WO2017042780A1

WO2017042780A1 - System and method for the detection of alerts and correction of errors in industrial processes

Info

Publication number: WO2017042780A1
Application number: PCT/IB2016/056373
Authority: WO
Inventors: Ignacio CHANG; Nicholas BELIZ; Mauricio Martinez; Edgardo SANTANA
Original assignee: Universidad Tecnológica De Panamá
Priority date: 2016-07-20
Filing date: 2016-10-24
Publication date: 2017-03-16

Abstract

System and method for the detection of alerts and correction of errors in industrial processes. The system takes as input the values measured by a network of sensors that form a monitoring system. The values measured are compared with reference values to generate an alert if there is discrepancy between both values. If there is not an already existing control law to deal with the anomaly, the system generates and stores a new control law using artificial intelligence techniques, with an automatic learning process to resolve similar future alerts. The model is adaptable, adjustable, flexible and dynamic with the system, facilitating its replicability and allowing its operation to not be affected by the acquisition of any component or that it cannot be connected to the model. This achieves a smart control and monitoring system.

Description

SYSTEM AND METHOD FOR THE DETECTION OF ALERTS AND CORRECTION OF ERRORS IN INDUSTRIAL PROCESSES

FIELD OF THE INVENTION The present invention is included in the field of automated systems for detection of risk alerts and alarms in industrial environments and correction of the errors detected.

BACKGROUND OF THE INVENTION

Companies are in a changing environment in all areas, both technological and management. This entails that an important effort must be made to adapt as quickly as possible to the new situations to continue to be competitive and efficient. Industrial facilities are a focus of possible accidents and risks that must be considered to decrease them. Thus, new management approaches have appeared on which interests are centred, such as the continued improvement of products, processes and in general of all systems. The automation of the detection and correction of errors allows the prevention of accidents and the significant reduction of occupational risk, and the implementation of a smart control system.

Patent document US20070049259-A1 discloses a portable communication device with an evacuation route display system to guide a user to a safe area in emergency situation. The system also shows emergency information. This invention transmits the alert to multiple devices; however, it is limited to the detection of smoke or certain gases, indicating an evacuation route.

Patent US7181 192-B2 presents an automated alert signal, which is composed of a portable communication device connected to the wireless telecommunication network wireless, a dynamic sensor which generates acceleration profiles of the device processes and a memory to store the generated profile. Additionally, this system allows verifying if the profile is already stored, to determine if it is necessary to notify the event to one or several emergency centres. This invention is limited to communication of the emergency to an alert attention centre.

Patent document US20070275690-A1 discloses a method for disseminating emergency notification from a source. The method sends the notification with the emergency content of the source to at least one transmitting party, selecting a subset of users for the dissemination of the emergency notification from the transmitting party, to the corresponding devices of the selected users.

Patent document US4257038-A presents an electronic security system for sending to various devices of coded intrusions and emergency signals for an alarm station. The alarm station decodes the intrusion signal, producing an alert from an alarm generator. The electronic alarm controls discriminate the intrusions from other alarm signals, blocking or delaying the intrusion signals, allowing the entry or exit from the protected area. The alarm control also presents alarm warning according to the alarm priority, the signals are received to allot the authorized input and provides disarming signals if they are not cancelled by a second coded transmission.

Patent US7714733-B2 consists of a risk alarm and includes a hazard alarm notification system and a signal detector that detects the alert messages from an external source. The hazard alarm system processes the alert message, emitting an alert through notifications. This risk alarm system determines if, how, where and/or when the alert is emitted from the alarm system using the notification appliance.

Patent document CN201020530731 discloses a security monitoring system in the home based on a network of wireless sensors that use the master/slave configuration to obtain the information, which is then shown in a computer.

Utility model CN202661822U uses a master/slave configuration in the control and monitoring of aquafarming systems. To do this, it uses embedded systems based on ARM microprocessors. The system includes the implementation of multiple sensors, which send their information to a slave node and this sends it to a master module.

DESCRIPTION OF THE INVENTION The present invention relates to a system and method for the detection of alerts and correction of errors using artificial intelligence paradigms.

The system for the detection of alerts and correction of errors in industrial processes comprises a sensorial unit with a plurality of sensors and a transmitter responsible for sending the data measured by the sensors to an alert activation unit. The alert activation unit is configured to compare the data measured by the sensors with reference values, determine in accordance with said comparison if there is an alert situation, and should an alert situation exist generate an alert signal (7) with the indication of the error produced. A control unit, comprising a control law database where control laws are stored, is configured to, in the event of reception of an alert signal, consult the control law database to verify if the error defined in the alert signal can be corrected using an already existing control law. If there is not an already existing control law to resolve the error, the control unit generates a new control law using artificial intelligence techniques, and it stores said control law generated in the control law database. Finally, the control unit applies the existing or generated control law.

The alert activation unit may comprise a database of reference values where the reference values of the sensors are stored. The alert activation unit may comprise a memory where the data measured by the sensors are stored. In a possible embodiment, the alert activation unit may comprise a comparator module responsible for comparing the data measured by the sensors with the reference values.

A second aspect of the present invention relates to a method for the detection of alerts and correction of errors in industrial processes. The method comprises:

Capturing data measured by a plurality of sensors installed in an industrial plant. Sending said data to an alert activation unit.

Comparing, in the alert activation unit, the data measured by the sensors with reference values.

Determining, in accordance with said comparison, if there is an alert situation.

Should an alert situation exist, generating an alert signal with the indication of the error produced and send said alert signal to a control unit.

Consulting, the control unit, a control law database to verify if the error defined in the alert signal can be corrected using an already existing control law.

If there is not an already existing control law to resolve the error, generating a new control law using artificial intelligence techniques and storing said control law generated in the control law database. Applying the existing or generated control law.

Therefore, the present invention is based on display systems integrating artificial intelligence for the formation of an autonomous system, developed in web environments, and adaptable to devices currently used, and thus be able to monitor and verify the alarms of a system being monitored and controlled. The system verifies the database of the readings in real time and compares it with the previous readings of the system which are suggested as ideal. Likewise, it seeks an ideal operation not contemplated in the reference database. Once the alert is generated, the control law is generated per the artificial intelligence paradigms. The present invention comprises a sensorial unit which captures the sensitive information transmitted by any of the current media (wireless, via internet, Ethernet, etc.) to an alert activation unit, where it is stored in a database. In this last unit it compares, according to different artificial intelligence paradigms, the data received from the sensorial unit with the desired values of the significant variables stored in another database. If the variable measured is greater than its desired value an alert signal is generated producing an error.

With the alert activated, it enters a control unit, where a control law is applied based on the artificial intelligence paradigms to correct the error. Should there exist the control law adapted to the alert generated, according to the type of plant or process to control the necessary time must be waited so that the adjustment made is effective. If there is any new learning in the application of the control law, it is modified with the new information. If there is no control law in existence to correct the error the system must generate it by an automatic learning technique storing it and informing the operator about the existence of the new control law. If no new law is found it informs the operator that the alert has not been corrected to analyse it manually.

The process allows the connection to an interface to display the data of the measured variables stored in a database (the numerical data or histograms can be presented). On the other hand, it is also possible to observe the historic evolution of each one of the variables. This process generates an autonomous model with automated and optimized decision-making. The system that implements the process integrates components such as sensors and actuators that fulfil the decision-making and generates an efficient control based on each one of its components. It is based on each one of the requirements and artificial intelligence, which define the tools and mechanisms it is necessary to work with. As it is also an adaptable, adjustable, flexible and dynamic model, it allows the application of other tools, techniques, methods and applications of knowledge engineering in artificial intelligence.

DESCRIPTION OF FIGURES A series of drawings is described very briefly below, which help to better understand the invention and are expressly related to an embodiment of said invention which is presented as a non-limiting example thereof.

Figure 1 shows a general view of the schematic diagram of the entire monitoring and control system used for the operation of the system. Figure 2 represents the flow diagram of the process involved in the alert activation unit.

Figure 3 illustrates the flow diagram of the process involved in the control unit.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to a system for the detection of alerts and correction of errors. Figure 1 shows the components of the detection and control system of the industrial plant to control. A sensorial unit 1 comprises a plurality of sensors (1a, 1 b, ... 1 n) of various types according to the application, and a transmitter 2 responsible for sending, to an alert activation unit 3, a data signal with the data from the DSensor sensors obtained from the reading of the sensors (1 a, 1 b, ... 1 n). The transmitter 2 can use different electronic and communication technologies (wireless, via internet, Ethernet, etc.) to form an embedded system.

The alert activation unit 3 stores the data measured by the sensors of the sensorial unit 1 in a memory 4 (e.g. in a database). A database of reference values 5 stores the desired values of the significant variables corresponding to the sensors. A comparator module 6 performs the comparison between the values of the reading of the sensors and the desired values stored, respectively, in the memory 4 and in the database of reference values 5.

Should there be any error (e.g. if the variable measured by the sensors is greater than its desired value) the alert activation unit 3 generates and sends an alert signal 7 to a control unit 8. The control unit 8 receives the alert signal 7, when it generates and seeks a control law contained in a control law database 9. Should said control law not exist in the control law database 9, the control unit 8 is responsible for automatically generating it. Later, the control unit 8 applies the control law found or generated.

Figure 2 shows the flow diagram that describes the process involved in the alert activation unit 3 and the form in which an alert signal 7 is generated. The sensor data (Dsensor) stored in the memory 4 are compared in the comparator module 6 with the reference or desired data (Datum) contained in the database of reference values 5. If there is any error in the comparison 6a of both data (in particular if Dsensor>Datum), the Dalert alert counter in increased by one in the sentence 6b. Otherwise, it is considered that there is no error (ideal function 6c). Finally, the comparison 6d generates an alert signal 7 which is sent to the control unit 8 if the alert counter (Dalert) is greater than a reference or limit value.

Figure 3 represents the flow diagram of the control unit 8. The control unit verifies 8a if the error can be corrected according to an already existing control law, in which case the control unit 8 applies the control law 8b to correct the error. If there is no control law to resolve the type of error or alert detected, the control unit 8 starts a generation process of a new law 8c automatically. It is verified 8d if the new control law could be generated, in which case it stores the new control law recently generated in the control law database 9 and it applies 8b the new control law to correct the error. Otherwise, it informs 8e of the existing situation to the operator to correct the error manually. The system of the present invention is applicable to any industrial process. By means of the use of artificial intelligence paradigms, the control unit 8 generates new control laws that correct the errors detected. Furthermore, the error alert and correction system has the capacity to automatically learn, on storing the control laws generated for future uses when errors or similar alert situations occur.

Claims

1. System for the detection of alerts and correction of errors in industrial processes, comprising: a sensorial unit (1) comprising a plurality of sensors (1 a, 1 b, ... , 1 n) and a transmitter (2) responsible for sending the data measured by the sensors (DSensor) sending to an alert activation unit (3); an alert activation unit (3) configured to: compare the data measured by the sensors (DSensor) with reference values (Datum); determining, in accordance with said comparison, if there is an alert situation; should an alert situation exist, generating an alert signal (7) with the indication of the error produced; a control unit (8) comprising a control law database (9) where control laws are stored, the control unit (8) being configured to, in the event of reception of an alert signal (7): consult the control law database (9) to verify (8a) if the error defined in the alert signal (7) can be corrected using an already existing control law; if there is not a an already existing control law to resolve the error, generating (8c) a new control law using artificial intelligence techniques, and storing said control law generated in the control law database (9); applying (8b) the existing or generated control law.

2. System according to claim 1 , where the alert activation unit (3) comprises a database of reference values (5) where the reference values (Datum) of the sensors are stored.

3. System according to any of the preceding claims, where the alert activation unit (3) comprises a memory (4) where the data measured by the sensors (DSensor) are stored.

4. System according to any of the preceding claims, where the alert activation unit (3) comprises a comparator module (6) responsible for comparing the data measured by the sensors (DSensor) with the reference values (Datum).

5. Method for the detection of alerts and correction of errors in industrial processes, comprising: capturing data measured (DSensor) by a plurality of sensors (1 a, 1 b, ... , 1 n) installed in an industrial plant; sending said data to an alert activation unit (3); comparing, in the alert activation unit (3), the data measured by the sensors (DSensor) with reference values (Datum); determining, in accordance with said comparison, if there is an alert situation; should an alert situation exist, generating an alert signal (7) with the indication of the error produced and sending said alert signal (7) to a control unit (8); consulting, the control unit (8), a control law database (9) to verify (8a) if the error defined in the alert signal (7) can be corrected using an already existing control law; if there is not an already existing control law to resolve the error, generating (8c) a new control law using artificial intelligence techniques and storing said control law generated in the control law database (9); applying (8b) the existing or generated control law.

6. Method according to claim 5, where the reference values (Datum) of the sensors are stored in a database of reference values (5).

7. System according to any of claims 5 to 6, where the data measured by the sensors (DSensor) are stored in a memory (4).