Search

DE-102024210917-A1 - Method for verifying the plausibility of a signal from a sensor in a technical system, as well as a control device for a technical system

DE102024210917A1DE 102024210917 A1DE102024210917 A1DE 102024210917A1DE-102024210917-A1

Abstract

Method for verifying the plausibility of a signal from a sensor (38) in a technical system, in particular a fuel system (12) of an internal combustion engine (10), wherein the signal characterizes a first quantity detected by the sensor (38) which characterizes a state of a first element or area (34) of the technical system (10, 12), comprising the following steps: a. Detecting or determining a value of the first quantity by means of the sensor (38); b. Detecting or determining a value of at least one second quantity which characterizes a state of a second element or area (28, 32, 36, 40, 50, 52) of the technical system (10, 12), wherein the second element or area (28, 32, 36, 40, 50, 52) is not the first element or area (34); c. d. Processing the value of at least one second quantity using a model to generate a value of the first quantity; d. Comparing the value of the first quantity generated in step c and the value of the first quantity recorded in step a; e. Initiating an action depending on the result of the comparison.

Inventors

  • Lyu Zhang
  • Gijo Pulikkottil Peter

Assignees

  • Robert Bosch Gesellschaft mit beschränkter Haftung

Dates

Publication Date
20260513
Application Date
20241113

Claims (10)

  1. Method for verifying the plausibility of a signal from a sensor (38) in a technical system, in particular a fuel system (12) of an internal combustion engine (10), wherein the signal characterizes a first quantity detected by the sensor (38) that characterizes a state of a first element or area (34) of the technical system (10, 12), comprising the following steps: a. Detecting (54) or determining a value (55) of the first quantity by means of the sensor (38); b. Detecting or determining (56) a value of at least a second quantity that characterizes a state of a second element or area (28, 32, 36, 40, 50, 52) of the technical system (10, 12), wherein the second element or area (28, 32, 36, 40, 50, 52) is not the first element or area (34); c. Processing the value of at least one second quantity using a model (58) to generate a value (59) of the first quantity; d. Comparing (62) the value (59) of the first quantity generated in step c and the value (55) of the first quantity recorded in step a; e. Initiating (64) an action depending on the result of the comparison.
  2. Procedure according to Claim 1 , characterized in that the model (58) used in step c comprises an artificial intelligence, for example machine learning, in particular a neural network, and in particular a deep neural network.
  3. Method according to at least one of the preceding claims, characterized in that in a technical system in the form of a fuel system (12) of an internal combustion engine (10) the first quantity is a pressure in a fuel rail (34) for gaseous fuel.
  4. Procedure according to Claim 3 , characterized in that for further processing the larger value of the generated value of the first quantity and the recorded value of the first quantity is used (64.1).
  5. Procedures according to at least one of the Claims 3 - 4 , characterized in that the second quantity is at least one of or characterized from the following group: operating position of a safety valve (28) arranged upstream of the fuel rail (34); operating position of a low-pressure pressure regulating valve (32) arranged upstream of the fuel rail (34); activation time of a fuel injector (40); pressure in a region upstream of the fuel rail (34); load requirement of the internal combustion engine (10); speed of a crankshaft of the internal combustion engine (10); gas temperature in the fuel rail (34).
  6. Method according to at least one of the preceding claims, characterized in that the value (59) of the first quantity generated in step c is made plausible by a value (55) of the first quantity which is expected for a certain operating state of the technical system (10, 12).
  7. Procedure according to Claim 6 , characterized in that the specific operating state of a technical system in the form of a fuel system (12) of an internal combustion engine (10) is at least one of the following group: idling of the internal combustion engine (10), starting of the internal combustion engine (10), switching off of the internal combustion engine (10).
  8. Method according to at least one of the preceding claims, characterized in that the value of at least a second quantity is checked to see if it is within a permissible range.
  9. Method according to at least one of the preceding claims, characterized in that the action (64) in step c is at least one from the following group: output of an error message; termination of operation of the technical system; bringing the technical system into a safe emergency operation.
  10. Control device (48), in particular for a fuel system (12) of an internal combustion engine (10), comprising a processor and a memory on which a computer program product is stored, characterized in that the computer program product includes instructions which, when the computer program product is executed by the control device (48), cause it to execute the method according to at least one of the preceding claims.

Description

State of the art The invention relates to a method for verifying the plausibility of a signal from a sensor in a technical system, in particular a fuel system of an internal combustion engine, and a control device for such a technical system. The DE 10 2021 210 001 A1 This describes a method for operating an internal combustion engine that runs on gaseous fuel, such as hydrogen. Air is supplied to multiple combustion chambers. Gaseous hydrogen is injected directly into the combustion chambers, or it is injected into the intake ports (port fuel injection), and the hydrogen-air mixture is ignited in the combustion chambers. The hydrogen can be stored in liquid form under relatively high pressure in a tank-like fuel storage unit of the internal combustion engine's fuel system. From there, it passes in gaseous form via a high-pressure pressure regulator with a pressure-reducing valve to a low-pressure pressure regulator and then to a fuel rail, which is functionally similar to the fuel rail in an internal combustion engine with gasoline direct injection or port fuel injection. Several fuel injectors are connected to the fuel rail, which deliver the gaseous fuel to the combustion chambers. Disclosure of the invention The problem underlying the invention is solved by a method and a control device with the features of the dependent claims. Advantageous embodiments are specified in the subclaims. The invention enables the plausibility check of a signal from a sensor that detects a first quantity, without requiring a further sensor for the same first quantity or a specially designed sensor. Instead, the plausibility check is achieved based on other second quantities, from which the first quantity is inferred. This improves the safety of the technical system while simultaneously keeping costs comparatively low. Specifically, these advantages are achieved through a plausibility check of a sensor signal in a technical system. This plausibility check verifies the sensor signal for accuracy, consistency, and reliability to ensure its reliability and meaningfulness. The technical system could be, for example, the fuel system of an internal combustion engine, but virtually any other technical system in which sensors play a role is also suitable. The signal characterizes a first physical quantity detected by the sensor, which characterizes the state of a first element (e.g., the position of a valve) or the state of a first region (e.g., the pressure or temperature of a gas) within the technical system. The method according to the invention comprises the following steps: a. Detecting or determining a value of the first quantity using the sensor. This corresponds to the typical operation of a sensor. The first quantity can be directly derived from the sensor signal or be the sensor signal itself, or it can be determined from the sensor signal, e.g., by conversion. b. Determining or capturing a value of at least one second quantity that characterizes the state of a second element or area of the technical system, where the second element or area is not the first element or area. This second quantity may, but need not, be based on a sensor signal. Physically, the second quantity may be the same as the first quantity, for example, pressure, voltage, etc. Functionally, however, it characterizes the state of a different element or area of the technical system than the first quantity. c. Processing the value of at least one second quantity using a model to generate a value for the first quantity. This step thus infers the value of the first quantity from the value of the second quantity(ies). With such a model, the behavior of the technical system and the relationship between the second quantity(ies) and the first quantity can, for example, be mathematically simulated, predicted, or estimated. d. Comparing the value of the first quantity generated in step c with the value of the first quantity recorded in step a. This is the actual plausibility check. If the generated value deviates from the recorded value by more than a predefined limit, and this deviation persists for at least a predefined period, there is in most cases at least an increased probability that the recorded value is implausible or incorrect, and thus a sensor error is present. e. Initiating an action depending on the outcome of the comparison. This step is the This is a logical consequence of the previous step. If there is an increased probability that the value detected by the sensor is implausible or incorrect, measures can be taken in this present step, for example, to establish a safe state of the technical system despite the error. In a further training program, the model used in step c is intended to include artificial intelligence, for example, machine learning, in particular a neural network, and especially a deep neural network. With artificial intelligence models, the value of the first quantity can be generated from the second quantity or quantities with h