Method for operating a sensor element for detecting at least one property of a measuring gas in a measuring gas chamber

Abstract

A method is proposed for operating a sensor element (10) for detecting at least one property of a measurement gas in a measurement gas chamber, in particular for detecting a proportion of a gas component in the measurement gas or a temperature of the measurement gas. The sensor element (10) has at least one solid electrolyte body (12), a pump cell (40), and a Nernst cell (46). The method comprises the steps of defining an upper limit characteristic curve (58) and a lower limit characteristic curve (60) for a pump current (I P ) and a pump voltage (U P ) of the pump cell (40), measuring a pump current (I P ), measuring a pump voltage (U P ), forming pairs of values for the measured pump current (I P ) and the measured pump voltage (U P), comparing the value pairs with the upper limit characteristic curve (58) and the lower limit characteristic curve (60), and detecting a suspected offset if a predetermined number of value pairs lie outside a tolerance band (66) between the upper limit characteristic curve (58) and the lower limit characteristic curve (60).

Description

State of the art

[0001] A large number of sensor elements and methods for detecting at least one property of a measurement gas in a measurement gas chamber are known from the prior art. In principle, these can be any physical and / or chemical properties of the measurement gas, with one or more properties being able to be detected. The invention is described below in particular with reference to a qualitative and / or quantitative detection of a proportion of a gas component of the measurement gas, in particular with reference to a detection of an oxygen content in the measurement gas component. The oxygen content can be detected, for example, in the form of a partial pressure and / or in the form of a percentage. Alternatively or additionally, however, other properties of the measurement gas can also be detected, such as the temperature.

[0002] Ceramic sensor elements based on the electrolytic properties of certain solids, i.e., the ion-conducting properties of these solids, are known from the prior art. These solids can be ceramic solid electrolytes, such as zirconium dioxide (ZrO2), especially yttrium-stabilized zirconium dioxide (YSZ) and scandium-doped zirconium dioxide (ScSZ), which may contain small amounts of aluminum oxide (Al2O3) and / or silicon oxide (SiO2).

[0003] For example, such sensor elements can be designed as so-called lambda sensors, as described, for example, in Konrad Reif (ed.): Sensors in Motor Vehicles, 1st edition 2010, pp. 160-165. Broadband lambda sensors, especially planar broadband lambda sensors, can be used to determine the oxygen concentration in the exhaust gas over a wide range, thus providing information about the air-fuel ratio in the combustion chamber. The air-fuel ratio λ describes this air-fuel ratio.

[0004] Internal combustion engines have a catalytic converter in the exhaust system to comply with emissions regulations. To control the internal combustion engine, a broadband lambda sensor is typically positioned upstream of the catalytic converter's exhaust flow and a so-called step-through sensor downstream of the catalytic converter's exhaust flow. The pumping current is output from the engine control unit's component package and used for lambda control. The pumping current is converted into a lambda signal in the engine control unit, which is used for control. The current calibration of the broadband lambda sensors upstream of the catalytic converter was previously performed via the step-through sensor downstream of the catalytic converter. When the step-through sensor downstream of the catalytic converter, which is very accurate at λ = 1, is active and λ = 1 is set, the measured value of the broadband lambda sensor upstream of the catalytic converter can be corrected.The broadband lambda sensor upstream of the catalyst is less accurate at λ = 1 than the step sensor downstream of the catalyst, but can also measure lambda values other than 1.

[0005] There are also methods that identify an offset of the broadband lambda sensor upstream of the catalytic converter without the need for a second lambda sensor downstream of the catalytic converter as a reference, as described, for example, in EP 0 444 674 B1. This can be referred to as self-adjustment of the broadband lambda sensor because only signals from the broadband lambda sensor itself are used. In these methods, in addition to the pump current signal, from which the lambda value of the sample gas is calculated, the pump voltage is also evaluated. The pump voltage signal serves as a reference and it is assumed that the pump voltage (particularly at lambda = 1.0) is not subject to aging or interference effects. By adjusting the pump current and pump voltage at lambda = 1.0 (which theoretically corresponds to a pump voltage of approximately 0 mV), a pump current offset and thus also a lambda offset is determined.

[0006] Other methods, such as those described in DE 10 2013 202 161 A1 and DE 10 2013 202 260 A1, are also based on the pump current and the pump voltage, but calculate a lambda value from each of these variables. The difference between these two lambda values is interpreted as the lambda offset of the broadband lambda sensor, with the lambda value calculated from the pump voltage serving as a reference. Instead of calculating a lambda offset, however, it is also possible to calculate the self-adjustment of the broadband lambda sensor based on the pump current in the form of a pump current offset, as described, for example, in DE 10 2018 205 494 A1, which directly identifies the actual physical fault pattern of a leakage current or an injected current.

[0007] Despite the advantages of the sensor elements and methods for operating them known from the state of the art, they still have potential for improvement. For example, the current use of the step-in sensor downstream of the catalytic converter requires a functional lambda sensor to calibrate the broadband lambda sensors upstream of the catalytic converter. A challenge in self-calibration of the broadband lambda sensor is tolerances, especially in the pump voltage, which are reflected in the underlying parameters, such as the pump current / pump voltage or lambda / pump voltage curves. These tolerances arise, among other things, from production variations or aging / poisoning of the sensor.Since the tolerances of the broadband lambda sensor's self-adjustment system increase significantly with increasing distance from lambda = 1.0, the broadband lambda sensor's self-adjustment can generally only be operated robustly within a narrow activation range around exhaust lambda = 1.0. However, this activation range can be permanently exceeded due to control and adaptation interventions of the lambda control system, particularly in the case of an offset error of the broadband lambda sensor, so that the self-adjustment of the broadband lambda sensor cannot run. Disclosure of the invention

[0008] A method for operating a sensor element for detecting at least one property of a measuring gas in a measuring gas chamber is therefore proposed, which method at least largely avoids the disadvantages of known methods and in which a suspected error can be indicated even if the self-adjustment of the broadband lambda sensor is not able to run due to a potential offset error.

[0009] The method according to the invention is suitable for operating a sensor element for detecting at least one property of a measurement gas in a measurement gas chamber, in particular for detecting a proportion of a gas component in the measurement gas or a temperature of the measurement gas. The sensor element comprises at least one solid electrolyte body, a pump cell, and a Nernst cell.

[0010] The method comprises the following steps, preferably in the order given, whereby the respective steps can be carried out multiple times or repeated or carried out in parallel: - Defining an upper limit characteristic and a lower limit characteristic for a pump current and a pump voltage of the pump cell, - Measuring a pump current, - Measuring a pump voltage, - Forming pairs of values for the measured pump current and the measured pump voltage - comparing the pairs of values with the upper limit characteristic curve and the lower limit characteristic curve, and - Detection of a suspected offset if a predetermined number of value pairs lie outside a tolerance band between the upper limit characteristic curve and the lower limit characteristic curve.

[0011] A basic idea of the invention is to generate a suspected offset error of the sensor element based on measured value pairs of pump current and pump voltage compared to upper and lower limit characteristic curves of pump current and pump voltage, which represent the nominal or normal behavior of the system. A suspected offset of the sensor element is detected if a predetermined and sufficiently large number of the measured pump current / pump voltage value pairs lie outside the applied limit characteristic curves. This allows a suspected offset error to be reliably detected. The advantage of this invention is that a suspected offset is formed using limit curves derived from the sensor-typical characteristic curve.

[0012] The predetermined number of value pairs can be greater than two, preferably greater than three, and even more preferably greater than five. A suspected offset of the sensor element is thus detected if a large proportion of the measured pump current / pump voltage value pairs lie outside the applied limit characteristics.

[0013] The method may further include debouncing the detected offset suspicion. Debouncing can generate a robust offset suspicion.

[0014] Debounce can include classifying value pairs within the upper and lower limit characteristic curves as OK, classifying value pairs outside the upper and lower limit characteristic curves as NOT OK, and accumulating the value pairs classified as OK and the value pairs classified as NOT OK using a debounce counter. If a predetermined threshold is exceeded for the value pairs classified as NOT OK, an offset suspicion can be declared.

[0015] In other words, debouncing the suspected offset can be achieved by accumulating a dimensionless debounce counter. Pump voltage / pump current value pairs classified as OK lead to the counter accumulating with a pump voltage-dependent increment, while value pairs classified as not OK lead to the counter accumulating with a pump voltage-dependent increment of the opposite sign. If the debounce counter exceeds a threshold, the suspected offset is declared.

[0016] The predetermined threshold may further include a hysteresis to cancel the suspected offset. Hysteresis can thus be applied to cancel the suspected offset to prevent bouncing at the threshold.

[0017] Higher-magnitude pump voltage values of the value pairs can be weighted with smaller-magnitude increments of the debounce counter. At the outer pump voltage ranges, offset pump voltage / pump current value pairs may fall within the tolerance band due to the characteristically flatter shape of the sensor element's reference characteristic curve and thus cannot be reliably detected. These measured values are therefore weighted lower or not weighted in the accumulation. This means that at higher-magnitude pump voltages, the increments for each pump voltage / pump current value pair are considered with smaller-magnitude increments of the counter for measured values classified as OK and not OK.

[0018] As an alternative to debouncing by means of a debouncing counter, the debouncing may comprise determining a weighting factor for each pair of values classified as in order and each pair of values classified as out of order as a function of its pump voltage value, weighting each of the pairs of values classified as in order and each of the pairs of values classified as out of order with the determined weighting factor, accumulating the weighted pairs of values classified as in order and the weighted pairs of values classified as out of order, and evaluating the total weights of the weighted pairs of values classified as in order and the weighted pairs of values classified as out of order by forming a ratio feature between the total weights.As an alternative debouncing method for suspected offsets, an OK or non-OK individual weight is determined for each pump voltage / pump current value pair, depending on the classification, based on its pump voltage. The individual weights have values between 0 (no weight) for the flatter pump voltage ranges in the reference characteristic curve, and an end value (maximum weight) for the pump voltage ranges in the reference characteristic curve close to 0 V. All OK and non-OK individual weights are accumulated over a fixed number of measured value pairs to form OK and non-OK total weights. The evaluation is based on ratio characteristics between OK and non-OK total weights.

[0019] If a predetermined threshold for the ratio characteristic is exceeded, suspicion of an offset can be declared. Thus, suspicion of an offset is only confirmed if there is a high probability that an offset is present.

[0020] Suspected offsets can only be declared if the total weights each exceed a predetermined minimum sum. The suspected offset should only be updated when a minimum value for the sum of the acceptable and non-acceptable total weights is reached. This avoids updating the suspected offset during operation in the less meaningful, flat areas of the reference characteristic curve, since this sum has numerically smaller values here.

[0021] If a suspected offset is detected, the pump cell can be regulated to a predetermined setpoint for the pump cell's pump voltage. When the predetermined setpoint for the pump cell's pump voltage is reached, the sensor element can be self-adjusted. In conjunction with a preferably pump voltage-based regulation to a setpoint for the pump cell's pump voltage, this results in a completely independent closed-loop sensor element adaptation.

[0022] The sensor element can be used in the exhaust system of an internal combustion engine, where the predetermined target value for the pump voltage is adjusted to lambda = 1 for the exhaust gas of the internal combustion engine. This suspected fault can trigger a suitable lambda intervention (e.g., by switching to pump voltage-based lambda control), which allows the activation range of the self-tuning to be reached again. In conjunction with a pump voltage-based lambda control, a completely independent closed-loop sensor element adaptation is achieved.

[0023] Adjustment to the predetermined measured value can be achieved by controlling the pump voltage. Control based on the pump voltage is particularly robust and less prone to error if a fault is suspected due to a pump current offset. This means that if an offset suddenly occurs, the pump voltage-based lambda control automatically adjusts to the pump voltage switch-on range required for self-adjustment after a suspected fault has been identified. This enables robust detection of the offset. As a result, the value pairs of pump current / pump voltage are measured within the limit characteristic curve again after adaptation, and the suspected fault is automatically eliminated after debouncing. Meanwhile, for the duration of the pump voltage-based lambda control, an emissions advantage results compared to an operating point outside of real stoichiometric operation with lambda control based on an offset pump current.

[0024] Furthermore, a computer program is proposed which is configured to carry out each step of the method according to the invention.

[0025] Furthermore, an electronic storage medium is proposed on which such a computer program is stored.

[0026] Furthermore, an electronic control unit is proposed which comprises such an electronic storage medium.

[0027] In the context of the present invention, a solid electrolyte body is understood to mean a body or object with electrolytic properties, i.e., with ion-conducting properties, for example, oxygen ion-conducting properties. In particular, it can be a ceramic solid electrolyte. For example, the solid electrolyte body can comprise stabilized zirconium dioxide and / or scandium-stabilized zirconium dioxide. The solid electrolyte body can also be formed from multiple solid electrolyte layers. A layer is understood to be a uniform mass with a flat extension at a certain height, located above, below, or between other elements.

[0028] In the context of the present invention, an electrode is generally understood to be an element capable of contacting the solid electrolyte in such a way that a current can be maintained or a voltage can be measured through the solid electrolyte and the electrode. Accordingly, the electrode can comprise an element at which the ions can be incorporated into the solid electrolyte and / or removed from the solid electrolyte. Typically, the electrodes comprise a noble metal electrode, which can be applied to the solid electrolyte, for example, as a metal-ceramic electrode or can be connected to the solid electrolyte in some other way. Typical electrode materials are platinum cermet electrodes. However, other noble metals, such as gold or palladium, can also be used in principle.

[0029] In the context of the present invention, a Nernst cell is an electrochemical measuring cell that uses a solid electrolyte as a membrane between two electrodes. This cell utilizes the property of the solid electrolyte to electrolytically transport ions of the gas to be measured, such as oxygen ions, from one electrode to the other above a certain temperature, creating a so-called Nernst potential. This property allows the difference in the partial pressure of the gas to be measured on the different sides of the membrane to be determined. In the lambda sensor, one side of the membrane is exposed to the gas to be measured, while the other side is connected to a reference.

[0030] In the context of the present invention, a pump cell is understood to be an electrochemical cell in which the content of a component of the measurement gas, such as oxygen, in a measuring gap is determined on the one hand by the measurement gas acting through a diffusion channel and on the other hand is influenced by the current flow of the pump cell. Depending on the polarity, the pump current pumps measurement gas from the measurement gas side of the solid electrolyte membrane into the measuring gap or out of it. The pump current is regulated by an external controller such that the lambda value in the measurement gas precisely balances the measurement gas flow through the diffusion channel and keeps the measurement gas in the measuring gap constant at a predetermined value such as λ = 1. A lambda value of 1 is always given, for example, when the voltage across the Nernst cell is 0.45 V. The pump current pumps measurement gas ions into the measurement gas in the measuring gap when the mixture is rich and out when the mixture is lean.

[0031] In the context of the present invention, a heating element is understood to be an element that serves to heat the solid electrolyte and the electrodes to at least their functional temperature and preferably to their operating temperature. The functional temperature is the temperature above which the solid electrolyte becomes conductive to ions and is approximately 350°C. This is to be distinguished from the operating temperature, which is the temperature at which the sensor element is typically operated and which is higher than the functional temperature. The operating temperature can be, for example, 700°C to 950°C.

[0032] The heating element can comprise a heating region and at least one supply line. In the context of the present invention, a heating region is understood to be the region of the heating element which, in the layer structure, overlaps with an electrode along a direction perpendicular to the surface of the sensor element. The heating region usually heats up more during operation than the supply line because it has a higher electrical resistance, so that they can be distinguished. The different heating can therefore be achieved, for example, by the heating region having a higher electrical resistance than the supply line. The heating region and / or the supply line are designed, for example, as electrical resistance tracks and heat up when an electrical voltage is applied. The heating element can be made, for example, from a platinum cermet.

[0033] For the purposes of this invention, debouncing is a technique used in electronics to reduce the number of unwanted signals or triggers generated by mechanical switches or noisy signals. It is used to prevent multiple signals from being generated when a switch is operated multiple times in rapid succession. It is a type of signal conditioning technique used in a wide variety of applications, from controlling the speed of motors to controlling the input of digital signals to microcontrollers. Debounce works by adding a delay to the signal so that multiple transitions within a short period of time are ignored.This delay can be added in a variety of ways, from using a resistor-capacitor combination to using software algorithms to capture and filter the signal. Debounce is a useful technique for eliminating unwanted signals and preventing system malfunctions. This is especially important in digital systems, as multiple signals can cause system malfunctions or unpredictable behavior.

[0034] In the context of the present invention, a debounce counter is understood to be a component designed to temporarily store data and interacts with measuring devices and control units. If a control unit detects an error or incorrect data, these are registered and first stored in the debounce counter. The control units can use multiple microprocessors, which can make several thousand queries per second. An error is therefore first temporarily stored in the debounce counter to check whether it occurs again in subsequent queries. If this was only a one-time error, the debounce counter is cleared. If the error occurs again, an error code is generated and stored in the actual error memory.Furthermore, the debounce time can be measured with a debounce counter, which is started when a signal first occurs and incremented at regular intervals as long as the signal is present. Only when a predefined threshold is exceeded is the signal not interpreted as interference, but is further processed by the system. Debounce times are used to prevent short interferences from being mistakenly recognized as signals.

[0035] In the context of the present invention, hysteresis is understood to be an effect that occurs with a delay compared to a change in its cause (e.g., in the case of thermostat-controlled heating, the difference between the on and off temperature). Hysteresis characterizes a variant-delayed behavior of the effected output variable (in the case of heating, the actual temperature) - related to the input variable (in the case of heating) - which has reached its maximum or minimum. Generally speaking, hysteresis is a system behavior in which the output variable depends not only on the independently changing input variable, but also on the previous state of the output variable. The system can therefore assume one of several possible states with the same input variable, depending on its previous history. This behavior is also called path dependency.Hysteresis occurs in many natural and technical processes, particularly in the magnetization of a magnet, in control engineering, and cybernetics. Hysteresis behavior is typically characterized by the occurrence of a hysteresis loop, which is created when the causative variable is moved back and forth between two different values. The most well-known phenomenon is the hysteresis behavior of a ferromagnet in a magnetic field: If a non-magnetized ferromagnet is exposed to an external field and then removed, the ferromagnet retains a positive or negative magnetization depending on the polarity (i.e., direction) of the external field. This residual magnetization is called remanence. Short description of the drawings

[0036] Further optional details and features of the invention emerge from the following description of preferred embodiments, which are shown schematically in the figures.

[0037] They show: Fig. 1 a cross-section of a sensor element, Fig. 2 a schematic representation of a definition of pump voltage-pump current reference characteristic and limit characteristics, Fig. 3 a schematic representation of a principle of detecting offset measured values via the pump current, Fig. 4 a schematic representation of a principle of detecting offset measured values via the pump voltage, Fig. 5 a schematic representation of a principle of debouncing the offset suspicion via a debouncing counter, and Fig. 6 a schematic representation of a principle of debouncing the offset suspicion via a fixed number of value pairs with pump voltage weighted in order / not in order ratio features. Embodiments of the invention

[0038] Fig.1 shows a cross-sectional view of a sensor element 10 according to a first embodiment of the invention. Fig.The sensor element 10 shown in Figure 1 can be used to detect physical and / or chemical properties of a measurement gas, wherein one or more properties can be detected. The invention is described below in particular with reference to a qualitative and / or quantitative detection of a gas component of the measurement gas, in particular with reference to a detection of an oxygen content in the measurement gas. The oxygen content can be detected, for example, in the form of a partial pressure and / or in the form of a percentage. In principle, however, other types of gas components can also be detected, such as nitrogen oxides, hydrocarbons and / or hydrogen. Alternatively and / or additionally, however, other properties of the measurement gas can also be detected.The invention can be used in particular in the field of automotive engineering, so that the measuring gas chamber can in particular be an exhaust tract of an internal combustion engine and the measuring gas can in particular be an exhaust gas.

[0039] The sensor element 10 comprises a solid electrolyte body 12. The sensor element 10 further has a gas access path 14. The gas access path 14 has a gas access hole 16 that extends from an outer side or surface 18 of the solid electrolyte body 12 into the interior of the solid electrolyte body 12. An electrode cavity 20 is provided in the solid electrolyte body 12, which adjoins and is connected to the gas access hole 16. The electrode cavity 20 is, for example, cuboid-shaped. The electrode cavity 20 is part of the gas access path 14 and can be connected to the measurement gas chamber via the gas access hole 16. For example, the gas inlet hole 16 extends as a cylindrical blind hole perpendicular to the surface 18 of the solid electrolyte body into the interior of the solid electrolyte body 12. Between the gas inlet hole 16 and the electrode cavity 20, a channel 22 is arranged, which is also part of the gas inlet path 14.The channel 22 or the electrode cavity 20 is arranged radially or perpendicularly with respect to the gas inlet hole 16. A diffusion barrier 24 is arranged in this channel 22. The diffusion barrier 24 reduces or even prevents the flow of measurement gas from the measurement gas chamber into the electrode cavity 20 and only allows the measurement gas to diffuse. A reference gas channel 26 or exhaust air channel is formed in the solid electrolyte body 12 and separated from the electrode cavity 20.

[0040] Furthermore, the sensor element 10 has a heating element 28. The heating element 28 is arranged in the solid electrolyte body 12 below the electrode cavity 20 and the reference gas channel 26, in an imaginary extension of the direction in which the gas inlet hole 16 extends. The heating element 28 has a heating region 30, a first supply path 32, and a second supply path 34. The first supply path 32 is connected to a positive pole 36 of the heating region 30. The second supply path 34 is connected to a negative pole 38 of the heating region 30.

[0041] The sensor element 10 further comprises a pump cell 40. The pump cell 40 has a first electrode 42 and a second electrode 44 arranged in the electrode cavity 20. The first electrode 42 is arranged on the surface 18 of the solid electrolyte body 12 that is exposed to the measurement gas. The second electrode 44 is arranged in the electrode cavity 20 on a side facing the first electrode 42. The pump cell 40 further comprises the part of the solid electrolyte body 12 between the first electrode 42 and the second electrode 44. A limiting current in the pump cell 40 can be set via the diffusion barrier 24. The limiting current represents a current flow between the first electrode 42 and the second electrode 44 via the solid electrolyte body 12.

[0042] The sensor element 10 further comprises a Nernst cell 46. The Nernst cell 46 comprises a third electrode 48 and a fourth electrode 50. The third electrode 48 is located adjacent to the heating region 30 of the heating element 28 in the electrode cavity 20. The fourth electrode 50 is arranged in the reference gas channel 26. The fourth electrode 50 can be arranged as a so-called pumped reference in the reference gas channel 26. This means that the reference gas channel 26 is not a macroscopic reference gas channel, but rather a pumped reference, i.e., an artificial reference. The second electrode 44 and the third electrode 48 are coupled to one another via the solid electrolyte body 12. The third electrode 48, the fourth electrode 50 and the part of the solid electrolyte body 12 between the third electrode 48 and the fourth electrode 50 form, for example, the Nernst cell 46.By means of the pump cell 40, for example, a pump current through the pump cell 40 can be adjusted such that the condition λ = 1 or another known composition prevails in the electrode cavity 20. This composition is in turn detected by the Nernst cell 46 by measuring a Nernst voltage between the third electrode 48 and the fourth electrode 50. Since there is an excess of oxygen in the reference gas channel 26 or in the fourth electrode 50, which serves as the reference electrode, the measured voltage can be used to determine the composition in the electrode cavity 20. According to the described structure, the first electrode 42 can also be referred to as the outer pump electrode, the second electrode 44 as the inner pump electrode, the third electrode 48 as the Nernst electrode, and the fourth electrode 50 as the reference electrode.

[0043] As can be seen from the above description of the structure of the sensor element 10, this is designed as a so-called two-cell, in which the pump cell 40 and the Nernst cell 46 are formed separately.

[0044] The sensor element 10 can be part of a broadband lambda probe and is arranged, for example, in an exhaust tract of an internal combustion engine 52 (not shown in detail). The sensor element 10 is connected to an electronic control unit 54 of the internal combustion engine 52 or integrated into its control and regulation system.

[0045] A method according to the invention for operating the sensor element 10 is described below.

[0046] Fig. Figure 2 shows a schematic representation of a definition of the pump voltage-pump current reference characteristic and limit characteristics. Fig.2 shows a reference characteristic curve 56 typical for the sensor element 10. First, the method defines an upper limit characteristic curve 58 and a lower limit characteristic curve 60 for a pump current I P and a pump voltage U P the pump cell 40 is defined. Fig. 2 shows on the X-axis 62 the pump current I P and on a Y-axis 64 the pump voltage U P The limit characteristic curves 58, 60 can be determined based on tolerance analyses, such as the 4-sigma approach. The upper limit characteristic curve 58 and the lower limit characteristic curve 60 span a tolerance band 66 around the reference characteristic curve 56.

[0047] Furthermore, the method comprises measuring a pump current I P and measuring a pump voltage U P during operation of the sensor element 10. For the measured pump current I P and the measured pump voltage U PPairs of values are then formed. The resulting pairs of values can then be displayed in the Fig. 2. The value pairs are then compared with the upper limit characteristic curve 56 and the lower limit characteristic curve 58. Non-offset value pairs 68 lie within the tolerance band 66, while offset value pairs 70 lie outside the tolerance band 66.

[0048] Fig. 3 a schematic representation of a principle of detecting offset measured values via the pump current I P . For a given pump voltage U P (k) in step S30, a maximum pump current I P,max (U P ) depending on the pump voltage U P and in parallel in step S32 a minimum pump current I P,min (U P ) depending on the pump voltage U P read and in step S34 with a measured pump current I P(k). The comparison in step S34 includes checking whether the measured pump current I P greater than the determined minimum pump current I P,min and smaller than the determined maximum pump current I P,max Then, in step S36, the information is output as to whether the value pair lies within or outside the tolerance band 66. The index "k" indicates a number or figure for the value pair, ie, the k-th value pair, such as the fifth value pair.

[0049] Fig. 4 shows a schematic representation of a principle of detecting offset measured values via the pump voltage U P . For a given pump current I P (k) in step S40 a maximum pump voltage U P,max (I P ) depending on the pump current I P and in parallel in step S42 a minimum pump voltage U P,min (I P ) depending on the pump voltage U Pread and in step S44 with a measured pump voltage U P (k). The comparison in step S44 includes checking whether the measured pump voltage U P greater than the determined minimum pump voltage U P,min and less than the determined maximum pump voltage U P,max Then, in step S46, the information is output as to whether the value pair lies within or outside the tolerance band 66.

[0050] Both in the Fig. 3 and Fig. The four variants shown are equivalent. If the measured value pair lies within the tolerance band of 66, it is classified as OK. If the measured value pair lies outside the tolerance band of 66, it is classified as not OK.

[0051] In order to prevent sensor elements 10 with marginal tolerances in the field, in conjunction with physical phenomena such as measurement noise or dynamic hysteresis, from leading to individual measured values outside the tolerance band and triggering a suspected offset, debouncing must be carried out over several measured values.

[0052] A suspected offset is detected if a predetermined number of value pairs lies outside a tolerance band 66 between the upper limit characteristic curve 56 and the lower limit characteristic curve 58. The predetermined number of value pairs is greater than two, preferably greater than three, and even more preferably greater than five, such as 10, 15, or even more. A suspected offset is detected in particular if a large portion of the measured pump current / pump voltage value pairs lie outside the applied limit characteristic curves 58, 60.

[0053] Fig.Figure 5 shows a schematic representation of the principle of debouncing the suspected offset using a debouncing counter 70. The debouncing counter 70 is, for example, an up-down debouncing counter. Debouncement involves classifying value pairs within the upper and lower limit characteristic curves as OK, classifying value pairs outside the upper and lower limit characteristic curves as NOT OK, and accumulating the value pairs classified as OK and the value pairs classified as NOT OK using a debouncing counter 70. If a predetermined threshold value is exceeded for the value pairs classified as NOT OK, the suspected offset is declared.

[0054] More precisely, the debouncing of the suspected offset can be achieved by accumulating a dimensionless debounce counter 70. Pump voltage / pump current value pairs that are classified as OK, after incrementation in step S50 and passing through an OK / NOK increment selection logic in step S52, lead to the accumulation of the counter in step S54 with a pump voltage-dependent increment, while value pairs that are classified as NOK, after incrementation in step S56 and passing through an OK / NOK increment selection logic in step S52, lead to the accumulation of the counter in step S54 with a pump voltage-dependent increment of the respective other sign. For this purpose, the OK / NOK increment selection logic is supplied with the information in step S58 as to whether the value pair lies within or outside the tolerance band 66.Then, in step S60, a comparison is made with a predetermined threshold for the suspicion. If the debounce counter 70 exceeds a threshold, a suspicion of offset is declared in step S62. To prevent bouncing at the threshold, a hysteresis is also applied to clear the suspicion of offset.

[0055] At the outer pump voltage ranges, the characteristically flatter shape of the reference characteristic curve 56 may cause offset pump voltage / pump current value pairs to fall within tolerance band 66 and thus not be reliably detected. These measured values are therefore weighted lower or not included in the accumulation. This means that for higher-magnitude pump voltages, the increments for each pump voltage / pump current value pair are considered with smaller-magnitude increments of the counter for measured values classified as OK and not OK.

[0056] Fig.Figure 6 shows a schematic representation of a principle for debouncing the suspected offset using a fixed number of value pairs with pump voltage-weighted OK / NOT OK ratio features. Thus, debouncing can be used as an alternative to the variant in Fig. 5 Determining a weighting factor for each pair of values classified as in order and each pair of values classified as out of order as a function of its pump voltage value, weighting each of the pairs of values classified as in order and each of the pairs of values classified as out of order with the determined weighting factor, accumulating the weighted pairs of values classified as in order and the weighted pairs of values classified as out of order, and evaluating the total weights of the weighted pairs of values classified as in order and the weighted pairs of values classified as out of order by forming a ratio feature between the total weights.

[0057] Depending on the classification in step S70, for each individual pump voltage / pump current value pair, an OK or an OK individual weight is determined in step S72 depending on its pump voltage. The individual weights have values between 0, i.e. no weight, which is used for the flatter pump voltage ranges in the reference characteristic curve 56, and an end value, i.e. maximum weight, which is used for the pump voltage ranges in the reference characteristic curve close to 0 V. In step S74, the OK individual weights are then weighted, and in parallel, in step S76, the OK individual weights are weighted. In step S78, all OK and OK individual weights are then accumulated over a fixed number of measured value pairs to form OK and OK total weights.

[0058] The assessment is based on the ratio of the total weight to the total weight that is acceptable or not acceptable. Possible ratios and the respective conditions for determining a suspected offset are shown in Table 1. Table 1 # | Ratio characteristic | Condition for setting a Suspicion of offset 1 ViO=iOiO+niO Ratio characteristic < suspicion threshold 2 VniO=niOiO+niO Ratio characteristic > Suspicion threshold 3 V'iO=iOniO Ratio characteristic < suspicion threshold 4 V'niO=niOiO Ratio characteristic > Suspicion threshold

[0059] In Table 1, V or V' represents the ratio characteristic. The abbreviation "OK" denotes a pair of values classified as "in order." The abbreviation "not in order" denotes a pair of values classified as "not in order."

[0060] As can be seen from Table 1, for each definition of the suspicious feature V or V', the offset suspicion is then expressed or confirmed in step S80 if it exceeds or falls below a predetermined threshold value for the suspicion.

[0061] The suspected offset is preferably updated only when a minimum value for the sum of the total weights that are OK and those that are not OK is reached. This avoids updating the suspected offset during operation in the less meaningful, flat areas of the reference characteristic curve 56, since this sum has numerically smaller values here.

[0062] In the case of detection and in particular expression or confirmation of a suspicion of offset, the pump cell 40 is then regulated to a predetermined setpoint value for the pump voltage U P of the pump cell 40, such as substantially 0 mV. When the predetermined setpoint for the pump voltage U P The pump cell 40 then carries out a self-adjustment of the sensor element 10.

[0063] In the present case, in which the sensor element 10 is used as part of a broadband lambda probe and the sensor element 10 is used in an exhaust tract of an internal combustion engine 52, the predetermined measured value is lambda = 1 for the exhaust gas of the internal combustion engine 52. The adjustment to the predetermined measured value is preferably carried out by controlling the pump voltage U P .

[0064] In conjunction with pump voltage-based lambda control, this results in completely independent closed-loop sensor element adaptation. This means that if a sudden offset occurs, the pump voltage-based lambda control automatically adjusts to the pump voltage switch-on range required for self-adjustment after a suspected fault has been generated. This enables robust offset detection. As a result, the pump current / pump voltage value pairs are measured within the limit characteristic curve after adaptation, and the suspected fault is automatically removed after debouncing. Meanwhile, for the duration of the pump voltage-based lambda control, an emissions advantage results compared to an operating point outside of real stoichiometric operation with lambda control based on an offset pump current.

[0065] The use of the invention can be verified if, for example, a suspected offset is detected when a leakage current is introduced or a coupling to the pump current occurs. In conjunction with the pump voltage-based lambda control, the activation range of the sensor element's self-adjustment is reached, the applied offset is adapted, and the suspected error is subsequently deleted. With the method, an error entry is made in the control unit's error memory if the offset is sufficiently large. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] EP 0 444 674 B1

[0005] DE 10 2013 202 161 A1

[0006] DE 10 2013 202 260 A1

[0006] DE 10 2018 205 494 A1

[0006] Cited non-patent literature

[0000] Konrad Reif (ed.): Sensors in motor vehicles, 1st edition 2010, pp. 160-165

[0003]

Claims

[1] Method for operating a sensor element (10) for detecting at least one property of a measurement gas in a measurement gas space, in particular for detecting a proportion of a gas component in the measurement gas or a temperature of the measurement gas, wherein the sensor element (10) has at least one solid electrolyte body (12), a pump cell (40), and a Nernst cell (46), the method comprising the following steps: - Defining an upper limit characteristic curve (58) and a lower limit characteristic curve (60) for a pump current (I P ) and a pump voltage (U P ) of the pump cell (40), - Measuring a pump current (I P ), - Measuring a pump voltage (U P ), - Formation of value pairs for the measured pump current (I P ) and the measured pump voltage (U P ), - comparing the pairs of values with the upper limit characteristic curve (58) and the lower limit characteristic curve (60), and - Detecting a suspected offset if a predetermined number of value pairs lie outside a tolerance band (66) between the upper limit characteristic curve (58) and the lower limit characteristic curve (60). [2] Method according to the preceding claim, wherein the predetermined number of value pairs is greater than two, preferably greater than three and more preferably greater than five. [3] Method according to one of the preceding claims, further comprising debouncing the detected offset suspicion. [4] Method according to the preceding claim, wherein the debouncing comprises classifying value pairs within the upper limit characteristic curve (58) and lower limit characteristic curve (60) as OK, classifying value pairs outside the upper limit characteristic curve (58) and lower limit characteristic curve (60) as NOT OK and accumulating the value pairs classified as OK and as NOT classified value pairs by means of a debouncing counter (70), wherein if a predetermined threshold value is exceeded for the value pairs classified as NOT OK, the offset suspicion is expressed. [5] Method according to the preceding claim, wherein the predetermined threshold value further comprises a hysteresis for removing the offset suspicion. [6] Method according to one of the two preceding claims, wherein higher pump voltage values of the value pairs are weighted with smaller increments of the debounce counter (70). [7] The method of claim 3, wherein debouncing comprises determining a weighting factor for each in-order value pair and each out-of-order value pair depending on its pump voltage value, weighting each of the in-order value pairs and each of the out-of-order value pairs with the determined weighting factor, accumulating the weighted in-order value pairs and the weighted out-of-order value pairs, and evaluating the total weights of the weighted in-order value pairs and the weighted out-of-order value pairs by forming a ratio feature between the total weights. [8] Method according to the preceding claim, wherein the suspicion of offset is expressed when a predetermined threshold value for the ratio feature is exceeded. [9] Method according to the preceding claim, wherein the suspicion of offset is only expressed if the total weights each exceed a predetermined minimum sum. [10] Method according to one of the preceding claims, wherein, in the event of a suspected offset being detected, the pump cell (40) is regulated to a predetermined setpoint value for the pump voltage (U P ) of the pump cell (40), wherein upon reaching the predetermined setpoint value for the pump voltage (U P ) of the pump cell (40) a self-adjustment of the sensor element (10) is carried out. [11] Method according to the preceding claim, wherein the sensor element is used in an exhaust tract of an internal combustion engine (52), wherein the predetermined setpoint value for the pump voltage (U P ) of the pump cell (40) is adjusted to lambda = 1 for the exhaust gas of the internal combustion engine (52). [12] Method according to one of the two preceding claims, wherein the adjustment to the predetermined setpoint value is carried out by regulating the pump voltage (U P ) takes place. [13] A computer program arranged to carry out each step of the method according to any one of the preceding claims. [14] Electronic storage medium on which a computer program according to the preceding claim is stored. [15] Electronic control device (54) comprising an electronic storage medium according to the preceding claim.

Images