Emissions monitoring

Abstract

An exhaust monitoring unit, in an engine control module, is provided for management of an emissions load of an internal combustion engine. During operation of the engine the exhaust monitoring unit is

Description

Field of the disclosure The disclosure relates to the field of internal combustion engines, and in particular to emissions load in the exhaust of internal combustion engines. Background It is known that soot can be present in the exhaust of internal combustion engines. A high soot load may be caused by a variety of engine conditions or ambient conditions. For example, high soot loads may arise if an engine operates at a low duty factor and in cold ambient conditions, or if there are issues with engine calibration at low or high altitudes, or if there is deviation from nominal engine out conditions, or if face-plugging occurs, or in an event of component failure. Conventionally, soot levels are detected (for example the soot load of a diesel particulate filter (DPF) is measured) and in an event that the soot levels rise above a soot load threshold, corrective action may be taken by the engine control module to try to reduce the DPF soot levels. Reaching the soot load threshold takes time, and the corrective action may be effective or may have limited effect given the relatively high soot load. A high soot load may impact engine performance. Similarly, NOx conversion issues may arise in a Selective Catalytic Oxidation (SCR) system due to low operating temperatures, low DEF dosing rates, catalyst poisoning, and so on. Corrective actions may be taken to improve SCR performance. Conventionally, corrective action may be taken in an event that NOx levels rise above a NOx threshold. Summary of the disclosure Against this background, there is provided: an exhaust monitoring unit for management of an emissions load of an internal combustion engine, wherein the exhaust monitoring unit is comprised in an engine control module of the internal combustion engine and wherein during operation of the engine the exhaust monitoring unit is configured to: a. receive engine operating data from the engine recorded over a pre-defined time interval At0 starting at a time point tn; b. based on the engine operating data, identify a representative engine operating cycle, the representative engine operating cycle having a length shorter than the pre-defined time interval; c. based on the representative engine operating cycle, use a neural network to determine engine out boundary conditions; d. using the engine out boundary conditions, predict an emissions load over a future time interval; e. determine whether the predicted emissions load exceeds an emissions threshold within the future time interval; f. in an event that the predicted emissions load exceeds an emissions threshold within the future time interval, instruct an engine control module of the engine to take corrective action; and g. return to step a); wherein: at the first iteration of step a) after engine turn on, n=0; and n is incremented at each iteration of step a). In this way, high loads of emissions may be predicted, such that appropriate actions may be taken to lower the emissions load earlier than waiting to detect high emissions loads. Taking earlier appropriate actions may help to avoid downtime of a vehicle or machine, may improve engine performance, and may provide earlier warnings that engine servicing may be required. There is also provided a method of managing an emissions load of an internal combustion engine wherein during operation of the engine the method comprises: a. receiving engine operating data from the engine recorded over a pre-defined time interval At0 starting at a time point tn; b. based on the engine operating data, identifying a representative engine operating cycle, the representative engine operating cycle having a length shorter than the pre-defined time interval; c. based on the representative engine operating cycle, using a neural network to determine engine out boundary conditions; d. using the engine out boundary conditions, predicting an emissions load over a future time interval; e. determining whether the predicted emissions load exceeds an emissions threshold within the future time interval; f. in an event that the predicted emissions load exceeds an emissions threshold within the future time interval, instructing an engine control module of the engine to take corrective action; and g. returning to step a); wherein: at the first iteration of step a) after engine turn on, n=0; and n is incremented at each iteration of step a) Brief description of the drawings A specific embodiment of the disclosure will now be described, by way of example only, with reference to the accompanying drawings in which: Figure 1 shows a flowchart illustrating steps carried out by an exhaust monitoring unit according to an embodiment of the present disclosure. Figure 2 shows a schematic of time periods over which engine operating data is received from the engine according to an embodiment of the present disclosure. Figure 3 shows a flowchart illustrating steps carried out by an exhaust monitoring unit according to an embodiment of the present disclosure. Detailed description An exhaust monitoring unit is provided for management of an emissions load of an internal combustion engine. The exhaust monitoring unit is comprised in an engine control module of the internal combustion engine. The emissions load may comprise an emissions load of a specific emission component. For example, in certain embodiments, the emissions load may comprise a soot load. In other embodiments, the emissions load may comprise a NOx load. The emission load may comprise an emission load of another specific emission component. With reference to Figure 1, during operation of the engine the exhaust monitoring unit is configured to carry out steps 110 to 160. Steps 110 to 160 may be repeated during operation of the engine. Operation of the engine may be when the engine is turned on, or may be when a vehicle comprising the engine is turned on, or may be when the engine is running, or may be a shift of a work vehicle comprising the engine. Operation of the engine may be when the engine control module (ECM) is providing data. In an example, a period of operation of the engine may be from when the engine is turned on (such as when the ignition key is turned on) until the engine is turned off. Steps 110 may start at the beginning of a period of operation of the engine. Steps 110 to 160 may then be repeated during operation of the engine and may stop at the end of a period of operation of the engine. At step 110, the exhaust monitoring unit is configured to receive engine operating data from the engine. The engine operating data is recorded over a pre-defined time interval starting at a time point t0. For the first iteration of step 110 during operation of the engine, n = 0. For each subsequent iteration of step 110, n is incremented by one. Based on the engine operating data received at step 110, at step 120 the exhaust monitoring unit is configured to identify a representative engine operating cycle. The representative engine operating cycle has a length shorter than the pre-defined time interval. Based on the representative engine operating cycle identified at step 120, at step 130 the exhaust monitoring system is configured to use a neural network to determine engine out boundary conditions. Using the engine out boundary conditions, the exhaust monitoring system is configured to predict an emissions load over a future time interval at step 140. At step 150, the exhaust monitoring system is configured to determine whether the predicted emissions load exceeds an emissions threshold within the future time interval. In an event that the predicted emissions load exceeds an emissions threshold within the future time interval, at step 160 the exhaust monitoring system is configured to instruct an engine control module of the engine to take corrective action. The exhaust monitoring system is configured to return to step 110. With reference to Figure 2, at step 110, the engine operating data is received over the predefined time interval At0 starting at a time point t0 for the first iteration of step 110, such that the engine operating data is received from time t0 to time t0 + At0. The engine operating data may comprise data recorded by the ECM over the pre-determined time interval At0. The pre-determined time interval At0 may be of any length. The pre-determined time interval At0 may be shorter than a typical length of time over which the engine is operated. In a specific example, the pre-determined time interval At0 may be in the region of 2 hours However, the pre-determined time interval At0 may be greater than 2 hours, or less than 2 hours. t0 may be the start of operation of the engine, for example ignition on. At step 120, the exhaust monitoring unit is configured to identify a representative engine operating cycle Co based on the data received over the pre-defined time interval At0. At the next iteration of step 110, n=1 and the engine operating data may be received over the pre-defined time interval At0 starting at a time point = + At1; wherein A^ is a time period. In other words, engine operating data is received from time = + A^ to time t1 + At0. At step 120 where n=1, the exhaust monitoring unit is configured to identify a representative engine operating cycle Ci based on the data received over the pre-defined time interval At0. In certain embodiments, time period A^ may be the same length as the pre-determined time interval At0, such that step 110 is repeated every pre-determined time interval At0. In other embodiments, time period A^ may be a different length to the predetermined time interval, wherein steps 110 to 160 are repeated at a time that is a time period A^ after t0. In a particular example, the pre-determined time interval Atomay be at least two hours. The time period A^ may be around 30 minutes, such that the steps 110 to are repeated every 30 minutes and the engine operating data over the pre-defined time interval may be updated to reflect the new start time of the pre-defined interval. However, the pre-determined time interval and / or the time period A^ may differ from that example. At the next iteration of step 110, n=2 and the engine operating data may be received over the pre-defined time interval At0 starting at a time point t2 = + At2, wherein At2 is a time period. In other words, engine operating data is received from time t2 = t1 + At2 to time t2 + At0. At step 120 where n=2, the exhaust monitoring unit is configured to identify a representative engine operating cycle C2 based on the data received over the pre-defined time interval At0. In certain embodiments, time period At2 may be the same length as A^. In other embodiments, time period At2 may be a different length to A^. At an iteration of step 110 where n=n, the engine operating data may be received over the pre-defined time interval At0 starting at a time point tn = + Mn, wherein Atn is a time period. In other words, engine operating data is received from time tn = tn_1 + btn to time tn + At0. At step 120 where n=n, the exhaust monitoring unit is configured to identify a representative engine operating cycle Cn based on the data received over the pre-defined time interval At0. In certain embodiments, time period may be the same length for any value of n, such that step 110 is repeated every htn. In that case, the pre-determined time interval At0 may start at a time point tn = t0 + nMn. In other embodiments, time period Atn may be a different length for different values of n. Identifying a representative engine operating cycle Cn at step 120 uses the engine operating data received at step 110. The engine operating data my comprise engine speed and engine load over the pre-determined time period At0. The representative engine operating cycle Cn has a length that is shorter than the pre-determined time period over which the engine operating data was received. Identifying the representative engine operating cycle Cn at step 120 may be via a data sorting algorithm. The data sorting algorithm may comprise time series data over each pre-defined time interval of engine speed and load, segregated into different groups of speed and load. With reference to Figure 3, a flowchart illustrates a method comprising the method steps illustrated in Figure 1, wherein the data sorting algorithm may comprise assessing repeatability of engine speed and loads over the pre-determined time interval at step 310; weighting engine speed and load points based on a length of time that the engine spends at each speed and load at step 320; and identifying transient and steady portions of engine speed and load at step 330. The representative engine operating cycle may be stitched using the transient and steady portions of the engine speed and load. Steps 310, 320, 330 and the stitching may be carried out by the data sorting algorithm to create the representative engine operating cycle. In a particular example, the pre-determined time interval may be at least two hours in length. The representative engine operating cycle may, for example, be 15 minutes or 30 minutes in length. The representative engine operating cycle may be updated with each iteration of step 120, based on the updated data received at step 110. At step 130, the exhaust monitoring system is configured to use a neural network to determine engine out boundary conditions. A neural network is an empirical model, and may be used to predict engine performance in quick time. For example, a neural network may be used to predict on or more of specific fuel consumption, power, temperature, and emissions. To build a neural network model, Design of Experiments (DoE) test bed data may be used. DoE is a method used to run tests with combinations of inputs to generate a response. Here, combinations of inputs such as engine speed, engine load, engine rail pressure, main timing, and so on, may be used. Tests with combinations of those inputs may be used to generate an output of the engine, such as engine out emissions, temperature, power, fuel consumption and so on. In this case, the DoE may be used to generate engine out boundary conditions. A plurality of DoEs may be constructed by creating boundaries with variable inputs, such that the DoEs provide characteristic behaviour of engine performance as a function of the inputs. The variable inputs may comprise one or more of engine speed, engine load and / or one or more combustion variables such as start of injection, rail pressure, etc. The DoEs may provide characteristic behaviour of performance of the engine as a function of the DoE inputs. The DoEs may be used to create neural networks, wherein the neural networks predict engine performance. The neural networks predict engine out boundary conditions such as temperature, emissions and mass flow. If required, correction factors may be applied to represent the real scenario. At step 140, the emissions load is predicted at a future time interval. The future time interval may start when step 140 is carried out, or may start at a later time. The future time interval may, for example, start at around t = t0 + nM1 + pre-determined time interval. The future time interval may be of the order of 10 hours. For example, the future time interval may be 10 to 50 hours in length. The emissions load may be predicted over the future time interval by predicting an emissions load at time points within the future time interval. For example, the emissions load may be predicted for time points that are every 10 hours within the future time interval. The time points may be closer together, or further apart. Or, the emissions load may be predicted continuously over the future time interval. The emissions load may be predicted by looping the representative engine operating cycle using the engine out boundary conditions. This may be passed to a physics based model. In other words, the emissions load may be based on an after treatment physics-based model by passing engine out conditions from a neural network to the representative engine operating cycle. A physics-based model is based on first principles (momentum, energy, mass flow conservation). An after-treatment system may, for a diesel engine, comprise a diesel oxidation catalyst (DOC), a diesel particulate filter (DPF), a selective catalytic reduction (SCR) and an ammonia oxidation catalyst (AMOx). The emissions load may be a load of any specific emissions component found in or removed by any component of the after-treatment system. The physics-based model may be for one or more components of the after-treatment system. In an event that the emissions load comprises a soot load, the physics-based model may comprise a diesel particulate filter performance model. In certain embodiments, the physics-based model comprises a diesel oxidation catalyst performance model, wherein outputs of the diesel oxidation catalyst (DOC) performance model are passed to the diesel particulate filter (DPF) performance model. The DOC model may predict the DOC NO2 make up as a function of the engine out conditions. The outputs of the DOC model may be passed to the DPF model, which may compute the soot load by accounting some or all of incoming soot, outgoing soot, soot lost to chemical reactions (such as passive regeneration), DPF inlet temperature, mass flow rates and initial soot load. The DOC and / or DPF models may comprise thermal and / or kinetic models. A thermal model is configured to predict output temperatures of a DOC / DPF catalyst brick based on input conditions such as mass flow, temperatures and sizing of the catalyst brick (length, diameter and cells per square inch (CPSI)). A kinetic model predicts how chemistry influences the DOC / DPF catalyst bricks and predicts how output concentrations of exhaust emissions like soot, NOx (NO ,NO2), HC and CO change when they pass through DOC / DPF. Temperature, mass flow and sizing of catalyst brick may each influence exhaust species conversion, and may all be included in the model. In an event that the emissions load comprises a NOx load, wherein the physics-based model comprises a selective catalytic reduction performance model. In this way, tailpipe (TP) NOx may be forecasted. In an event that low NOx conversion is predicted to occur, corrective action may be taken such as changing an engine calibration. The corrective action carried out at step 160 may comprise one or more of: changing an engine calibration; active regeneration of after-treatment components (such as a diesel particulate filter or an SCR); and engine servicing. The exhaust monitoring unit may be configured to instruct the corrective action and / or notify a user that the corrective action is required. There is also provided a method of managing an emissions load of an internal combustion engine. The method comprises receiving engine operating data from the engine recorded over a pre-defined time interval starting at a time point tn. Based on the engine operating data, the method further comprises identifying a representative engine operating cycle, the representative engine operating cycle having a length shorter than the pre-defined time interval. Based on the representative engine operating cycle, the method comprises using a neural network to determine engine out boundary conditions. The method further comprises using the engine out boundary conditions to predict an emissions load over a future time interval. The method determines whether the predicted emissions load exceeds an emissions threshold within the future time interval. In an event that the predicted emissions load exceeds an emissions threshold within the future time interval, the method comprises instructing an engine control module of the engine to take corrective action. The method returns to the first step of receiving engine operating data, wherein at the first iteration of the step after engine turn on, n=0, and wherein is incremented at each iteration of the step. The method may comprise any steps described with relation to the exhaust monitoring unit.

Claims

:CMCM1. An exhaust monitoring unit for management of an emissions load of an internal combustion engine, wherein the exhaust monitoring unit is comprised in an engine5 control module of the internal combustion engine and wherein during operation of the engine the exhaust monitoring unit is configured to:receive engine operating data from the engine recorded over a pre-defined timeinterval At0 starting at a time point tn;based on the engine operating data, identify a representative engine operating cycle, the representative engine operating cycle having a length shorter than the pre-defined time interval;based on the representative engine operating cycle, use a neural network to determine engine out boundary conditions;using the engine out boundary conditions, predict an emissions load over a future time interval;determine whether the predicted emissions load exceeds an emissions threshold within the future time interval;in an event that the predicted emissions load exceeds an emissions threshold within the future time interval, instruct an engine control module of the engine to take corrective action; andreturn to step a);wherein:at the first iteration of step a) after engine turn on, n=0; and n is incremented at each iteration of step a).

252. The exhaust monitoring unit of claim 1, wherein at the n,h iteration of step a) the exhaust monitoring unit is configured to receive engine operating data from the engine recorded over a pre-defined time interval At0 starting at a time point tn = Ln_r +30 3. The exhaust monitoring unit of claim 1 or 2, wherein the emissions load comprises asoot load.

4. The exhaust monitoring unit of claim 1 or 2, wherein the emissions load comprises a NOx load.a.b.101520c.d.e.f.

355. The exhaust monitoring unit of any preceding claim, wherein the engine operating data comprises engine speed and engine load over the pre-determined time interval.

6. The exhaust monitoring unit of claim 5, wherein identifying the representative engine5 operating cycle is via a data sorting algorithm comprising:assessing repeatability of engine speed and loads over the pre-determined time interval;weighting engine speed and load points based on a length of time that the engine spends at each speed and load; and10 identifying transient and steady portions of engine speed and load.

7. The exhaust monitoring unit of any preceding claim, wherein the neural network used to predict engine out boundary conditions is trained using engine data.15 8. The exhaust monitoring unit of claim 7, wherein the neural network is built using Designof Experiments, DoE, test bed data, wherein:a plurality of DoEs are constructed by creating boundaries with variable inputs, such that the DoEs provide characteristic behaviour of engine performance as a function of the inputs; and20 the DoEs are used to create neural networks to predict engine out boundaryconditions; andwherein optionally correction factors are applied to the neural network based on the operation of the engine.25 9. The exhaust monitoring unit of any preceding claim, wherein predicting the emissionsload is based on an after treatment physics-based model by passing engine out conditions from a neural network to the representative engine operating cycle.

10. The exhaust monitoring unit of claim 9, wherein the emissions load comprises a soot30 load and wherein physics-based model comprises a diesel particulate filter performance model, wherein optionally the physics-based model comprises a diesel oxidation catalyst performance model, wherein outputs of the diesel oxidation catalyst performance model are passed to the diesel particulate filter performance model.

11. The exhaust monitoring unit of claim 10, wherein the diesel particulate filter performance model uses as input at least one of:incoming soot level;outgoing soot level;soot consumption via chemical reaction;temperature;mass flow rates;emissions; andinitial soot load.1012. The exhaust monitoring unit of claim 9, wherein the emissions load comprises a NOx load and wherein the physics-based model comprises a selective catalytic reduction performance model.CM15CM13. The exhaust monitoring unit of any preceding claim, wherein the corrective action comprises one or more of:changing an engine calibration;active regeneration of an after-treatment component; andengine servicing.2014. The exhaust monitoring unit of any preceding claim, wherein the future time interval is between the next 10 and 50 hours.

15. A method of managing an emissions load of an internal combustion engine wherein 25 during operation of the engine the method comprises:a. receiving engine operating data from the engine recorded over a pre-defined time interval At0 starting at a time point fn;b. based on the engine operating data, identifying a representative engine operating cycle, the representative engine operating cycle having a length 30 shorter than the pre-defined time interval;c. based on the representative engine operating cycle, using a neural network to determine engine out boundary conditions;d. using the engine out boundary conditions, predicting an emissions load over a future time interval;CMCMe. determining whether the predicted emissions load exceeds an emissions threshold within the future time interval;f. in an event that the predicted emissions load exceeds an emissions threshold within the future time interval, instructing an engine control module of the engine5 to take corrective action; andg. returning to step a);wherein:at the first iteration of step a) after engine turn on, n=0; and n is incremented at each iteration of step a).

Images