COMPUTER-ASSISTED CONFIGURATION OF A TECHNICAL SYSTEM

Field of the Invention

The present invention relates to a computerized method for configuring a technical system comprising a sequence of sys tem components, whereby each sequence of system components can be assigned to a sequence pattern and whereby the se quence comprises at least one system component. The invention further specifies a computer program product and a computer- readable storage medium.

Background of the Invention

Creation and configuration of technical systems (e.g. in form of technical projects, e.g. engineering projects) is a very complex and time-consuming task, requiring vast amounts of domain-specific knowledge. Due to the sheer number of steps involved in configuring a typical technical system, lack of training data, as well as computational complexity, it is in feasible to automate the whole process. However, it is possi ble to automate simple steps typically performed by an engi neer working on a specific type of technical systems.

A system for enabling the training of personalized software components (which can be refer to as "assistants") that would take over the task of performing simple steps in the process of configuring/designing a technical system for a given user within a given engineering tool would facilitate the process a lot. An example of a step that such an assistant could take over is the introduction of a specific programming block into the system (e.g., checking the air pressure of a pneumatic element) given certain pre-conditions (e.g., presence in the system of pneumatic elements) . The state of the art are two approaches to fulfil the pro posed functionality:

a) Users can manually define each rule that they want to be applied throughout their technical projects/ technical sys tems, which is a lot more labour-intensive approach.

b) An engineering platform for technical projects/ technical systems can be supplied with a recommendation engine main tained by the platform development team. In this case, the recommendations are not personalized, and the user has no control over the quality of recommendations.

Summary of the Invention

The objective of the present invention is to improve and fa cilitate the configuration of technical systems.

The invention is given by the features of the independent claims. Advantageous embodiments are provided in the depend ent claims. Further features, application possibilities, and advantages of the invention result from the following de scription .

According to the invention the objective is accomplished by a method automatically introducing system components of a tech nical system during the configuration of the technical system based on previous system configurations used to train the method. The method is assisting a user during the configura tion of the technical system.

The invention claims a computerized method for configuring the construction/ creation of a technical system/ designing a technical system comprising a sequence of system components, whereby the sequence of system components is assignable to a sequence pattern and whereby the sequence (Pi) of system com ponents (Si) comprises at least one system component.

The method comprises the following steps:

- specifying a support threshold for sequence patterns by a user, - starting the configuration of the technical system by choosing the sequence of system components by the user,

- analysing the chosen sequence of system components and ex tracting sequence patterns by an automation unit, whereby an alysing the chosen sequence of system components includes a pattern mining procedure and

- introducing at least one further system component to the chosen sequence of system components based on the extracted sequence pattern and according to the support threshold by the automation unit.

According to the embodiment the automation unit is trained by the following steps in advance and while used:

- configuring the technical system by choosing the sequence of system components by the user,

- storing the chosen sequence of system components in a data base, and

- training the automation unit by using the sequence of sys tem components stored in the database.

The advantage of this embodiment of the method is a personal ization for a given user within a given engineering tool for configuration of a technical system.

Summarized, the method supports the user with configuring the construction/ creation of the technical system based on the chosen sequence of system components and the at least one further introduced system component.

The purposed method has the advantage to automate simple steps typically performed by an engineer/ user working on the configuration and construction of a specific type of tech nical system/ engineering projects.

According to a further embodiment the system components in clude technical operations and/ or hardware units. An example of a step that the automation unit (or assistant) could take over is the introduction of a specific system component in form of a programming block into the technical system (e.g., checking the air pressure of a pneumatic element) given cer tain preconditions (e.g., presence in the system of pneumatic elements) .

According to a further embodiment at least one equivalence relation function between sequence patterns is specified by the user before starting the configuration of the technical system.

With no equivalence relation function specified, the method requests exact equivalence of the sequence of system compo nents for sequence patterns to be equal. No specified equiva lence relation function is equal to a trivial equivalence re lation function. In this case the method requests exact equivalence of the sequence of system components for sequence patterns to be equal. This means a trivial sequence pattern equivalence relation function specification requests a common first system component directly followed by a common second system component for sequence patterns to be equal.

According to a further embodiment a sequence pattern equiva lence relation function specification is configured to re quest at least one equal first system component directly or non-directly followed by at least one equal second system component for sequence patterns to be equal (non-trivial equivalence relation function) .

Further, the method allows for automatically detecting parts of the user' s workflow that could be performed automatically by the automation unit.

According to a further embodiment the support threshold of a sequence pattern is a value that indicates how frequent a se quence pattern must be in the database that the automation unit introduces / for the automation unit (A) to introduce the at least one a further system component based on the ana lysed sequence pattern. Each sequence of system components chosen by the user is stored in the database with its occurrence (during training, see above) . The occurrence is used to check if the support threshold is reached. If the support threshold is met the au tomation unit introduces a system component as response to the sequence pattern chosen by the user.

Per each new/ yet uncovered sequence pattern chosen by the user, the automation unit saves a new introduction. Optional ly, the user can preciously be asked to validate the automat ic introduction of the system component based on a specific analysed sequence pattern in future configurations.

According to a further embodiment the sequence pattern de scribes a series of successive system components.

According to a further embodiment the database is organized in form of a search tree. The search tree keeps all discov ered sequence patterns and their current support measures (number of occurrence) . In case the user creates a new engi neering project for a new technical system, the method will not re-calculate the whole search tree from scratch, but it will rather update the existing search tree. Update of the search tree includes re-calculation of support measures for sequence patterns already in the search tree as well as fur ther growth of the search tree. Depending on the choice of the equivalence relation function, the search tree might look differently.

According to a further embodiment the database stores the se quences of system component of one user or of multiple users. This means that analyzation of the pattern of the sequence of system components chosen by the user in combination with the database by the automation unit is done based on database da ta of one user (personalization) or of multiple (similar) us- ers . If it is only stored for one user, the method has the ad vantage of a personalization for the given user within a giv en engineering tool for configuration of a technical system.

Given consent, trained automation unit may be shared among similar users (e.g., working on similar technical systems (projects, same application area, same business unit, etc.)) . This has the advantage that similarities between technical systems lead to synergy effects.

According to a further embodiment the technical system is an engineering system, an automatization system, a production system and/ or an industrial system and/ or includes a tech nical project on the technical system.

The present invention further claims a computer program prod uct comprising instructions which, when the program is exe cuted by a computational device, cause the computational de vice to carry out the steps of the inventive method.

The present invention further claims a computer-readable storage medium comprising instructions which, when executed by a computational device, cause the computational device to carry out the steps of the inventive method.

Compared to the state of art, the purposed method is substan tially less labour-intensive. Instead of asking the users to manually define the rules to be applied throughout future projects, the purposed method automatically detects parts of the user' s workflow that could be automated while the user continues to interact with the engineering software a famil iar way.

Further, state of the art methods involve a recommendation unit trained on aggregated information from various users and thus provides general recommendations, while the purposed method learns the steps that are customary to a specific us- er . Summarized the purposed method has the potential of reducing the time needed for designing/configuring a technical project by automating simple and repetitive steps performed by users, as well as reduce the frustration levels resulting from per forming such repetitive steps. In addition, the purposed method may reduce the number of errors introduced into tech nical projects due to forgetfulness of the responsible engi neer .

Further benefits and advantages of the present invention will become apparent after a careful reading of the detailed de scription with appropriate reference to the accompanying drawings .

Brief Description of the Drawings

Fig. 1 shows a flow diagram of the purposed method,

Fig. 2 shows sequences of system components,

Fig. 3 shows a trivial sequence pattern equivalence rela tion function specification,

Fig. 4 shows a non-trivial sequence pattern equivalence relation function specification,

Fig. 5 shows a search tree resulting from sequence pat tern mining, and

Fig. 6 shows the overall workflow of the proposed method.

Detailed Description of the Invention

Fig. 1 shows a flow diagram of the purposed computerized method for configuring a technical system comprising a se quence of system components, whereby each sequence of system components of system components can be assigned to a sequence pattern and whereby the sequence comprises at least one sys tem component .

The method comprises the following steps:

- Method step Ml: specifying a support threshold for sequence patterns and optionally at least one equivalence relation function between sequence patterns by a user,

- Method step M2: starting the configuration of the technical system by choosing the sequence of system components by the user,

- Method step M3: analysing the chosen sequence of system components and extracting sequence patterns by an automation unit, whereby analysing (M3) the chosen sequence of system components includes a pattern mining procedure and

- Method step M4 : introducing at least one further system component based on the extracted sequence pattern and accord ing to the support threshold (and the optional equivalence relation function) by the automation unit.

Every project of configuring a technical system can be repre sented as a sequence of operations performed consecutively/ sequences of system components PI, P2, and P3 chosen by a us er (see Fig . 2 ) .

Fig. 2 shows exemplary sequences of system components PI, P2, and P3 chosen by a user, e.g. an engineer in the process of configuring an engineering project. Each sequences of system components PI, P2, and P3 has multiple system components SI to S9. The shadings of the circles indicate the type of the system component/ action that the user chooses.

Fig. 3 shows a trivial sequence pattern equivalence relation function specification T. The trivial sequence pattern equiv alence relation function specification T requests exact equivalence of the sequence of system components for sequence patterns to be equal. This means a trivial sequence pattern equivalence relation function specification T requests a com mon first system component directly followed by a common sec ond system component for sequence patterns to be equal. Sys- tem component in between of the first system component and the second system component are not allowed. The trivial se quence pattern equivalence relation function specification T can be used as default.

Fig. 4 shows a non-trivial sequence pattern equivalence rela tion function specification NT. The non-trivial sequence pat tern equivalence relation function specification NT requests at least one equal first system component (shaded with in the example of Fig. 4) directly or non-directly followed by at least one equal second system component (shaded with "+" in the example of Fig. 4) for sequence patterns to be equal.

In the example of Fig. 4, the user wants all sequence pat terns that start with the specified first system component and end with the specified second system component steps to be equivalent in case they are separated by not more than 1 other system component. A support threshold is a value that indicates how frequent a sequence pattern must be in a pro ject database that an automation unit introduce a system com ponent based on a sequence pattern analysed. For example, the support measure (measure to be compared with the support threshold) of the pattern shown in Fig. 3 (shaded with shaded with "+") is one third in the project database depict ed in Fig. 2 if the user chooses the trivial sequence pattern equivalence relation function specification T (Fig. 3) . How ever, if the user chooses the non-trivial sequence pattern equivalence relation function specification NT (Fig. 4), then the support measure of this pattern is 1.

Fig. 5 shows a search tree ST resulting from sequence pattern mining on a collection of configurations of technical systems (projects) in the database DB . The search tree has multiple layers LI to L3. Each further layer includes one further sys tem component .

The method of the invention monitors the sequence of opera tions performed consecutively/ sequences of system components chosen by a user in the engineering software/ tools and ac tively maintains the search tree ST. The search tree ST keeps all discovered sequence patterns and their current support measures (measure to be compared with the support threshold) . In case the user creates a new engineering project for a new technical system, the method of the invention will not re calculate the whole search tree ST from scratch, but it will rather update the existing search tree ST. Update of the search tree ST includes re-calculation of support measures for sequence patterns already in the search tree ST as well as further growth of the search tree ST. Depending on the choice of the equivalence relation function, the search tree ST might look differently. For example, if the user chooses the trivial sequence pattern equivalence relation function specification (Fig. 3), then e.g. the right pattern in layer 2 L2 and the right pattern in layer 3 L3 in Fig. 5 will be treated as separate by the method of the invention. However, if the user chooses the non-trivial sequence pattern equiva lence relation function specification (Fig. 4), the method of the invention will not include sequence the right pattern in layer 3 L3 in the search tree ST, since it is equivalent to the right pattern in layer 2 L2.

Fig. 6 shows the overall workflow of the training within the proposed method. A user U works with an engineering software ES . The user U choses/ defines a new sequences of system com ponents P/ new project for a new technical system. The chosen sequence of system components P is stored in a database DB . The database DB is used to train an automation unit A in a "training gym" (= training algorithm) G. The trained automa tion unit A is used afterwards from the engineering software ES to support the user U with configuring technical systems. List of Reference Signs

A automation unit

DB database

ES Engineering software

G Training gym

Li layer i ( i e [ 1 ; 3 ] )

Mi Method step i (i e [1; 5])

NT non-trivial sequence pattern equivalence relation function specification

Si system component i (i e [1; 9])

ST search tree

T trivial sequence pattern equivalence relation func tion specification

U user

P new sequences of system components/ new project for a new technical system

Pi sequences of system components i (i e [1; 3])