safe hardware, which is formed, for example, by two redundant processors 64a, 64b, which enable the safe execution of an application in a conventional manner. Since the SPM does not execute any user programs for control purposes in the sense of the user programs of the automation system 10, the hardware can be designed particularly simply. In particular, special safe CPUs can be used for the processors 64a, 64b, for which proof of fail-safety can be practically provided, for example, through FMEA.

[0067] A second safety instance 41 is executed on the SPM 40. This is preferably implemented in a hardware-neutral manner and decoupled from the hardware by an abstraction layer 66. The second safety instance 41 can communicate with the first safety instance 38 of the automation system 10 via the communication interface 44.

[0068] The communication interface 44 can be a single-channel communication interface without special safety precautions. Communication security can be ensured by the interaction of the local safety instance 38 of the automation system 10 and the second safety instance 41 of the external safety device.

[0069] In a preferred embodiment, the secure runtime environment 22 and the SPM 40 communicate with each other in a fixed cycle with timeout monitoring. Upon request from the SPM 40, data (e.g., timer values, process control, etc.) is deliberately falsified at certain code locations. For this purpose, the secure runtime environment 22 has manipulators that can be activated by the SPM 40, for example, by the SPM 40 sending a corresponding manipulator ID to the secure runtime environment 22. The SPM 40 then expects a suitable error message from the secure runtime environment 22. In other words, the SPM 40 has certain expectations of the secure runtime environment 22 after a manipulator ID has been transmitted.

[0070] The safe runtime environment 22 must not exhibit any errors detectable by the user due to the manipulators, nor must it disable the outputs due to deliberate tampering. Therefore, the error must be intercepted by the safe runtime environment 22. This concept allows the system behavior to be tested at the real point in the program.

[0071] Fig. 4 shows in a state diagram the possible states that SPM 40 and the automation system 10 can assume during a check.

[0072] In the normal state A, the secure runtime environment 22 waits for an error in order to react accordingly. If the secure runtime environment 22 receives a manipulator ID, preferably including a defined bit mask, from the SPM 40 in this state, the corresponding manipulator is activated. The secure runtime environment 22 then transitions to state B, in which it is desensitized to the corresponding error. This is followed by the manipulation of the data to be processed to simulate an error. Preferably, the data is exclusively ORed with the received bit mask so that the faulty bit position can be rolled by the SPM 40.

[0073] Upon detection of the manipulation, the safe runtime environment 22 must generate the appropriate error message, which the SPM 40 waits for. If this expectation is not met, the safe runtime environment 22 and / or the SPM 40 can trigger a safe shutdown, for example, through a second shutdown path. In a preferred embodiment, the second shutdown path can be implemented through the interaction of the SPM 40 and the safe peripheral module 20. However, it is understood that other shutdown paths are also conceivable.

[0074] If the test is successfully completed, the secure runtime environment 22 returns to normal state A. If the error actually occurs during desensitization, it will be intercepted by the desensitization, but the test would not be successful, so that an unintercepted error message is transmitted to the SPM 40.

[0075] With the external tests performed by the SPM 40, all single-channel tests executed in the local safety instance can be separately verified. For example, the following single-channel tests can be verified with the external tests: input data comparisons, output data comparisons, RAM tests, ROM tests, timer tests, cycle monitoring, execution monitoring of the safe runtime environment 22, and execution monitoring of the user programs 24 and 26.

[0076] Fig. 5 shows a schematic representation of an automation system according to a second embodiment of the present invention.

[0077] The automation system according to the second embodiment is designated here in its entirety by reference numeral 100. Parts with the same reference numerals denote the same parts as in the first embodiment and will not be explained again below.

[0078] The automation system 100 consists of three individual components.

[0079] The first component is an automation system 10, which is based on a non-safe platform 12. In particular, the automation system is configured to implement a safety function 28 with the aid of two user programs 24, 26. The first component can thus comprise an automation system as described with reference to the first embodiment. However, it is understood that the first component is not limited to this configuration; rather, any automation system is suitable as the first component.

[0080] The second and third components form the monitoring device 70 and the fail-safe peripheral module 20. The monitoring device 70 can in particular be the previously described SPM, which was designated above with the reference number 40.

[0081] The monitoring device 70 and the fail-safe peripheral module are coupled to the automation system 10 via a first communication interface 72 and a second communication interface 74. In addition to communication with the automation system, the first communication interface 72 and the second communication interface 74 also enable communication between the monitoring device 70 and the fail-safe peripheral module 20. In other words, the monitoring device 70 can communicate with the fail-safe peripheral module 20 via the platform 12. The first and second communication interfaces 72, 74 are, in particular, single-channel communication interfaces, such as a commercially available USB interface. It is understood that other communication interfaces, such as Ethernet, RS232, etc., can equally be considered.

[0082] The monitoring device 70 is preferably, as shown here, a standalone hardware component configured to provide fail-safe services 76. The fail-safe services 76 can be implemented in software or hardware, with intrinsic fault tolerance being ensured by appropriate measures, for example, by the services being designed with multi-channel redundancy or by executing them on multi-channel redundancy hardware in a manner known per se, so that fail-safe operation of the services 76 can be ensured and tested. It is understood that the fail-safe configuration is not limited to the form shown here, but other measures known to those skilled in the art for achieving intrinsic fault tolerance are also conceivable.

[0083] The fail-safe services 76 are preferably simple services, such as, for example, a counter unit 78 and a coding unit 80. The fail-safe services 76 can thus be implemented on hardware that is simple and relatively uncomplex. The hardware of the monitoring device 70 can, for example, comprise one or more simple microcontrollers, ASICs, or similar computing units, or can be composed of discrete electrical components. In particular, when using a microcontroller, its command set is as limited as possible, so that certification with regard to intrinsic fault tolerance is possible cost-effectively. In a preferred embodiment, the hardware equipment of the monitoring device 70 is limited to being able to execute the counter unit 78 and coding unit 80.

[0084] The fail-safe services 76 interact with the fail-safe peripheral module 20 so that, regardless of the implementation of the safety function 28, safe shutdown can be ensured even in the event of a fault in the non-safe platform 12.

[0085] In a preferred embodiment, the monitoring device 70 itself does not have any further safety-related devices, i.e., in particular, no dedicated safe outputs that could directly trigger an additional shutdown path for the safety-critical process. The coupling to the safety-critical process preferably occurs exclusively via the fail-safe peripheral module 20.

[0086] In a particularly preferred embodiment, the monitoring device 70 additionally has a test device 82 in the form of a safety instance, which interacts with a safe runtime environment 22 implemented on the platform 12 to provide additional test facilities for the user programs 24, 26. Preferably, the safe runtime environment 22, as previously described in connection with the first embodiment, is implemented in a user-program-independent manner and is identical for different automation systems 10. Thus, with a uniform test device 82, the safety of different safety functions 28 on different platforms 12 can be tested independently, wherein the monitoring device 70 itself, as a last resort, together with the fail-safe peripheral module, can transfer the process to a safe state.The monitoring device 70 can thus be used in various automation systems without additional adaptations. Likewise, the monitoring device 70 can be replaced with another one without major effort in the event of a defect.

[0087] According to the second embodiment, the essential aspect of the invention is thus the division of the safe control system into various, particularly hardware-independent, components. The focus of the first component is to provide as much computing capacity as possible for implementing the safety function 28 and to enable simple development of the associated user programs 24, 26. The focus of the second and third components, in contrast, is to design the hardware as simply as possible so that the fault tolerance of the provided services 76 can be reliably and cost-effectively ensured.

[0088] Fig. 6 shows two embodiments of an automation system according to the second embodiment of the present invention.

[0089] In the first variant (in the image above), the three components of the automation system 100 are implemented as independent hardware components. Both the monitoring device 70 and the fail-safe peripheral module 20 are each configured as a hardware component. A first communication interface 72 connects the monitoring device 70 to the automation system 10, and a second communication interface 74 connects the fail-safe peripheral module 20 to the automation system 10. Communication between the monitoring device 70 and the fail-safe peripheral module 20 thus preferably takes place exclusively via the automation system 10.

[0090] The design with separate components has the advantage that they can be manufactured and sold separately. For example, a monitoring device 70 can be used for a variety of different peripheral modules 20. The automation system 100 can thus be adapted to the respective requirements, with only the peripheral module 20 needing to be replaced.

[0091] In the second variant (in the image below), the monitoring device 70 and the fail-safe module 20 are implemented as a shared hardware component. A first logical communication interface 72 and a second logical communication interface 74 are implemented here through a single physical connection to the automation system 10. Despite the shared integration on one piece of hardware, communication between the monitoring device 70 and the fail-safe peripheral module 20 takes place via the first and second communication interfaces 72, 74, i.e., here too, communication between them takes place via the automation system 10. In other words, the monitoring device 70 and the fail-safe peripheral module 20 are two independent logical units despite the integration.

[0092] Advantageously, this embodiment allows the monitoring device 70 and the fail-safe module 20 to use a common safe hardware basis. For example, functions of the fail-safe peripheral module and the monitoring device 70 can be implemented via shared processors that operate redundantly across multiple channels. Likewise, in this embodiment, one physical communication interface is sufficient to implement the first and second communication interfaces. In this way, a safe automation system 100 within the meaning of the present invention can be implemented particularly favorably.

[0093] Fig. 7 shows a circuit diagram of a preferred embodiment of the automation system according to the invention according to the second embodiment.

[0094] The automation system 100 here comprises a monitoring device 70, an automation system 10, and a fail-safe peripheral module 20, each as a separate hardware component. In this exemplary embodiment, communication between the individual components takes place via a USB interface. This means that the first communication interface 72 and the second communication interface 74 are both USB connections.

[0095] The monitoring device 70 further comprises a counter unit 78 and a coding unit 80 for implementing the fail-safe service 76. The counter unit 78 can continuously increment an internal counter and provide a counter value C to the coding unit 80. Furthermore, the counter value C can be transmitted to the fail-safe peripheral module 20 via the communication interface 72, 74.

[0096] The coding unit 80 receives a random value R from the automation system 10. The random value R is generated here by a random number generator 86 within the fail-safe I / O module 20. From the random value R and the counter value C, the coding unit 80 generates a key value S and transmits it to the fail-safe I / O module 20 via the automation system 10.

[0097] The monitoring device 70 checks whether the random value, which is preferably transmitted cyclically by the automation system 10, changes over time. Since the key value S is based on both the random value R and the continuous counter C, it can be ensured that all numbers are run through at least once. This eliminates the need for the monitoring device 70 to ensure that a random generator that generates the random value actually runs through all values. Such an evaluation would be computationally intensive and thus difficult to implement in a preferred embodiment of the monitoring device 70 with only limited computing capacity.

[0098] The fail-safe peripheral module 20 has complementary units to those provided in the monitoring device 70. These can be designed as dedicated units or implemented in the form of software modules.

[0099] The fail-safe peripheral module 20 comprises a decoding unit 88 and a comparator unit 90 as complementary units. The decoding unit 88 receives the continuous counter value C and a copy of the random value R and generates a comparison value S' from them. If the received key value S and the generated comparison value S' are identical, the fail-safe peripheral module 20 signals a safe state. In a preferred embodiment, the decoding is known only to the decoding unit 88 of the fail-safe peripheral module 20, and the coding is known only to the coding unit 78 of the monitoring device.

[0100] In the preferred embodiment shown here, the safe state is signaled with the aid of a first output circuit 91 with a first monostable multivibrator 92 and a second output circuit 93 with a second monostable multivibrator 94. The comparator 90 acts alternately via a switch 96 on the two output circuits 91, 93, i.e. here on the monoflops 92, 94. The fail-safe peripheral module 20 indicates a safe state when at least one of the two monoflops 92, 94 is triggered. During normal operation, when the comparator 90 continuously supplies a signal, i.e. a comparison between the key value S and the generated intermediate value S' is continuously successful, both monoflops 92, 94 are generally triggered, i.e. the cycle time within which new keys are sent is correspondingly coordinated with the hold time of the monoflops 92, 94.

[0101] To test whether the fail-safe peripheral module 20 correctly signals the safe state, the monitoring device 70 generates an incorrect key value by inverting a bit of the correct key value S. As a result, the comparator 90 detects an inequality between the key value S and the generated intermediate value S' and interrupts the signal to the switch 96.

[0102] As a result, one of the two monoflops initially drops out and no longer provides a signal at the output. This can be detected by the logic gate 98 following the monoflops, whereupon a timeout signal 102 is generated. The timeout signal 102 can be read back by the monitoring device 70, allowing it to determine whether an incorrect numerical value as the key value S leads to an expected drop in one of the two monoflops.

[0103] Since initially only one monoflop drops out during the test, the fail-safe peripheral module continues to signal a safe state provided the dropped monoflop is retriggered promptly by a correct key value S. If retriggering does not occur in a timely manner, the second monoflop also drops out, and the fail-safe peripheral module 20 signals a non-safe state. In the present exemplary embodiment, signaling a non-safe state would result in the safe outputs 104 being switched off. It is understood that the configuration with monoflops 92, 94 represents only one possible configuration of the output circuits 91, 93. The two branches can also be configured differently and merely need to be activatable and deactivatable. The output circuits are preferably configured to deactivate themselves after a defined period of time from activation.Alternatively, deactivation from outside is also conceivable.

[0104] In addition to the devices that implement the watchdog principle presented here, the fail-safe peripheral module 20 can have additional test devices that can be combined with the watchdog protection. For example, in addition to a first input register, two further input registers 108, 110 can be provided, which contain a previous and a subsequent value, based on which address corruption can be detected. Furthermore, the fail-safe peripheral module 20 can have a safe clock generator 112, which provides a safe clock that can be used for various safety-related applications. One such application can be, for example, a voltage monitoring device 114, which monitors a safe power supply and can, if necessary, influence the safe outputs 104 of the fail-safe peripheral module 20.It is also conceivable to implement a restart interlock 116, which prevents the safety-critical process from restarting on its own in the event of a fault.

[0105] It is understood that the implementation presented here is only one example of how an automation system 100 can be constructed in accordance with the invention. It is understood that other embodiments may also be encompassed by the subject matter of the invention. In principle, the scope of protection of the present application is defined by the claims and is not limited by the features shown in the description or the figures.

Claims

1. An automation system (100) for monitoring a safety-critical process, comprising - a non-failsafe platform (12) for executing user programs (24, 26), which implement a safety function (28), - a fail-safe peripheral module (20), via which the safety-critical process can be coupled to the user programs (24, 26), and - a monitoring device (70) configured to ensure fail-safe execution of the safety function (28) on the non-failsafe platform (12), wherein the monitoring device (70) is couplable to the non-failsafe platform (12) via a first communication interface (72) and the fail-safe peripheral module (20) is couplable to the non-failsafe platform (12) via a second communication interface (74), wherein the monitoring device (70) is configured to run at least one fail-safe service (76) independently of the non-failsafe platform (12) in order to exchange data generated cyclically by the fail-safe service (76) with the fail-safe peripheral module (20) according to a defined scheme, and wherein the fail-safe peripheral module (20) is configured to signal a safe state based on the implemented safety function (28) and based on the data exchange which takes place via the non-failsafe platform (12).

2. The automation system according to claim 1, wherein the fail-safe service (76) comprises a counter unit (78) configured to generate a continuous counter value (C) and an encoding unit (80) configured to generate a key value (S), in particular wherein the encoding unit (80) is configured to generate the key value (S) from a random value (R) and the continuous counter value (C).

3. The automation system according to claim 2, wherein the first communication interface is further configured to forward the continuous counter value (C) and the key value (S) to the non-failsafe platform (12), and wherein the fail-safe peripheral module (20) is configured to signal the safe state based on the implemented safety function, the key value (S) and the continuous counter value (C).

4. The automation system according to any one of claims 1 to 3, wherein the monitoring device (70) is a stand-alone hardware component detachable from the non-fail-safe platform (12) by the user, in particular a hardware dongle.

5. The automation system according to any one of claims 1 to 3, wherein the monitoring device (70) is integrated into the fail-safe peripheral module (20).

6. The automation system of any one of claims 1 to 5, wherein the first and / or second communication interface (72, 74) is a non-failsafe communication interface.

7. The automation system according claim 2, wherein the monitoring device (70) is configured to transmit a false value that does not correspond to the key value (S) as the key value (S) at a defined time, in particular wherein the monitoring device (70) is configured to invert a bit of the key value (S) to generate the false value, preferably in a rolling manner.

8. The automation system according to any one of claims 2 to 7, wherein the fail-safe peripheral module (20) comprises a decoding unit (88) configured to generate an intermediate value (S') from the continuous counter value (C) and a random value (R), and a comparator unit (90) configured to compare the intermediate value (S') with the key value (S).

9. The automation system according to claim 8, wherein the decoder unit (88) and the comparator unit (90) are implemented in a fail-safe manner, in particular wherein the fail-safe peripheral module (20) comprises a first output circuit (91) and second output circuit (93) that are activatable and deactivatable and on which the comparator unit (90) acts in an alternating manner, so that one of the first and second output circuits (91, 93) is alternately activated, and wherein the fail-safe peripheral module (20) signals a safe state when one of the first and second output circuits (91, 93) is activated and wherein preferably the fail-safe peripheral module (20) is configured, when an incorrect value is transmitted as a key value (S), to generate a time-out signal via one of the two output circuits (91, 93) and to simultaneously signal the safe state via the other output circuit.

10. The automation system according to any one of claims 1 to 9, wherein the fail-safe peripheral module (20) comprises a safe random number generator (86) to generate a random value (R) or is configured to receive a random value (R).

11. The automation system according to any one of claims 1 to 10, wherein the fail-safe peripheral module (20) is a stand-alone hardware component.

12. The automation system according to any one of claims 1 to 11, wherein the monitoring device (70) and / or the fail-safe peripheral module (20) is of multi-channel redundant design and / or wherein the fail-safe peripheral module (20) comprises a fail-safe clock generator (112).

13. A method for monitoring a safety-critical process with an automation system, comprising the steps: - providing a non-failsafe platform (12) for executing user programs (24, 26) that implement a safety function (28); - coupling of the safety-critical process with the user programs (24, 26) via a fail-safe peripheral module (20); - ensuring fail-safe execution of the safety function (28) on the non-failsafe platform (12) by a monitoring device (70); wherein the monitoring device (70) is couplable to the non-failsafe platform (12) via a first communication interface (72) and the fail-safe peripheral module (20) is couplable to the non-failsafe platform (12) via a second communication interface (74), wherein the monitoring device (70) runs at least one fail-safe service (76) independently of the non-failsafe platform (12) in order to exchange data generated cyclically by the fail-safe service (76) with the fail-safe peripheral module (20) according to a defined scheme, and wherein the fail-safe peripheral module (20) signals a safe state based on the implemented safety function (28) and based on the data exchange which takes place via the non-failsafe platform (12).