JP2019167948A - Catalyst state estimation device, method for estimating state of catalyst, and computer program - Google Patents

Abstract

To improve accuracy of estimation in a technique for estimating purification performance of a catalyst for purifying exhaust of an internal combustion engine.SOLUTION: A catalyst state estimation device comprises: a first acquisition unit provided in a main flow passage through which exhaust from an internal combustion engine is passed, and acquiring information of a catalyst for purifying a hazardous material in exhaust; a storage unit for previously storing a catalyst state estimation model including at least one mathematical model; and an estimation unit for estimating purification performance of the catalyst by applying the information of the catalyst acquired by the first acquisition unit to the catalyst state estimation model.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technique for estimating the state of a catalyst.

  A technique for quantitatively predicting a phenomenon under various conditions using a mathematical model is known. As one of such mathematical models, a neural network (NN: Neural Network) in which a neural network in the human brain is expressed by a mathematical model called an artificial neuron is known.

  For example, in Patent Document 1, each value of ambient temperature, manifold pressure and temperature, fuel consumption rate, and engine rotation speed is input to NN and discharged from a selective reduction catalyst (SCR catalyst: Selective Catalytic Reduction catalyst). It is described that the amount of nitrogen oxides (NOx) is predicted. For example, in Patent Document 2, the EGR valve lift amount command value, the supercharging pressure, the intake air temperature, the exhaust pressure, the fuel injection amount, the intake air flow rate, and the engine speed are input to NN, and NOx occlusion reduction. It is described that the amount of NOx trapped in a catalyst (NSR catalyst: NOx Storage Reduction catalyst) is predicted. For example, in Patent Document 3, the values of engine speed, fuel injection amount, fuel injection timing, intake air amount, air-fuel ratio, exhaust temperature, and supercharging pressure are input to NN, and NOx discharged from the engine It is described to predict the amount of.

JP 2003-328732 A JP 2009-180086 A JP 2011-132915 A

  By the way, for example, SCR catalyst, NSR catalyst, three-way catalyst (Three-Way Catalyst), particulate matter removal filter (DPF: Diesel Particulate Filter), oxidation catalyst (DOC catalyst: Diesel Oxidation Catalyst), etc. It is known to purify the exhaust of an internal combustion engine using a catalyst alone or in combination. There has been a demand for estimating (predicting) the purification performance of such a catalyst by a mathematical model.

  However, in each of the techniques described in Patent Documents 1 to 3, only parameters relating to the internal combustion engine (engine) and its intake system are assumed as inputs to the NN. Here, since the purification performance of the catalyst fluctuates depending on the temperature of the catalyst, the amount of the additive adsorbed on the catalyst, etc., all of the techniques described in Patent Documents 1 to 3 use the catalyst. There has been a problem that the purification performance of water cannot be estimated with high accuracy.

  Further, in the techniques described in Patent Documents 1 to 3, in a configuration for purifying exhaust gas of an internal combustion engine using a plurality of catalysts (for example, a configuration in which two or more SCR catalysts are provided), the purification performance by the plurality of catalysts is estimated. It is not considered to do.

  The present invention has been made to solve at least a part of the above-described problems, and an object of the present invention is to improve estimation accuracy in a technique for estimating the purification performance of a catalyst that purifies exhaust gas from an internal combustion engine. .

  SUMMARY An advantage of some aspects of the invention is to solve at least a part of the problems described above, and the invention can be implemented as the following forms.

(1) According to one aspect of the present invention, a catalyst state estimation device is provided. This catalyst state estimation device is provided in a main flow path through which exhaust from an internal combustion engine flows, and includes a first acquisition unit that acquires information on a catalyst that purifies harmful substances in the exhaust, and at least one mathematical model. A storage unit that preliminarily stores a catalyst state estimation model; and an estimation unit that estimates the purification performance of the catalyst by applying the catalyst information acquired by the first acquisition unit to the catalyst state estimation model. Prepare.

  The purification performance of the catalyst varies depending on the information of the catalyst (for example, the temperature of the catalyst and the amount of additive adsorbed on the catalyst). According to this configuration, the estimation unit can accurately estimate the catalyst purification performance by applying the catalyst information that affects the catalyst purification performance to the catalyst state estimation model. That is, according to this configuration, it is possible to improve the estimation accuracy in the technique for estimating the purification performance of the catalyst that purifies the exhaust gas of the internal combustion engine.

(2) The catalyst state estimation apparatus of the above aspect further includes a second acquisition unit that acquires information on the exhaust gas flowing into the catalyst, and the estimation unit further includes the first information in addition to the information on the catalyst. 2 The purification performance of the catalyst may be estimated by applying the exhaust gas information acquired by the acquisition unit to the catalyst state estimation model. The purification performance of the catalyst fluctuates in addition to the information on the catalyst, and is affected by the information on the exhaust gas flowing into the catalyst (for example, the exhaust gas temperature, the exhaust gas flow rate, and the amount of nitrogen oxide in the exhaust gas). According to this configuration, the estimation unit applies the catalyst information that affects the catalyst purification performance and the information on the exhaust gas flowing into the catalyst to the catalyst state estimation model, thereby improving the catalyst purification performance. Further, it can be estimated with high accuracy.

(3) The catalyst state estimation apparatus of the above aspect further includes a third acquisition unit that acquires information on the additive supplied to the catalyst, and the estimation unit further includes the information on the catalyst. The purification performance of the catalyst may be estimated by applying the information on the additive acquired by the third acquisition unit to the catalyst state estimation model. According to this configuration, the estimation unit estimates the purification performance of the catalyst by applying the information on the additive (for example, the amount of the additive) to the catalyst state estimation model in addition to the information on the catalyst. For this reason, for example, the purification performance of catalysts that purify harmful substances using additives such as selective reduction catalyst (SCR catalyst: Selective Catalytic Reduction catalyst) and storage reduction catalyst (NSR catalyst: NOx Storage Reduction catalyst). In the case of estimation, the accuracy of estimation can be further improved.

(4) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is configured by a machine learning model, and as the catalyst information, the temperature of the front end of the catalyst in the main channel, the temperature of the catalyst, Including a first model that receives at least one of the adsorbed amount of the additive adsorbed on the catalyst one hour before and outputs the purification rate of nitrogen oxides in the catalyst; May estimate the purification rate of the nitrogen oxides as the purification performance of the catalyst using the first model. In machine learning models, if there is a causal relationship between input and output variables, classification and regression cannot be performed unless complex function approximation is performed (for example, if phenomena that are difficult to specify with physical formulas are included, describe the phenomenon) When a plurality of physical formulas are required, the output (estimation result) can be obtained with a low calculation load even if the input to the physical formula increases because the influence of many factors is assumed. According to this configuration, since the estimation unit estimates the nitrogen oxide purification rate using the first model configured by the machine learning model, the estimation unit estimates the nitrogen oxide purification rate affected by many factors. , Low computational load and high speed. Moreover, the estimation part can estimate the purification rate of nitrogen oxides with high accuracy by using the fully learned machine learning model as the first model.

(5) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is configured by a machine learning model, and as the catalyst information, the temperature of the front end of the catalyst in the main channel, the temperature of the catalyst, At least one of the adsorption amount of the additive adsorbed on the catalyst one hour before, the time differential value of the temperature of the catalyst, and the ratio of the adsorption amount to the saturated adsorption amount of the additive in the catalyst. The estimation unit estimates the outflow amount as the purification performance of the catalyst using the second model. The second model uses the second model as an output and outputs the outflow amount of the additive flowing out from the catalyst. May be. According to this configuration, the estimation unit estimates the outflow amount of the additive flowing out from the catalyst using the second model configured by the machine learning model. , Low computational load and high speed. Moreover, the estimation part can estimate the outflow amount with high accuracy by using the sufficiently learned machine learning model as the second model.

(6) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is configured by a physical model, and as the catalyst information, the temperature of the front end of the catalyst in the main channel and the temperature of the catalyst A third model that receives at least one of them and outputs an amount of an additive that does not contribute to the nitrogen oxide purification reaction in the catalyst, and the estimation unit uses the third model to generate the catalyst. The amount of the additive that does not contribute to the nitrogen oxide purification reaction in the catalyst as the purification performance may be estimated. Since the physical model is a law that is based on the similar physical rule based on the actual physical law, the output (estimated result) obtained by the physical model satisfies the physical law. On the other hand, since the machine learning model is a model constructed as a result of learning a huge amount of data, an output (estimation result) that satisfies the physical rule may not be obtained. According to this configuration, the estimation unit estimates the amount of the additive that does not contribute to the nitrogen oxide purification reaction in the catalyst using the third model configured by the physical model. The amount of the agent can be estimated with high accuracy according to the physical law.

(7) In the catalyst state estimation device of the above aspect, the catalyst state estimation model includes a current purification rate of nitrogen oxides estimated by the first model and a current outflow estimated by the second model. And the amount of additive that does not contribute to the nitrogen oxide purification reaction in the current catalyst estimated by the third model, and the amount of additive adsorbed on the catalyst at the next time, The model may be obtained using a physical law, and the estimation unit may estimate the adsorption amount at the next time as the purification performance of the catalyst using the catalyst state estimation model. According to this configuration, the estimation unit uses the first and second models that are machine learning models to estimate the results (nitrogen oxide purification rate, the outflow amount of the additive flowing out from the catalyst), and the third physical model. Combined with the estimation result by the model (the amount of additive that does not contribute to the nitrogen oxide purification reaction in the catalyst), the adsorption amount of the additive to the catalyst is estimated using the physical law. For this reason, according to this configuration, it is possible to estimate the amount of adsorption affected by more factors with high accuracy and at high speed while satisfying the physical law. Further, the amount of additive adsorbed on the catalyst varies under the influence of the amount of adsorption at the previous time (in other words, varies under the influence of the time history). According to this configuration, the estimation unit uses the current estimation results of the first to third models to estimate the amount of adsorption at the next time, and therefore, based on the purification performance of the catalyst at the previous time, The purification performance of the catalyst at the time can be estimated with high accuracy.

(8) In the catalyst state estimation device according to the above aspect, the first model and the second model may further include, as the exhaust information, the temperature of the exhaust, the flow rate of the exhaust, and the oxidation of nitrogen contained in the exhaust. At least one of the quantity of the substance is input, and as the information on the additive, the inflow of the additive flowing into the catalyst is input, and the third model is further used as the information on the exhaust. The temperature of the exhaust gas may be input, and the amount of the additive flowing into the catalyst may be input as the additive information. According to this configuration, in the first to third models, estimation is performed in consideration of various kinds of information, so that the purification performance of the catalyst can be estimated with higher accuracy.

(9) In the catalyst state estimation device of the above aspect, the catalyst state estimation model includes a plurality of the first models created using teacher data acquired from a plurality of the catalysts having different degrees of deterioration, and a degree of deterioration. A plurality of second models created using teacher data obtained from a plurality of different catalysts. According to this configuration, the catalyst state estimation model includes a plurality of first and second models created using teacher data acquired from a plurality of catalysts having different degrees of deterioration, and accordingly, according to the degree of deterioration of the catalyst. The optimal first model and second model can be adopted. As a result, according to this configuration, the purification performance of the catalyst can be estimated with higher accuracy.

(10) According to one aspect of the present invention, a method for estimating the state of a catalyst is provided. In this method, an information processing apparatus including a storage unit that stores in advance a catalyst state estimation model including at least one mathematical model is provided in a main flow path through which exhaust from an internal combustion engine flows, and harmful substances in the exhaust are removed. Obtaining information on the catalyst to be purified, and estimating the purification performance of the catalyst by applying the information on the catalyst obtained in the obtaining step to the catalyst state estimation model.

(11) According to one aspect of the present invention, a computer program is provided. In this computer program, an information processing apparatus including a storage unit that stores in advance a catalyst state estimation model including at least one mathematical model is provided in a main flow path through which exhaust from an internal combustion engine flows, and harmful substances in the exhaust And a function of estimating the purification performance of the catalyst by applying the information of the catalyst obtained in the obtaining step to the catalyst state estimation model. .

(12) In the catalyst state estimation device according to the above aspect, when a plurality of the same type of catalysts are provided in the main flow path, the first acquisition unit is configured to provide information on the catalyst located at the most upstream position in the main flow path. The estimation unit may estimate the purification performance of the plurality of catalysts as a whole by applying the information on the catalyst acquired by the first acquisition unit to the catalyst state estimation model. . According to this configuration, when a plurality of catalysts are provided in the main flow path, the estimation unit applies information on the catalyst located in the uppermost stream to the catalyst state estimation model, thereby The purification performance as can be estimated. That is, according to this configuration, the purification performance can be estimated by regarding a plurality of catalysts on the main flow path as one catalyst, and therefore, compared with the case where the purification performance of each catalyst is estimated, the catalyst information acquisition unit ( The number of sensors, etc.) and the number of catalyst state estimation models prepared in advance can be reduced, and the calculation load on the catalyst state estimation device can be reduced. In addition, when the plurality of catalysts provided in the main flow path are the same type of catalyst, the estimation unit regards these as one catalyst and estimates the purification performance. If the catalyst is of the same type, there is no difference in the information items of the catalyst that affect the purification performance (for example, the temperature of the catalyst and the amount of additive adsorbed on the catalyst), so the estimation unit accurately estimates the purification performance. can do.

(13) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is configured by a machine learning model, and as the catalyst information, the temperature of the front end of the catalyst located at the uppermost stream and the temperature of each catalyst And at least one of the total adsorbed amounts of additives adsorbed on the plurality of catalysts one hour before the input, and an output of the total amount of nitrogen oxide purification in the plurality of catalysts The estimation unit may estimate a purification amount of the nitrogen oxide as the purification performance using the first model. According to this configuration, since the estimation unit estimates the total amount of nitrogen oxide purification in the plurality of catalysts using the first model configured by the machine learning model, the nitrogen oxide affected by many factors The amount of purification can be estimated at a low computational load and at high speed. Further, by using the fully learned machine learning model as the first model, the estimation unit can estimate the nitrogen oxide purification amount with high accuracy.

(14) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is configured by a machine learning model, and as the catalyst information, the temperature of the front end of the catalyst located in the uppermost stream and the temperature of each catalyst And the sum of the adsorption amounts of the additives adsorbed on the plurality of catalysts one hour before, the time differential value of the temperature of each catalyst, and the sum of the saturated adsorption amounts of the additives in the plurality of catalysts. A second model having at least one of the total ratio of the adsorbed amounts as an input and outputting an outflow amount of the additive flowing out from the catalyst located on the most downstream side in the main flow path; May estimate the outflow amount as the purification performance using the second model. According to this configuration, the estimation unit estimates the amount of the additive flowing out from the catalyst located at the most downstream using the second model configured by the machine learning model, and therefore, more factors affect the estimation unit. The amount of outflow can be estimated with a low computational load and high speed. Moreover, the estimation part can estimate the outflow amount with high accuracy by using the sufficiently learned machine learning model as the second model.

(15) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is configured by a physical model, and as the catalyst information, the temperature of the front end of the catalyst located in the uppermost stream, the temperature of each catalyst, Among the additives supplied to the most upstream catalyst, the amount of the additive that does not contribute to the nitrogen oxide purification reaction in the plurality of catalysts is output. Including a third model, the estimating unit may estimate the amount of the additive that does not contribute to the nitrogen oxide purification reaction as the purification performance using the third model. According to this configuration, the estimation unit uses the third model configured by the physical model to perform the nitrogen oxide purification reaction in the plurality of catalysts among the additives supplied to the catalyst located at the uppermost stream. Since the amount of additive that does not contribute is estimated, it is possible to estimate the amount of the additive that has few influential factors with high accuracy in accordance with physical laws.

(16) In the catalyst state estimation device of the above aspect, the catalyst state estimation model includes a current total amount of nitrogen oxide purification estimated by the first model and a current amount estimated by the second model. Additives adsorbed on the plurality of catalysts at the next time with the outflow amount and the amount of additive that does not contribute to the current nitrogen oxide purification reaction estimated by the third model as inputs Is a model for obtaining the total amount of adsorption using physical laws, and the estimation unit estimates the total amount of adsorption at the next time as the purification performance using the catalyst state estimation model. Also good. According to this configuration, the estimation unit estimates the results of the first and second models that are machine learning models (the total amount of nitrogen oxide purification in the plurality of catalysts, the amount of the additive flowing out from the catalyst located at the most downstream position, Outflow amount) and estimation results based on the third model, which is a physical model (the amount of additive that does not contribute to the nitrogen oxide purification reaction in a plurality of catalysts among the additives supplied to the catalyst located in the uppermost stream) In combination, and the physical quantity is used to estimate the total adsorbed amount of the additive adsorbed on the plurality of catalysts. For this reason, according to this configuration, it is possible to estimate the total amount of adsorption affected by a larger number of factors with high accuracy and high speed while satisfying the physical law. Further, the total amount of additives adsorbed by the plurality of catalysts varies under the influence of the total amount of adsorption at the previous time (in other words, varies under the influence of the time history). According to this configuration, the estimation unit uses the current estimation results of the first to third models to estimate the total amount of adsorption at the next time, and therefore, based on the purification performance of the catalyst at the previous time, The purification performance of the catalyst at the next time can be estimated with high accuracy.

(17) In the catalyst state estimation device according to the above aspect, the second acquisition unit that acquires information on the exhaust gas flowing into the most upstream catalyst, and the addition that is supplied to the most upstream catalyst A third acquisition unit that acquires information on the agent, and the first model and the second model further include the exhaust temperature, the exhaust flow rate, and the exhaust as the exhaust information. At least one of the amount of nitrogen oxides to be input, and as the information on the additive, the inflow amount of the additive flowing into the uppermost catalyst is input, and the third model is Further, the exhaust gas temperature may be input as the exhaust gas information, and the additive flow rate of the additive flowing into the most upstream catalyst may be input as the additive gas information. In addition to catalyst information, the purification performance of the catalyst further includes information on the exhaust flowing into the catalyst (for example, exhaust temperature, exhaust flow rate, amount of nitrogen oxide in the exhaust), and information on additives (for example, Fluctuates under the influence of the amount of additive). According to this configuration, the estimation unit applies the catalyst information that affects the catalyst purification performance, the exhaust information, and the additive information to the catalyst state estimation model, thereby improving the catalyst purification performance. Further, it can be estimated with high accuracy.

(18) In the catalyst state estimation device of the above aspect, the catalyst state estimation model is the most upstream in the exemplary main flow channel for a catalyst group including a plurality of the same type of catalysts arranged in the exemplary main flow channel. Teacher data obtained from the catalyst group, assuming that the inlet of the catalyst located is regarded as the inlet of the catalyst group, and that the outlet of the catalyst located at the most downstream in the exemplary main flow path is regarded as the outlet of the catalyst group. It may include at least one model of the first model and the second model created using. According to this configuration, the first and / or second model of the catalyst state estimation model can be constructed by learning a plurality of catalysts arranged in the exemplary main flow path as one catalyst (catalyst group). For this reason, in the first and / or second model of the catalyst state estimation model, the influence of the main flow path (exhaust pipe or the like) between the catalysts belonging to the catalyst group can be taken into consideration. By using the catalyst state estimation model constructed in this way, the estimation unit does not need information such as exhaust pipes between the catalysts on the main flow path, and accurately estimates the purification performance as a whole of a plurality of catalysts. can do.

(19) The catalyst state estimation apparatus according to the above aspect may further include a temperature estimation unit that estimates the temperature of the other catalyst provided in the main flow path from the temperature of the catalyst located in the uppermost stream. According to this configuration, since the temperature estimation unit can estimate the temperature of the other catalyst provided in the main flow path from the temperature of the catalyst located at the uppermost stream, the acquisition unit acquires the temperature of the other catalyst. (Sensors etc.) can be omitted.

  It should be noted that the present invention can be realized in various modes. For example, a catalyst state estimation device and system, an exhaust purification device and an exhaust purification system including the catalyst state estimation device, a control method for these devices and systems, these The present invention can be realized in the form of a computer program executed in the apparatus and system, a server device for distributing the computer program, a non-temporary storage medium storing the computer program, and the like.

1 is a block diagram of an exhaust purification system as one embodiment of the present invention. It is a figure explaining a 1st model. It is a figure explaining a 3rd model. It is a flowchart which shows the procedure of the estimation process by an estimation part. It is a figure explaining step S24 of an estimation process. It is a block diagram of the exhaust gas purification system in 2nd Embodiment. It is a block diagram of the exhaust gas purification system in 3rd Embodiment. It is a block diagram of the exhaust gas purification system in 4th Embodiment. It is a block diagram of the exhaust gas purification system in 5th Embodiment. It is a block diagram of the exhaust gas purification system in 6th Embodiment. It is a flowchart which shows the procedure of the estimation process by the estimation part in 6th Embodiment. It is a block diagram of the exhaust gas purification system in 7th Embodiment.

<First Embodiment>
FIG. 1 is a block diagram of an exhaust purification system 1 as an embodiment of the present invention. The exhaust purification system 1 of this embodiment includes a combustion state control unit 91, an internal combustion engine 92, an exhaust purification device 20, and a catalyst state estimation device 10. The exhaust purification device 20 of this embodiment is a device that purifies harmful substances (nitrogen oxides: NOx) in the exhaust of the internal combustion engine 92, and includes a selective reduction catalyst (SCR catalyst) 40 as a catalyst. As an additive, urea water that generates ammonia (NH ₃ ) is used. The catalyst state estimation device 10 of the present embodiment can estimate the state of the SCR catalyst 40 mounted on the exhaust purification device 20, specifically, the purification performance of the SCR catalyst 40.

  The internal combustion engine 92 is, for example, a diesel engine or a lean burn operation type gasoline engine. The combustion state control unit 91 controls the air / fuel ratio in the internal combustion engine 92 to lean, stoichiometric, and rich states by controlling the injection of air and fuel to the internal combustion engine 92. The combustion state control unit 91 is implemented by, for example, an electronic control unit (ECU, Electronic Control Unit). In the following description, the side of the exhaust purification device 20 that is closer to the internal combustion engine 92 is referred to as “upstream side”, and the side that is farther from the internal combustion engine 92 is referred to as “downstream side”. In the case of FIG. 1, the left side corresponds to the upstream side, and the right side corresponds to the downstream side.

  The exhaust purification device 20 includes an exhaust pipe 30 extending from the internal combustion engine 92, an SCR catalyst 40 provided on the exhaust pipe 30, a urea pump unit 62, and a urea nozzle 64. The exhaust pipe 30 forms a main flow path through which exhaust from the internal combustion engine 92 flows. Exhaust gas from the internal combustion engine 92 passes through the main flow path in the exhaust pipe 30, passes through the SCR catalyst 40, and is released to the outside air. The SCR catalyst 40 receives the supply of the additive and purifies NOx in the exhaust. The SCR catalyst 40 corresponds to a “catalyst”. The urea pump unit 62 stores therein urea water as an additive and incorporates a pump for sending urea water to the urea nozzle 64. The urea nozzle 64 is an injection port for urea water, and is provided on the upstream side of the SCR catalyst 40 to supply urea water to the SCR catalyst 40.

  The catalyst state estimation device 10 includes a CPU 11, a storage unit 12, a flow rate acquisition unit 52, a NOx concentration acquisition unit 54, a front end temperature acquisition unit 56, and a temperature acquisition unit 58. CPU11 and the memory | storage part 12 are mounted by ECU, for example.

  CPU11 controls each part of the catalyst state estimation apparatus 10 by expand | deploying and executing the computer program stored in ROM on RAM. In addition, the CPU 11 functions as the estimation unit 110, receives each acquired value acquired from the flow rate acquisition unit 52, the NOx concentration acquisition unit 54, the front end temperature acquisition unit 56, and the temperature acquisition unit 58, and executes an estimation process to be described later. To do. The storage unit 12 includes a flash memory, a memory card, a hard disk, and the like. The storage unit 12 stores a catalyst state estimation model 120 in advance. The catalyst state estimation model 120 includes a first model 121, a second model 122, and a third model 123.

  FIG. 2 is a diagram for explaining the first model 121. The first model 121 is a model for estimating the NOx purification rate in the SCR catalyst 40, and is configured by a machine learning model, specifically a neural network (NN). As shown in FIG. 2, the first model 121 of this embodiment is composed of three layers, an input layer, an intermediate layer, and an output layer. The intermediate layer is a single layer, and a sigmoid function is used for the elements of the intermediate layer. The case is illustrated. Note that the sigmoid function may not be used for the element in the intermediate layer.

The vector U that is an input variable of the first model 121 is composed of n components (u ₁ , u ₂ ,..., U _n , n is a natural number). Parameters shown in the following items a1 to a3 can be adopted for each component of the vector U.
(A1) Information on catalyst: At least one of the temperature of the front end of the SCR catalyst 40, the temperature of the SCR catalyst 40, and the adsorption amount of NH ₃ adsorbed on the SCR catalyst 40 one hour ago (a2 ) Exhaust information: at least one of the temperature of the exhaust from the internal combustion engine 92, the flow rate of the exhaust, and the amount of NOx contained in the exhaust (a3) Information on the additive: NH flowing into the SCR catalyst 40 ₃ inflow

  Here, for the vector U, it is necessary to input at least one or more of the catalyst information shown in the item a1, and the exhaust information and additive information shown in the items a2 and a3 are as follows. At least one or more may be input, or all may be omitted. However, in order to improve the estimation accuracy of the NOx purification rate in the first model 121, it is preferable that the number of input parameters is large.

After each component of the vector U is input to the input layer, the intermediate layer adds up the product of each component u _i of the vector U and the weight constant W _ij (Equation 1). Thereafter, as shown in Equation 2, the eigenvalue θ _j (bias) for each node in the intermediate layer is taken into account, and the sigmoid function is passed through and output. In the output layer, the product of the input value Y _j and the weight constant W _jk is added, and the sum with the eigenvalue θ _k (bias) is obtained and output (Formula 3). The output variable Z of the first model 121 is an estimated value of the NOx purification rate of the SCR catalyst 40 under various conditions represented by the input variable (vector U). Note that n represents the number of input variables, and m represents the number of intermediate layer nodes. The first model 121 corresponds to a “first model”.

The first model 121 causes the NN to learn the relationship between the input variable and the output variable so that the output variable Z matches the physical quantity to be estimated (NOx purification rate in the case of the first model 121). Values W _ij , θ _j , W _jk , θ _k are determined in advance. For learning the first model 121, it is preferable to use data acquired during transient operation of the internal combustion engine 92 as teacher data. Note that the number of nodes in the intermediate layer can be determined in consideration of, for example, accuracy with respect to teacher data and accuracy with respect to data that has not been used as teacher data.

The second model 122 is a model for estimating the outflow amount of NH ₃ flowing out from the SCR catalyst 40, and is configured by a machine learning model, specifically, NN. The second model 122 has the same configuration as the first model 121 described with reference to FIG.

The vector U as an input variable of the second model 122 is composed of n components (u ₁ , u ₂ ,..., U _n , n is a natural number). For each component of the vector U, parameters shown in the following items b1 to b3 can be adopted.
(B1) Information on the catalyst: the temperature of the front end of the SCR catalyst 40, the temperature of the SCR catalyst 40, the adsorption amount (actual adsorption amount) of NH ₃ adsorbed on the SCR catalyst 40 one hour ago, and the SCR catalyst 40 And the ratio of the adsorption amount (actual adsorption amount) to the saturated adsorption amount of NH ₃ in the SCR catalyst 40 (b2) Exhaust information: The exhaust information from the internal combustion engine 92 Information on additive (b3) of at least one of temperature, flow rate of exhaust gas, and amount of NOx contained in exhaust gas: Inflow amount of NH ₃ flowing into SCR catalyst 40

  For the vector U of the second model 122, as in the first model 121, it is necessary to input at least one or more of the catalyst information shown in the item b1, and the exhaust gas shown in the items b2 and b3. About information and the information of an additive, at least 1 or more may be input and all may be abbreviate | omitted. However, in order to improve the estimation accuracy of the outflow amount in the second model 122, it is preferable that the number of input parameters is large.

The output variable Z of the second model 122 is an estimated value of the outflow amount of NH ₃ that flows out of the SCR catalyst 40 and is lost under various conditions represented by the input variable (vector U). The second model 122 corresponds to a “second model”.

FIG. 3 is a diagram for explaining the third model 123. The third model 123 is a model for estimating the amount of the additive (NH ₃ ) that does not contribute to the NOx purification reaction in the SCR catalyst 40, and is configured by a physical model. It is known that NH ₃ is lost due to thermal decomposition and oxidation reaction particularly in a high temperature environment and does not contribute to the NOx purification reaction in the SCR catalyst 40. In the third model 123, the amount of NH ₃ that decreases due to the thermal decomposition and oxidation reaction is estimated as the amount of the additive (NH ₃ ) that does not contribute to the NOx purification reaction. As shown in FIG. 3, the third model 123 is configured by the Arrhenius equation. Of the Arrhenius equation, values obtained by experiments or the like in advance can be used for the frequency factor (A) and the activation energy (E).

The vector U as an input variable of the third model 123 is composed of n components (u ₁ , u ₂ ,..., U _n , n is a natural number). Parameters shown in the following items c1 to c3 can be adopted for each component of the vector U.
(C1) Information on the catalyst: at least one of the temperature of the front end of the SCR catalyst 40 and the temperature of the SCR catalyst 40 (c2) Information on exhaust: Temperature of exhaust from the internal combustion engine 92 (c3) Additive Information: NH ₃ inflow into SCR catalyst 40

Similar to the first model 121, it is necessary to input at least one or more of the catalyst information shown in the item c1 for the vector U of the third model 123, and the exhaust gas shown in the items c2 and c3. About information and the information of an additive, at least 1 or more may be input and all may be abbreviate | omitted. However, in order to improve the estimation accuracy of the amount of NH ₃ that does not contribute to the NOx purification reaction in the third model 123, it is preferable that the number of input parameters is large.

The output variable Z of the third model 123 is an estimated value of the amount of NH ₃ that does not contribute to the NOx purification reaction in the SCR catalyst 40 under various conditions represented by the input variable (vector U). The third model 123 corresponds to a “third model”.

  Returning to FIG. 1, the description will be continued. The flow rate acquisition unit 52 acquires the flow rate of exhaust from the internal combustion engine 92. The flow rate acquisition unit 52 may be realized, for example, by acquiring a measurement signal measured by a Pitot tube type flow meter provided in the exhaust pipe 30. The flow rate acquisition unit 52 may estimate the flow rate of the exhaust gas from the intake air amount signal to the internal combustion engine 92 or the fuel injection amount signal. The exhaust temperature acquisition unit 53 is a sensor that measures the temperature of exhaust from the internal combustion engine 92. The NOx concentration acquisition unit 54 is a sensor that measures the NOx concentration in the exhaust gas flowing into the SCR catalyst 40. Note that the NOx concentration acquisition unit 54 may estimate the NOx concentration in the exhaust gas from the combustion state (lean, stoichiometric, rich) of the internal combustion engine 92 instead of measurement by the sensor. The front end temperature acquisition unit 56 is a sensor that measures the temperature in the vicinity of the inlet (front end) of the SCR catalyst 40. The temperature acquisition unit 58 is a sensor that measures the bed temperature of the SCR catalyst 40.

  The flow rate acquisition unit 52, the exhaust temperature acquisition unit 53, and the NOx concentration acquisition unit 54 correspond to a “second acquisition unit” that acquires information on the exhaust gas flowing into the SCR catalyst 40. The front end temperature acquisition unit 56 and the temperature acquisition unit 58 correspond to a “first acquisition unit” that acquires information on the SCR catalyst 40.

  FIG. 4 is a flowchart showing a procedure of estimation processing by the estimation unit 110. The estimation process is a process for estimating the purification performance of the SCR catalyst 40, and is executed at an arbitrary timing. For example, the estimation process may be executed according to a request from a user of the exhaust purification system 1 or the catalyst state estimation device 10 or may be executed according to a request from another control unit in a vehicle on which the exhaust purification system 1 is mounted. Good. The estimation process shown in FIG. 4 is periodically executed.

  In the following description, among the items a1 to a3, all parameters except for the temperature at the front end of the SCR catalyst 40 are adopted as the input of the first model 121, and the items b1 to b1 are input as the inputs of the second model 122. The case where all the parameters described in b3 are employed and all the parameters described in items c1 to c3 are employed as the input of the third model 123 is illustrated. In the following description, Δt represents a unit time in the catalyst state estimation device 10 (for example, a calculation cycle of the CPU 11 and a sampling cycle in each of the acquisition units 52 to 58 described above), and time t = kΔt represents the current time, t = (k + 1) Δt means one time later (after one unit time), and time t = (k−1) Δt means one time before (one unit time before). k is an integer.

  In step S10, the estimation unit 110 determines whether a start condition for the estimation process is satisfied. Specifically, for example, the estimation unit 110, when the temperature acquisition unit 58 is normal, the urea pump unit 62 and the urea nozzle 64 are normal, and the NOx concentration acquisition unit 54 is in an active state, It can be determined that the estimation processing start condition is satisfied. When the estimation process start condition is satisfied (step S10: YES), the estimation unit 110 shifts the process to step S12. When the estimation process start condition is not satisfied (step S10: NO), the estimation unit 110 ends the process.

In step S <b> 12, the estimation unit 110 acquires the current flow rate Q [k] in the exhaust pipe 30 from the flow rate acquisition unit 52 (time t = kΔt). Estimation unit 110 in step S14 calculates the amount of NOx NOx_SCR _in flowing into the SCR catalyst 40 of the current (time t = k.DELTA.t). Specifically, the estimation unit 110 acquires the current NOx concentration [k] in the exhaust gas flowing into the SCR catalyst 40 from the NOx concentration acquisition unit 54. Next, the estimation unit 110 calculates the NOx amount NOx_SCR _in [k] from the acquired NOx concentration [k] in the exhaust and the flow rate Q [k] of the exhaust acquired in step S12.

In step S <b> 16, the estimation unit 110 acquires the current (time t = kΔt) bed temperature T [k] of the SCR catalyst 40 from the temperature acquisition unit 58. In step S <b> 18, the estimation unit 110 flows into the SCR catalyst 40 at the current time (time t = kΔt) from the injection amount of urea water injected from the urea nozzle 64 using a prepared formula, map, or the like. NH ₃ inflow amount NH ₃ —SCR _in [k] is calculated. In step S18, the estimation unit 110 also functions as a “third acquisition unit” that acquires information on additives.

In step S20, the estimation unit 110 acquires the current temperature T_SCR _in [k] of the exhaust gas flowing into the SCR catalyst 40 from the exhaust temperature acquisition unit 53 (time t = kΔt). Note that in step S20, the estimation unit 110 may acquire the temperature of the SCR catalyst 40 in the vicinity of the inlet of the SCR catalyst 40 (front end) instead of the temperature of the exhaust gas and set it as T_SCR _in [k].

  In step S <b> 22, the estimation unit 110 calculates the time differential value DiffT [k] of the bed temperature T of the SCR catalyst 40 at the current time (time t = kΔt) by the following mathematical formula 4. In Equation 4, Δt represents the sampling period of the temperature acquisition unit 58, T [k] represents the current bed temperature T of the SCR catalyst 40 acquired in step S16, and T [k−1] is acquired in step S16. Represents the bed temperature T of the SCR catalyst 40 one hour before. In step S22, the estimation unit 110 also functions as a “first acquisition unit” that acquires catalyst information.

FIG. 5 is a diagram illustrating step S24 of the estimation process. In FIG. 5, the horizontal axis represents the bed temperature of the SCR catalyst 40, and the vertical axis represents the saturated adsorption amount of NH ₃ in the SCR catalyst 40. As shown in the figure, the saturated adsorption amount of NH ₃ in the SCR catalyst 40 has a characteristic NP that decreases as the bed temperature increases. The relational expression representing the characteristic NP is obtained in advance through experiments or the like and stored in the storage unit 12.

In step S24 of FIG. 4, the estimation unit 110 calculates the ratio of the actual adsorption amount to the saturated adsorption amount of NH ₃ in the SCR catalyst 40. Specifically, the estimating unit 110 applies the current bed temperature T [k] of the SCR catalyst 40 acquired in step S16 to the relational expression representing the characteristic NP, and the current (time t = kΔt) SCR catalyst. The saturated adsorption amount StrtNH ₃ Ad [k] of 40 NH ₃ is obtained (FIG. 5). Next, the obtained saturated adsorption amount StrtNH ₃ Ad [k], using the current actual adsorption amount NH ₃ Ad of NH ₃ in the SCR catalyst 40 (time t = kΔt) [k], the real for saturated adsorption amount The adsorption amount ratio r [k] is calculated.

For the actual adsorption amount NH ₃ Ad [k] of NH ₃ of the SCR catalyst 40 at the current time (time t = kΔt), the estimation unit 110 calculates the estimated value obtained in the previously executed estimation process (FIG. 4). It can be used as an adsorption amount NH ₃ Ad [k]. For example, when there is no previous estimated value such as when the estimation process is executed for the first time, the estimating unit 110 can use a predetermined default value as the actual adsorption amount NH ₃ Ad [k]. Although a default value can be set arbitrarily, for example, 0 or a saturated adsorption amount can be used. In step S22, the estimation unit 110 also functions as a “first acquisition unit” that acquires catalyst information.

In step S26, the estimation unit 110 calculates an estimated value of the NOx purification rate of the SCR catalyst 40. Specifically, the estimation unit 110 adds the following values of the current (time t = kΔt) obtained in steps S12 to S20 to the input variable vector U of the first model 121 (FIG. 2) for estimating the NOx purification rate. Each value is set and the output variable Z is obtained. The estimation unit 110 sets the output variable Z obtained from the first model 121 as the NOx purification rate NOxConv [k] of the SCR catalyst 40 at the current time (time t = kΔt).
(A1) Catalyst information:
SCR catalyst 40 temperature (bed temperature) T [k] (step S16)
-Adsorption amount of NH ₃ adsorbed on SCR catalyst 40 one hour before (actual adsorption amount) NH ₃ Ad [k] (result of previous estimation process)
(A2) Exhaust information:
The temperature T_SCR _in [k] of the exhaust from the internal combustion engine 92 (step S20)
Exhaust flow rate Q [k] (step S12)
-NOx amount contained in exhaust gas NOx_SCR _in [k] (step S14)
(A3) Information on additives:
· Inflow NH of NH ₃ flowing into the SCR catalyst 40 ₃ _SCR _in [k] (Step S18)

  In this way, in step S26, the estimation unit 110 estimates the NOx purification rate NOxConv [k] as the purification performance of the SCR catalyst 40 using the first model 121. A machine learning model such as the first model 121 can be classified and regressed without complex function approximation as long as there is a causal relationship between input and output variables (for example, the input variable vector U and the output variable Z). (For example, when a phenomenon that is difficult to specify with a physical formula is included, when multiple physical formulas that describe the phenomenon are required, the influence of many factors is assumed, and input to the physical formula increases. ), An output (estimation result) can be obtained with a low calculation load. According to step S26, since the estimation unit 110 estimates the NOx purification rate NOxConv [k] using the first model 121 configured by the machine learning model, the estimation of the NOx purification rate influenced by many factors is performed. Can be implemented at low speed and with low computational load. Further, by using the fully learned machine learning model as the first model 121, the estimation unit 110 can estimate the NOx purification rate with high accuracy.

In step S <b> 28, the estimation unit 110 calculates an estimated value of the outflow amount of NH ₃ flowing out from the SCR catalyst 40. Specifically, the estimation unit 110 adds the following values of the current (time t = kΔt) obtained in steps S12 to S24 to the input variable vector U of the second model 122 for estimating the outflow amount of NH _3. To obtain the output variable Z. Estimation unit 110, the output variable Z obtained from the second model 122, a current (time t = k.DELTA.t) outflow of NH ₃ NH ₃ flowing out of the SCR catalyst 40 of _SCR _out [k].
(B1) Catalyst information:
SCR catalyst 40 temperature (bed temperature) T [k] (step S16)
-Adsorption amount of NH ₃ adsorbed on SCR catalyst 40 one hour before (actual adsorption amount) NH ₃ Ad [k] (result of previous estimation process)
-Time differential value DiffT [k] of the temperature of the SCR catalyst 40 (step S22)
Ratio r [k] of the adsorption amount (actual adsorption amount) to the saturated adsorption amount of NH ₃ in the SCR catalyst 40 (step S24)
(B2) Exhaust information:
The temperature T_SCR _in [k] of the exhaust from the internal combustion engine 92 (step S20)
Exhaust flow rate Q [k] (step S12)
-NOx amount contained in exhaust gas NOx_SCR _in [k] (step S14)
(B3) Additive information:
· Inflow NH of NH ₃ flowing into the SCR catalyst 40 ₃ _SCR _in [k] (Step S18)

In this way, in step S28, the estimation unit 110 estimates the NH ₃ outflow amount NH ₃ _SCR _out [k] as the purification performance of the SCR catalyst 40 using the second model 122. Like the first model 121 described above, the second model 122 is configured by a machine learning model, so that it is possible to estimate the outflow amount of NH ₃ affected by more factors at a low computational load and at high speed. . Further, by using the fully learned machine learning model as the second model 122, the estimation unit 110 can estimate the outflow amount with high accuracy.

In step S30, the estimation unit 110 calculates an estimated value of the amount of the additive (NH ₃ ) that does not contribute to the NOx purification reaction in the SCR catalyst 40. Specifically, the estimation unit 110 obtains the current value (time t = kΔt) obtained in steps S16 to S20 as the input variable vector U of the third model 123 for estimating the amount of NH ₃ that does not contribute to the NOx purification reaction. The following values are set to obtain an output variable Z. Estimation unit 110, the output variable Z obtained from the third model 123, a current (time t = k.DELTA.t), the amount NH ₃ _Thrmlytc of NH ₃ which does not contribute to purification reaction of NOx [k].
(C1) Catalyst information:
SCR catalyst 40 temperature (bed temperature) T [k] (step S16)
(C2) Exhaust information:
The temperature T_SCR _in [k] of the exhaust from the internal combustion engine 92 (step S20)
(C3) Additive information:
· Inflow NH of NH ₃ flowing into the SCR catalyst 40 ₃ _SCR _in [k] (Step S18)

In this way, in step S30, the estimation unit 110 estimates the amount NH ₃ —thrmlytc [k] of NH ₃ that does not contribute to the NOx purification reaction as the purification performance of the SCR catalyst 40 using the third model 123. A physical model such as the third model 123 is a law that is based on a similarity rule based on an actual physical law. For this reason, the output (estimation result) obtained by the third model 123 (physical model) satisfies the physical law. On the other hand, since the first model 121 and the second model 122 (machine learning model) are models constructed as a result of learning a huge amount of data, an output that satisfies the physical law (estimation result) may not be obtained. According to step S30, the estimation unit 110 estimates the amount of NH ₃ (additive) that does not contribute to the NOx purification reaction in the SCR catalyst 40 using the third model 123 configured by a physical model. Therefore, the amount of NH ₃ can be estimated with high accuracy according to physical laws.

In step S <b> 32, the estimation unit 110 calculates an estimated value of the adsorption amount of NH ₃ adsorbed on the SCR catalyst 40 after one time (time t = (k + 1) Δt) by the following Equation 5. Formula 5 functions as “catalyst state estimation model 120”.

Each value of Formula 5 is the purification performance (state) of the SCR catalyst 40 at the current time (time t = kΔt) estimated or calculated by the estimation unit 110 in steps S12 to S30. Specifically, NOxConv [k] is the NOx purification rate of the SCR catalyst 40 estimated by the first model 121. NH ₃ _SCR _out [k] is the amount of NH ₃ flowing out from the SCR catalyst 40 estimated by the second model 122. NH ₃ —thrmlytc [k] is the amount of NH ₃ that does not contribute to the NOx purification reaction estimated by the third model 123. NH ₃ Ad [k] is the adsorption amount (actual adsorption amount) of NH ₃ adsorbed on the SCR catalyst 40 obtained as a result of the previous estimation process. NH ₃ _SCR _in [k] is the amount of NH ₃ flowing into the SCR catalyst 40. NOx_SCR _in [k] is the amount of NOx contained in the exhaust gas flowing into the SCR catalyst 40.

After step S32 is completed, the estimation unit 110 outputs the NH ₃ adsorption amount NH ₃ Ad [k + 1] estimated in step S32. The output can be carried out in an arbitrary manner. For example, the output may be displayed on a display unit (not shown) included in the catalyst state estimation device 10 or may be recorded in an estimation history in the storage unit 12, and the exhaust purification system 1 is mounted. You may transmit a signal to the other control part in a vehicle. Further, the estimating unit 110, together with the amount of adsorption NH ₃ Ad [k + 1] of the NH ₃ estimated in step S32, or adsorption of NH NH _{3 3} instead of Ad [k + 1], by at least one of steps S12~S30 Obtained, calculated, and estimated results may be output. Thereafter, the estimation unit 110 ends the process.

As described above, according to the catalyst state estimation device 10 of the first embodiment, the estimation unit 110 uses the catalyst state estimation model 120 to adsorb NH ₃ (additive) adsorbed on the SCR catalyst 40. Estimate the quantity NH ₃ Ad [k + 1]. As is clear from Equation 5, the catalyst state estimation model 120 is an estimation result (NOx purification rate, NH ₃ flowing out from the SCR catalyst 40) of the first model 121 and the second model 122 which are machine learning models (NN models). The amount of NH ₃ that does not contribute to the NOx purification reaction in the SCR catalyst 40 in combination with the physical model, and one hour later (time t = (K + 1) Δt) is a model for estimating the adsorption amount NH ₃ Ad [k + 1] of NH ₃ (additive) adsorbed on the SCR catalyst 40. For this reason, according to the present embodiment, the estimation unit 110 determines the amount of adsorption NH ₃ Ad [k + 1] affected by a number of factors (for example, the factors listed in the items a1 to a3, b1 to b3, c1 to c3). The estimation can be performed with high accuracy and high speed while satisfying the physical laws.

Further, the adsorption amount of NH ₃ (additive) adsorbed on the SCR catalyst 40 varies under the influence of the adsorption amount NH ₃ Ad [k] at the previous time (time t = kΔt) (in other words, And fluctuate under the influence of time history). The estimation unit 110 according to the first embodiment performs the current estimation results NOxConv [k], NH ₃ _SCR _out [k], and NH ₃ _thrmlytc [k] based on the first model 121, the second model 122, and the third model 123. Is used to estimate the adsorption amount NH ₃ Ad [k + 1] after one time (time t = (k + 1) Δt). Therefore, the estimation unit 110 can estimate the purification performance of the SCR catalyst 40 at the next time with high accuracy based on the purification performance of the SCR catalyst 40 at the previous time (time t = kΔt).

Further, the purification performance of the SCR catalyst 40 varies under the influence of catalyst information (for example, the temperature of the SCR catalyst 40, the amount of NH ₃ adsorbed on the SCR catalyst 40). According to the catalyst state estimation device 10 of the first embodiment, the estimation unit 110 uses the first model 121 (item a1) and the second model 122 (item) as information on this catalyst that affects the purification performance of the SCR catalyst 40. It is applied to the catalyst state estimation model 120 by adopting it as a parameter of the vector U that is an input variable of b1) and the third model 123 (item a3). As a result, the estimation unit 110 can accurately estimate the purification performance of the SCR catalyst 40, and improve the estimation accuracy in the technique for estimating the purification performance of the SCR catalyst 40 that purifies the exhaust gas of the internal combustion engine 92. Can do.

Further, the purification performance of the SCR catalyst 40 varies in addition to the information on the catalyst, and is also affected by the information on the exhaust gas flowing into the catalyst (for example, the exhaust gas temperature, the exhaust gas flow rate, and the amount of NOx in the exhaust gas). . According to the catalyst state estimation device 10 of the first embodiment, the estimation unit 110 has information on these catalysts (items a1, b1, and c1) that affect the purification performance of the SCR catalyst 40, and information on the exhaust gas flowing into the catalyst. By applying both (items a2, b2, and c2) to the catalyst state estimation model 120, the purification performance of the SCR catalyst 40 can be estimated with higher accuracy. In addition to the catalyst information (items a1, b1, c1), the estimation unit 110 further applies additive information (for example, the amount of NH ₃ , items a3, b3, c3) to the catalyst state estimation model 120. By doing so, the purification performance of the SCR catalyst 40 is estimated. For this reason, for example, the purification performance of a catalyst that purifies harmful substances using an additive, such as the SCR catalyst 40 exemplified in the present embodiment, or other storage reduction catalyst (NSR catalyst: NOx Storage Reduction catalyst) is estimated. In some cases, the estimation accuracy can be further improved.

  Furthermore, if the data acquired during the transient operation of the internal combustion engine 92 is used as the teacher data when generating the first model 121 and the second model 122, the estimation unit 110 causes the SCR catalyst during the normal operation of the internal combustion engine 92. In addition to the purification performance of 40, the purification performance of the SCR catalyst 40 during transient operation of the internal combustion engine 92 can also be estimated.

Second Embodiment
FIG. 6 is a block diagram of an exhaust purification system 1a according to the second embodiment. In the first embodiment shown in FIG. 1, the estimation unit 110 uses the estimation results of the first to third models 121 to 123 as the purification performance of the SCR catalyst 40 one time later (time t = (k + 1) Δt The amount of NH ₃ adsorbed on the SCR catalyst 40 in FIG. However, the estimation unit 110a of the second embodiment uses the first model 121 to estimate the current (time t = kΔt) NOx purification rate NOxConv [k] of the SCR catalyst 40 as the purification performance of the SCR catalyst 40. To do.

  In the storage unit 12 of the second embodiment, a catalyst state estimation model 120a including only the first model 121 described in FIG. 2 is stored in advance. The estimation unit 110a uses the first model 121 of the catalyst state estimation model 120a to execute steps S10 to S24 and step S26 in the estimation process described in FIG. After execution of step S26, the estimation unit 110a outputs the NOx purification rate NOxConv [k] of the SCR catalyst 40 estimated in step S26 as the purification performance of the SCR catalyst 40. Thus, also in 2nd Embodiment, there can exist an effect similar to 1st Embodiment mentioned above.

  Note that the exhaust purification system 1 a may include the second model 122 instead of the first model 121 and / or together with the first model 121. In this case, the estimation unit 110a executes step S28 instead of step S26 and / or step S26, and outputs the estimation result as the purification performance of the SCR catalyst 40. Further, the exhaust purification system 1 a may include a third model 123 instead of the first model 121 and / or together with the first model 121. In this case, the estimation unit 110a executes step S30 instead of step S26 and / or together with step S26, and outputs the estimation result as the purification performance of the SCR catalyst 40. Further, the exhaust purification system 1a may include the first model 121, the second model 122, and the third model 123 without including the above-described mathematical formula 5. In this case, the estimation unit 110a executes steps S26 to S30 and outputs the estimation result as the purification performance of the SCR catalyst 40. Even if it does in this way, there can exist an effect similar to 1st Embodiment mentioned above.

<Third Embodiment>
FIG. 7 is a block diagram of an exhaust purification system 1b in the third embodiment. In the first embodiment shown in FIG. 1, the purification performance of the SCR catalyst is estimated. However, the exhaust purification system 1b of the third embodiment includes a storage reduction catalyst (NSR catalyst) instead of the SCR catalyst, and estimates the purification performance of the NSR catalyst.

The exhaust purification system 1b further includes a rich spike control unit 93, and includes a catalyst state estimation device 10b instead of the catalyst state estimation device 10 and an exhaust purification device 20b instead of the exhaust purification device 20. The rich spike control unit 93 is a control unit that generates rich spikes (extremely short fuel rich combustion) by controlling the injection of air and fuel to the internal combustion engine 92, and is implemented by, for example, an ECU. The rich spike control unit 93 causes the NSR catalyst 70 to store the carbon monoxide (CO), hydrogen (H ₂ ), and hydrocarbon (HC) supplied from the internal combustion engine 92 as a reducing agent by generating a rich spike. NOx that has been reduced is reduced and removed.

  The exhaust purification device 20b includes an NSR catalyst 70 instead of the SCR catalyst 40, and does not include the urea pump unit 62 and the urea nozzle 64. The NSR catalyst 70 purifies NOx in the exhaust by storing (storing) NOx contained in the exhaust into the storage material. The NSR catalyst 70 corresponds to a “catalyst”.

  The catalyst state estimation device 10b includes a front end temperature acquisition unit 56 instead of the front end temperature acquisition unit 56, a temperature acquisition unit 78 instead of the temperature acquisition unit 58, an estimation unit 110b instead of the estimation unit 110, and a catalyst state estimation model. Instead of 120, a catalyst state estimation model 120b is provided. The front end temperature acquisition unit 76 is a sensor that measures the temperature in the vicinity of the inlet (front end) of the NSR catalyst 70. The temperature acquisition unit 78 is a sensor that measures the bed temperature of the NSR catalyst 70. The front end temperature acquisition unit 76 and the temperature acquisition unit 78 correspond to a “first acquisition unit” that acquires information on the NSR catalyst 70.

Similar to the catalyst state estimation model 120 of the first embodiment, the catalyst state estimation model 120b includes a machine learning model, and includes a first model that outputs the NOx purification rate in the NSR catalyst 70 and a machine learning model. The second model that outputs CO, H ₂ , and HC outflow from the NSR catalyst 70 and the physical model and outputs the amounts of CO, H ₂ , and HC that do not contribute to the NOx purification reaction. And a third model. The estimation unit 110b executes the estimation process of FIG. 2 using such a catalyst state estimation model 120b, and determines the NOx occlusion amount in the NSR catalyst 70 by a physical law (for example, mass conservation law, mass balance law, heat balance). It is calculated using the balance law and energy balance law. Note that the NOx occlusion amount in the NSR catalyst 70 varies under the influence of the occlusion amount at the previous time (in other words, under the influence of the time history, similarly to the NH ₃ adsorption amount of the first embodiment). fluctuate).

  If it does in this way, also in 3rd Embodiment using NSR catalyst 70, there can exist an effect similar to 1st Embodiment mentioned above.

<Fourth embodiment>
FIG. 8 is a block diagram of an exhaust purification system 1c in the fourth embodiment. In the first embodiment shown in FIG. 1, the purification performance of the SCR catalyst is estimated. However, the exhaust purification system 1c of the fourth embodiment includes a three-way catalyst instead of the SCR catalyst, and estimates the purification performance of the three-way catalyst.

The exhaust purification system 1 c includes a catalyst state estimation device 10 c instead of the catalyst state estimation device 10, and an exhaust purification device 20 c instead of the exhaust purification device 20. The exhaust purification device 20c includes a three-way catalyst 80 instead of the SCR catalyst 40, and does not include the urea pump unit 62 and the urea nozzle 64. The three-way catalyst 80 removes CO, HC, and NOx contained in the exhaust gas by oxidizing or reducing, respectively, and purifies these harmful substances. Note that the three-way catalyst 80 of the present embodiment has an oxygen storage capacity (OSC) and can store (store) oxygen (O ₂ ) in the exhaust gas flowing into the three-way catalyst 80. .

The catalyst state estimation device 10c includes an oxygen concentration acquisition unit 84 instead of the NOx concentration acquisition unit 54, a front end temperature acquisition unit 86 instead of the front end temperature acquisition unit 56, and a temperature acquisition unit 88 instead of the temperature acquisition unit 58. An estimation unit 110c is provided instead of the estimation unit 110, and a catalyst state estimation model 120c is provided instead of the catalyst state estimation model 120. The oxygen concentration acquisition unit 84 acquires the O ₂ concentration in the exhaust from the internal combustion engine 92. The oxygen concentration acquisition unit 84 may be realized, for example, by acquiring a measurement signal measured by an A / F sensor provided in the exhaust pipe 30, or by acquiring a measurement signal measured by an oxygen sensor. It may be realized. The oxygen concentration acquisition unit 84 may estimate the O ₂ concentration in the exhaust gas from the intake air amount signal to the internal combustion engine 92 or the fuel injection amount signal. The oxygen concentration acquisition unit 84 also functions as a “third acquisition unit” that acquires information on additives. The front end temperature acquisition unit 86 is a sensor that measures the temperature in the vicinity of the inlet (front end) of the three-way catalyst 80. The temperature acquisition unit 88 is a sensor that measures the bed temperature of the three-way catalyst 80. The front end temperature acquisition unit 86 and the temperature acquisition unit 88 correspond to a “first acquisition unit” that acquires information about the three-way catalyst 80.

Similar to the catalyst state estimation model 120 of the first embodiment, the catalyst state estimation model 120c is configured by a machine learning model, and outputs at least one purification rate of CO, HC, NOx in the three-way catalyst 80 as an output. The first model is composed of a machine learning model, and is composed of a second model that outputs an O ₂ outflow from the three-way catalyst 80 and a physical model, and does not contribute to the purification reaction of CO, HC, and NOx. And a third model that outputs the amount of O ₂ . The estimation unit 110c executes the estimation process of FIG. 2 using such a catalyst state estimation model 120c, and determines the O ₂ occlusion amount in the three-way catalyst 80 by using physical laws (for example, mass conservation law, mass balance law). , Heat balance law, energy balance law). Note that the O ₂ occlusion amount in the three-way catalyst 80 varies under the influence of the occlusion amount at the previous time as in the case of the NH ₃ adsorption amount in the first embodiment (in other words, the influence of the time history). And fluctuate.)

  If it does in this way, also in 4th Embodiment using the three-way catalyst 80, there can exist an effect similar to 1st Embodiment mentioned above.

<Fifth Embodiment>
FIG. 9 is a block diagram of an exhaust purification system 1d according to the fifth embodiment. In the first embodiment shown in FIG. 1, the purification performance of the SCR catalyst 40 is estimated using the single first and second models 121 and 122. However, in the exhaust purification system 1d of the fifth embodiment, the purification performance of the SCR catalyst 40 is estimated by using a plurality of first and second models 121d and 122d.

  The exhaust purification system 1d includes a catalyst state estimation device 10d instead of the catalyst state estimation device 10. The catalyst state estimation device 10d includes an estimation unit 110d instead of the estimation unit 110, and includes a catalyst state estimation model 120d instead of the catalyst state estimation model 120. The catalyst state estimation model 120d includes a plurality of first models 121d, a plurality of second models 122d, and a third model 123. Each first model 121d is created using teacher data acquired from a plurality of SCR catalysts 40 having different degrees of deterioration. Specifically, for example, the first model 121d (1) is created from teacher data acquired from the SCR catalyst 40 with a low degree of deterioration, and the first model 121d (2) has a medium degree of deterioration. Created from the teacher data acquired from the SCR catalyst 40, the first model 121d (3) is generated from the teacher data acquired from the SCR catalyst 40 having a high degree of deterioration. The degree of deterioration (deterioration degree) may be determined from the state of the SCR catalyst 40, the replacement timing of the SCR catalyst 40, etc., may be determined from the NOx concentration in the exhaust discharged from the SCR catalyst 40, and It may be determined from the travel distance or the like. Each second model 122d is created using teacher data acquired from a plurality of SCR catalysts 40 having different degrees of degradation, similarly to the first model 121d.

In step S26 of the estimation process (FIG. 2), the estimation unit 110d estimates the NOx purification rate of the SCR catalyst 40 using the first model 121d corresponding to the degree of deterioration of the SCR catalyst 40. In step S28, the outflow amount of NH ₃ flowing out of the SCR catalyst 40 is estimated using the second model 122d corresponding to the degree of deterioration of the SCR catalyst 40.

  If it does in this way, also in 5th Embodiment, there can exist an effect similar to 1st Embodiment mentioned above. Further, according to the fifth embodiment, the catalyst state estimation model 120d is created using the teacher data acquired from the plurality of SCR catalysts 40 having different degrees of deterioration, and the plurality of first models 121d and second models 122d. Therefore, the optimal first model 121d and second model 122d can be adopted in accordance with the degree of deterioration of the SCR catalyst 40. As a result, according to the fifth embodiment, the purification performance of the catalyst can be estimated with higher accuracy.

<Sixth Embodiment>
FIG. 10 is a block diagram of an exhaust purification system 1e in the sixth embodiment. In the first embodiment shown in FIG. 1, the purification performance of one SCR catalyst is estimated. However, in the exhaust purification system 1e of the sixth embodiment, a plurality of SCR catalysts are mounted, and the purification performance of the plurality of SCR catalysts as a whole is estimated. The exhaust purification system 1e includes a catalyst state estimation device 10e instead of the catalyst state estimation device 10, and includes an exhaust purification device 20e instead of the exhaust purification device 20.

The exhaust purification device 20e further includes a second SCR catalyst 41 disposed on the downstream side of the SCR catalyst 40. Hereinafter, the SCR catalyst 40 is also referred to as a first SCR catalyst 40 for distinction. The second SCR catalyst 41 is a selective reduction catalyst of the same type as the first SCR catalyst 40. In the present embodiment, the “same catalyst” means a catalyst having the same or similar exhaust purification mechanism in the catalyst. The second SCR catalyst 41 purifies the harmful substances (NOx) exhausted without being purified by the first SCR catalyst 40 by the additive (NH ₃ ) exhausted without being used by the first SCR catalyst 40. The exhaust purification device 20e may be configured to include a urea nozzle that supplies urea water individually from the urea pump unit 62 to the second SCR catalyst 41. Hereinafter, a plurality of the same type of catalysts (first and second SCR catalysts 40 and 41) mounted on the exhaust purification device 20e are collectively referred to as “catalyst group CG”. In the configuration shown in FIG. 10, the first SCR catalyst 40 corresponds to “the catalyst located at the most upstream position”, and the second SCR catalyst 41 corresponds to “the catalyst located at the most downstream position”.

  The catalyst state estimation device 10e further includes a second temperature acquisition unit 59 including a sensor that measures the bed temperature of the second SCR catalyst 41. Further, the catalyst state estimation device 10e includes an estimation unit 110e instead of the estimation unit 110 and a catalyst state estimation model 120e instead of the catalyst state estimation model 120. The estimation unit 110e is different from the first embodiment (FIG. 4) in the content of the estimation process. The catalyst state estimation model 120e includes a first model 121e, a second model 122e, and a third model 123e.

  The first model 121e is a model for estimating the total NOx purification amount in a plurality of the same type of catalysts (that is, the first and second SCR catalysts 40 and 41) mounted on the exhaust purification device 20e. It is. The first model 121e is configured by a machine learning model, specifically, an NN, as in the first embodiment. Parameters shown in the following items a11 to a13 can be adopted for each component of the vector U that is an input variable of the first model 121e.

(A11) Information on the catalyst: the temperature of the front end of the first SCR catalyst 40 located at the uppermost stream, the temperatures of the first and second SCR catalysts 40 and 41, and NH ₃ adsorbed by a plurality of catalysts one hour before Information on at least one of (a12) exhaust gas: the temperature of the exhaust gas flowing from the internal combustion engine 92 into the first SCR catalyst 40 located at the most upstream, the flow rate of the exhaust gas, and the most upstream Information on the amount of NOx contained in the exhaust gas flowing into the first SCR catalyst 40 and information on at least one of the (a13) additives: the inflow amount of NH ₃ flowing into the first SCR catalyst 40 positioned in the uppermost stream

  As in the first embodiment, it is necessary to input at least one or more of the catalyst information shown in the item a11 for the vector U, and the exhaust information and the additive information shown in the items a12 and a13. About information, at least 1 or more may be input and all may be abbreviate | omitted. Processing after each component of the vector U is input to the input layer is the same as in the first embodiment. The output variable Z of the first model 121e is an estimated value of the total NOx purification amount in a plurality of catalysts (that is, the first and second SCR catalysts 40 and 41) under various conditions represented by the input variable (vector U). .

As in the first embodiment, the first model 121e causes the NN to learn the relationship between the input variable and the output variable so that the output variable Z matches the physical quantity to be estimated, and each value W _ij , θ _j , W _jk , θ _k are determined in advance. At the time of learning, in the present embodiment, the first SCR catalyst located in the uppermost stream with respect to a catalyst group including a plurality of the same type of catalysts (first and second SCR catalysts) arranged in the exemplary main flow path (exhaust pipe). It is preferable to use the teacher data acquired from the catalyst group by regarding the inlet of the catalyst group as the inlet of the catalyst group and the outlet of the second SCR catalyst located at the most downstream as the outlet of the catalyst group.

The second model 122e is outflow of the NH ₃ flowing out of the plurality of catalyst, in other words, a model for estimating the outflow of NH ₃ flowing out from the 2SCR catalyst 41 located downstream. Similar to the first embodiment, the second model 122e is configured by a machine learning model, specifically, an NN. Parameters shown in the following items b11 to b13 can be adopted for each component of the vector U that is an input variable of the second model 122e.

(B11) Information on the catalyst: the temperature of the front end of the first SCR catalyst 40 located on the most upstream side, the temperatures of the first and second SCR catalysts 40 and 41, and NH ₃ adsorbed by a plurality of catalysts one hour before The total adsorption amount (total actual adsorption amount), the time differential value of the temperature of the first SCR catalyst 40 and the time differential value of the temperature of the second SCR catalyst 41, and the sum of the saturated adsorption amounts of NH ₃ in a plurality of catalysts. Ratio of total adsorption amount (total actual adsorption amount) and at least one of them (b12) Exhaust information: temperature of exhaust gas flowing from the internal combustion engine 92 into the first SCR catalyst 40 located at the uppermost stream, Information on additive of at least one of the flow rate of exhaust gas and the amount of NOx contained in the exhaust gas flowing into the first SCR catalyst 40 located in the upstream (b13): the first SCR catalyst 40 located in the upstream Inflowing N H ₃ inflow

As in the first embodiment, it is necessary to input at least one or more of the catalyst information shown in the item b11 for the vector U, and the exhaust information and the additive information shown in the items b12 and b13. About information, at least 1 or more may be input and all may be abbreviate | omitted. The output variable Z of the second model 122e is the amount of NH ₃ flowing out from the plurality of catalysts lost under the conditions represented by the input variable (vector U), in other words, the second SCR catalyst located on the most downstream side. This is an estimated value of the amount of NH ₃ efflux that flows out of 41 and is lost. Note that the second model 122e, like the first model 121e, has a plurality of catalysts of the same type (disposed in the main flow channel (exhaust pipe) for example) when the NN learns the relationship between the input variable and the output variable. For the catalyst group consisting of the first and second SCR catalysts), the inlet of the first SCR catalyst located at the uppermost stream is regarded as the inlet of the catalyst group, the outlet of the second SCR catalyst located at the most downstream is regarded as the outlet of the catalyst group, It is preferable to use teacher data acquired from the catalyst group.

The third model 123e is a model for estimating the amount of the additive (NH ₃ ) supplied to the first SCR catalyst 40 located in the most upstream stream that does not contribute to the NOx purification reaction in a plurality of catalysts. It is. The third model 123e is configured by a physical model, specifically, an Arrhenius equation, as in the first embodiment. Parameters shown in the following items c11 to c13 can be adopted for each component of the vector U that is an input variable of the third model 123e.

(C11) Information on the catalyst: Information on at least one of the temperature of the front end of the first SCR catalyst 40 located at the most upstream and each temperature of the first and second SCR catalysts 40, 41 (c12) Information on the exhaust gas: Internal combustion Temperature of exhaust gas flowing into first SCR catalyst 40 located upstream from engine 92 (c13) Information on additive: Inflow amount of NH ₃ flowing into first SCR catalyst 40 located upstream

Similarly to the first embodiment, it is necessary to input at least one or more of the catalyst information shown in the item c11 for the vector U. The exhaust information and the additive information shown in the items c12 and c13 are necessary. About information, at least 1 or more may be input and all may be abbreviate | omitted. The output variable Z of the third model 123e is an estimated value of the amount of NH ₃ that does not contribute to the NOx purification reaction in the plurality of catalysts under the conditions represented by the input variable (vector U).

  FIG. 11 is a flowchart illustrating a procedure of estimation processing performed by the estimation unit 110e according to the sixth embodiment. The estimation process of the sixth embodiment is a process for estimating the purification performance of the plurality of catalysts as a whole, and is executed at an arbitrary timing, as in the first embodiment shown in FIG. In the following description, parameters a11 to a13, b11 to b13, c11 to c13 used for the description, Δt, time t = kΔt, time t = (k + 1) Δt, time t = (k−1) Δt Each definition is the same as in the first embodiment. In the following description, only processing different from the first embodiment shown in FIG. 4 will be described.

Estimation unit 110e in step S14e, the at present (time t = k.DELTA.t), calculates the amount of NOx NOx_SCR _in which flows into the 1SCR catalyst 40 located most upstream. In step S <b> 16 e, the estimating unit 110 e acquires the current (temperature t = kΔt) bed temperature T <b> 1 [k] of the first SCR catalyst 40 from the first temperature acquisition unit 58. In addition, the estimation unit 110e acquires the bed temperature T2 [k] of the second SCR catalyst 41 at the current time (time t = kΔt) from the second temperature acquisition unit 59. Estimation unit 110e in step S18e from the injection amount of urea water injected from the urea nozzle 64, the current (time t = k.DELTA.t) in the inflow amount of NH ₃ NH ₃ flowing into the first 1SCR catalyst 40 located at the most upstream _SCR _in [k] is calculated.

In step S20e, the estimation unit 110e acquires, from the exhaust temperature acquisition unit 53, the temperature T1_SCR _in [k] of the exhaust gas flowing into the first SCR catalyst 40 located at the uppermost stream at the current time (time t = kΔt). In step S22e, the estimation unit 110e calculates the time differential value DiffTn [k] of the bed temperature Tn of each SCR catalyst (first and second SCR catalysts 40, 41) at the current time (time t = kΔt) according to the following Equation 6. Calculate each. The variable n in Equation 6 is a natural number for distinguishing each catalyst.

In step S24e, the estimation unit 110e calculates a ratio of the total actual adsorption amount to the total saturated adsorption amount of NH ₃ in the plurality of catalysts. Specifically, the estimating unit 110e applies the current bed temperature T1 [k] of the first SCR catalyst 40 acquired in step S16e to the relational expression representing the characteristic NP, and the current (time t = kΔt) The saturated adsorption amount of NH ₃ of the 1SCR catalyst 40 is obtained (FIG. 5). Similarly, the estimating unit 110e applies the current bed temperature T2 [k] of the second SCR catalyst 41 acquired in step S16e to the relational expression representing the characteristic NP, and the current (time t = kΔt) second SCR catalyst. The saturated adsorption amount of 41 NH ₃ is determined. The estimation unit 110e sets the sum of the obtained saturated adsorption amounts as a total StrtNH ₃ Ad [k] of the saturated adsorption amounts of NH ₃ in the plurality of catalysts. Next, the estimation unit 110e calculates the total saturated adsorption amount StrtNH ₃ Ad [k] obtained and the total NH ₃ Ad [k] of the actual NH ₃ adsorption amounts of the plurality of catalysts at the present time (time t = kΔt). Is used to calculate the ratio r [k] of the total actual adsorption amount to the total saturated adsorption amount. In addition, regarding the total NH ₃ Ad [k] of the actual adsorption amounts of NH ₃ of a plurality of catalysts, the estimated value obtained by the previously executed estimation process can be used, or a predetermined default value when there is no estimated value. The point that can be used is the same as in the first embodiment.

  In step S26e, the estimation unit 110e calculates an estimated value of the total NOx purification amount in the plurality of catalysts (that is, the first and second SCR catalysts 40, 41). Specifically, the estimation unit 110e sets the current values (time t = kΔt) obtained in steps S12 and S14e to S20e to the input variable vector U of the first model 121e, and obtains the output variable Z. The estimation unit 110e sets the output variable Z obtained from the first model 121e as the total NOxConv [k] of the NOx purification amounts of the plurality of catalysts at the present time (time t = kΔt).

In step S <b> 28 e, the estimation unit 110 e estimates the amount of NH ₃ flowing out and lost from the plurality of catalysts, in other words, the amount of NH ₃ flowing out and lost from the second SCR catalyst 41 located on the most downstream side. Is calculated. Specifically, the estimation unit 110e sets the current values (time t = kΔt) obtained in steps S12 and S14e to S24e to the input variable vector U of the second model 122e, and obtains the output variable Z. The estimation unit 110e sets the output variable Z obtained from the second model 122e as the outflow amount NH ₃ _SCR _out [k] of NH ₃ that flows out and is lost from a plurality of catalysts at the present time (time t = kΔt).

In step S <b> 30 e, the estimation unit 110 e calculates an estimated value of the amount of additive that does not contribute to the NOx purification reaction in the plurality of catalysts among the additives (NH ₃ ) supplied to the first SCR catalyst 40 located in the uppermost stream. calculate. Specifically, the estimation unit 110e sets the current values (time t = kΔt) obtained in steps S16e to S20e to the input variable vector U of the third model 123e, and obtains the output variable Z. Estimation unit 110e is the output variable Z obtained from the third model 123e, as the current (time t = k.DELTA.t), the amount of NH ₃ which does not contribute to the purification reaction of NOx in a plurality of catalytic NH ₃ _thrmlytc [k] .

In step S32e, the estimation unit 110e explains the estimated value of the total amount of NH ₃ adsorbed on the plurality of catalysts after one time (time t = (k + 1) Δt) in step S32 of the first embodiment. It is calculated by the following formula 5. Here, each value applied to Formula 5 is the purification performance (or state) of the plurality of catalysts at the present time (time t = kΔt) estimated or calculated by the estimation unit 110e in steps S12 and S14e to S30e, respectively. . After the step S32E is finished, the estimation unit 110e outputs the total NH ₃ Ad adsorption amount of NH ₃ estimated in step S32e [k + 1]. Details are the same as in the first embodiment.

  As described above, the catalyst state estimation device 10e according to the sixth embodiment can achieve the same effects as those of the first embodiment described above. Further, according to the catalyst state estimation device 10e of the sixth embodiment, when the plurality of catalysts (first and second SCR catalysts 40, 41) are provided in the main flow path (exhaust pipe 30), the estimation unit 110e is By applying the information of the first SCR catalyst 40 located on the most upstream side to the catalyst state estimation model 120e, it is possible to estimate the purification performance as a whole of the plurality of catalysts. That is, according to the catalyst state estimation device 10e of the sixth embodiment, the plurality of catalysts (first and second SCR catalysts 40 and 41) on the main flow path are regarded as one catalyst (that is, the catalyst group CG), and the catalyst group CG. As compared with the case of estimating the purification performance of each catalyst, the catalyst information acquisition unit (for example, the flow rate acquisition unit of exhaust gas flowing into the second SCR catalyst 41, the exhaust temperature acquisition unit) , The number of sensors constituting the NOx concentration acquisition unit, the front end temperature acquisition unit of the second SCR catalyst 41, etc.) and the number of catalyst state estimation models 120e prepared in advance can be reduced, and the calculation load on the catalyst state estimation device 10e can be reduced. Can be reduced. In addition, when the plurality of catalysts provided in the main flow path are the same type of catalyst (SCR catalyst in the example of the sixth embodiment), the estimation unit 110e regards these as one catalyst (catalyst group CG). Estimate purification performance. In the case of the same type of catalyst, there is no difference in the information items of the catalyst that influence the purification performance (for example, the temperature of the catalyst, the amount of additive adsorbed on the catalyst), and therefore the estimation unit 110e accurately improves the purification performance. Can be estimated.

  Further, according to the catalyst state estimation device 10e of the sixth embodiment, a plurality of catalysts arranged in the exemplary main flow path are regarded as one catalyst (catalyst group), using the teacher data acquired from the catalyst group. By learning, the first model 121e and / or the second model 122e of the catalyst state estimation model 120e can be constructed. For this reason, in the 1st model 121e and / or the 2nd model 122e of the catalyst state estimation model 120e, the influence of the main flow path (exhaust pipe etc.) between each catalyst which belongs to a catalyst group can be considered. By using the catalyst state estimation model 120e constructed in this way in the estimation process, the estimation unit 110e allows the inter-catalyst (the first SCR catalyst 40 and the second SCR catalyst 41 between the main flow path (exhaust pipe 30) to be connected. The purification performance as a whole of the plurality of catalysts can be estimated with high accuracy without the need for information on the exhaust pipe and the like.

  In the sixth embodiment, two SCR catalysts (first and second SCR catalysts 40 and 41) are illustrated as specific examples of the plurality of catalysts. However, the exhaust purification system 1e may include three or more SCR catalysts. In addition, the exhaust purification system 1e may include a plurality of NSR catalysts or a plurality of three-way catalysts. The estimation process of the sixth embodiment described in FIG. 11 can be applied to a plurality of the same type of catalysts.

<Seventh embodiment>
FIG. 12 is a block diagram of an exhaust purification system 1f according to the seventh embodiment. In the sixth embodiment shown in FIG. 10, the temperature (bed temperature) of each catalyst provided in the exhaust purification device 20e is respectively determined by the first and second temperature acquisition units 58 and 59 provided for each catalyst. I got it. However, in the exhaust purification system 1f of the seventh embodiment, the catalyst state estimation device 10f replaces the second temperature acquisition unit 59 with a catalyst other than the first SCR catalyst 40 located on the most upstream side (in the example of FIG. 12). And a temperature estimating unit 159 for estimating the temperature of the second SCR catalyst 41).

  The temperature estimation unit 159 calculates the temperature T2 of the second SCR catalyst 41 from the temperature T1 of the first SCR catalyst 40 acquired by the first temperature acquisition unit 58 using a previously prepared calculation formula, map, or the like. In addition to the temperature T1 of the first SCR catalyst 40 located at the uppermost stream, the temperature estimation unit further reacts with the temperature of the exhaust from the internal combustion engine 92, the flow rate of the exhaust, and NOx in the exhaust at the first SCR catalyst 40. The temperature T2 of the second SCR catalyst 41 may be calculated in consideration of the reaction heat generated by the above and other arbitrary parameters. Similarly, even when three or more catalysts are mounted on the exhaust purification system 1f, the temperature estimation unit 159 similarly calculates the temperature prepared in advance from the temperature T1 of the first SCR catalyst 40 located at the uppermost stream. The temperature Tn of another catalyst (n is a natural number for distinguishing each catalyst) can be calculated using an equation, a map, or the like. The estimation unit 110f acquires the temperature Tn of the other catalyst from the temperature estimation unit 159 in step S16e of the estimation process described in FIG.

  As described above, the catalyst state estimation device 10f according to the seventh embodiment can achieve the same effects as those of the first and sixth embodiments described above. Further, according to the catalyst state estimation device 10f of the seventh embodiment, the temperature estimation unit 159 determines the second SCR catalyst 41 provided in the main flow path (exhaust pipe 30) from the temperature of the first SCR catalyst 40 located at the most upstream. Therefore, an acquisition unit (such as a sensor) for acquiring the temperature of the second SCR catalyst 41 can be omitted.

<Modification of this embodiment>
The present invention is not limited to the above-described embodiment, and can be implemented in various modes without departing from the gist thereof. For example, the following modifications are possible.

[Modification 1]
In the said embodiment, an example of the structure of the exhaust gas purification system was shown. However, various modifications can be made to the configuration of the exhaust purification system. For example, the exhaust gas purification system of the exhaust gas purification system is equipped with a combination of a plurality of SCR catalysts, NSR catalysts, and three-way catalysts, and the catalyst state estimation device provides the purification performance of the plurality of catalysts. Each may be estimated. Further, the exhaust purification device is equipped with a particulate matter removal filter (DPF: Diesel Particulate Filter) for removing particulate matter (PM), and the catalyst state estimation device may estimate the purification performance in this DPF. .

  For example, the temperature acquisition part which acquires the bed temperature of a catalyst may be provided in the front (near entrance) or back (near exit) of a catalyst.

For example, at least one of the flow rate, the NOx concentration, the front end temperature, and the catalyst temperature that the flow rate acquisition unit, the NOx concentration acquisition unit, the front end temperature acquisition unit, and the temperature acquisition unit respectively acquire is replaced with the measured value by the sensor. Alternatively, the temperature estimated using the catalyst state estimation model may be substituted. Specifically, for example, the temperature of the catalyst is affected by the time history as in the above-described NH ₃ adsorption amount in the SCR catalyst, NOx occlusion amount in the NSR catalyst, and O ₂ occlusion amount in the three-way catalyst. . For this reason, a catalyst state estimation model that can estimate the temperature of the catalyst may be created separately, and the temperature of the catalyst may be estimated using the catalyst state estimation model.

[Modification 2]
In the said embodiment, an example of the estimation process in an estimation part was shown (FIG. 4, FIG. 11). However, the content of the estimation process can be variously modified. For example, in the estimation process shown in FIG. 4, the execution order of steps S12 to S24 may be changed, and the execution order of steps S26 to S30 may be changed. Similarly, in the estimation process shown in FIG. 11, the execution order of steps S12 to S24e may be changed, and the execution order of steps S26e to S30e may be changed.

For example, the estimation unit may execute the following steps S100 and S102 in addition to the steps described above. Steps S100 and S102 can be executed at an arbitrary timing.
Step S100: The estimation unit determines whether or not the bed temperature T [k] of the SCR catalyst is equal to or higher than a predetermined temperature threshold value at which the saturated adsorption amount of NH _{3 in} the SCR catalyst becomes zero. When the bed temperature T [k] is equal to or higher than a predetermined temperature threshold, the estimation unit estimates the NH ₃ adsorption amount NH adsorbed on the SCR catalyst after one time (time t = (k + 1) Δt). ₃ Reset Ad [k + 1] to 0.
Step S102: The estimation unit sufficiently injects urea water from the urea pump unit and the urea nozzle until the saturated adsorption amount of NH _{3 in} the SCR catalyst reaches 100 (saturated state). Thereafter, the estimation unit resets the estimated value NH ₃ Ad [k + 1] of the adsorption amount of NH ₃ adsorbed on the SCR catalyst to 100 (saturated adsorption amount) after one time (time t = (k + 1) Δt). To do.
According to such steps S100 and S102, when the estimation process shown in FIGS. 4 and 11 is repeated, errors are accumulated, and when the estimated value deviates from the actual value, it can be reset.

  In the sixth and seventh embodiments, the first model is a model that estimates the total NOx purification amount in the plurality of catalysts of the same type of catalyst (first and second SCR catalysts). However, the first model may be defined as a model for estimating the NOx purification rates in a plurality of catalysts, as in the first embodiment. In this case, the NOx purification rate in the plurality of catalysts is, for example, for a catalyst group consisting of a plurality of catalysts of the same type, the inlet of the catalyst located at the most upstream is regarded as the inlet of the catalyst group, and the outlet of the catalyst located at the most downstream is It can be defined by multiplying 100 by “(NOx amount (or NOx concentration) at the inlet−NOx amount (or NOx concentration) at the outlet) / NOx amount (or NOx concentration) at the inlet)”, which is regarded as the outlet of the catalyst group.

  As mentioned above, although this aspect was demonstrated based on embodiment and a modification, embodiment of an above-described aspect is for making an understanding of this aspect easy, and does not limit this aspect. This aspect can be changed and improved without departing from the spirit and scope of the claims, and equivalents are included in this aspect. Further, if the technical feature is not described as essential in the present specification, it can be deleted as appropriate.

DESCRIPTION OF SYMBOLS 1,1a-1f ... Exhaust gas purification system 10, 10a-10f ... Catalyst state estimation apparatus 11 ... CPU
DESCRIPTION OF SYMBOLS 12 ... Memory | storage part 20, 20b, 20e ... Exhaust gas purification device 30 ... Exhaust pipe 40 ... SCR catalyst, 1st SCR catalyst 41 ... 2nd SCR catalyst 52 ... Flow volume acquisition part 53 ... Exhaust temperature acquisition part 54 ... NOx density | concentration acquisition part 56 ... Front end Temperature acquisition unit 58 ... temperature acquisition unit, first temperature acquisition unit 59 ... second temperature acquisition unit 62 ... urea pump unit 64 ... urea nozzle 70 ... NSR catalyst 76 ... front end temperature acquisition unit 78 ... temperature acquisition unit 80 ... three-way catalyst 84 ... oxygen concentration acquisition unit 86 ... front end temperature acquisition unit 88 ... temperature acquisition unit 91 ... combustion state control unit 92 ... internal combustion engine 93 ... rich spike control unit 110, 110a-110f ... estimation unit 120, 120a-120e ... catalyst state estimation Model 121, 121e ... 1st model 122, 122e ... 2nd model 123, 123e ... 3rd model 159 ... Temperature estimation part

Claims (19)

A catalyst state estimation device comprising:
A first acquisition unit that is provided in a main flow path through which exhaust from the internal combustion engine circulates and acquires information on a catalyst that purifies a harmful substance in the exhaust;
A storage unit for preliminarily storing a catalyst state estimation model including at least one mathematical model;
An estimation unit that estimates the purification performance of the catalyst by applying the catalyst information acquired by the first acquisition unit to the catalyst state estimation model;
A catalyst state estimating device. The catalyst state estimation device according to claim 1, further comprising:
A second acquisition unit for acquiring information of the exhaust gas flowing into the catalyst;
In addition to the catalyst information, the estimation unit further applies the exhaust information acquired by the second acquisition unit to the catalyst state estimation model, thereby estimating the purification performance of the catalyst. Estimating device. The catalyst state estimation apparatus according to claim 1 or 2, further comprising:
A third acquisition unit for acquiring information of the additive supplied to the catalyst;
The estimation unit further applies the information on the additive acquired by the third acquisition unit to the catalyst state estimation model in addition to the information on the catalyst, thereby estimating the purification performance of the catalyst. State estimation device. The catalyst state estimation device according to any one of claims 1 to 3,
The catalyst state estimation model is:
Consists of machine learning models,
As the catalyst information, at least one of the temperature of the front end of the catalyst in the main flow path, the temperature of the catalyst, and the adsorption amount of the additive adsorbed on the catalyst one hour before is input. age,
Including a first model whose output is a purification rate of nitrogen oxides in the catalyst;
The said estimation part is a catalyst state estimation apparatus which estimates the purification rate of the said nitrogen oxide as a purification performance of the said catalyst using the said 1st model. The catalyst state estimation device according to any one of claims 1 to 4,
The catalyst state estimation model is:
Consists of machine learning models,
As the information of the catalyst, the temperature of the front end of the catalyst in the main flow path, the temperature of the catalyst, the adsorption amount of the additive adsorbed on the catalyst before one time, and the time differential value of the temperature of the catalyst And at least one of the ratio of the adsorption amount to the saturated adsorption amount of the additive in the catalyst, and
Including a second model whose output is an outflow amount of the additive flowing out from the catalyst,
The said estimation part is a catalyst state estimation apparatus which estimates the said outflow amount as a purification performance of the said catalyst using the said 2nd model. The catalyst state estimation device according to any one of claims 1 to 5,
The catalyst state estimation model is:
Composed of physical models,
As the information of the catalyst, at least one of the temperature of the front end of the catalyst in the main channel and the temperature of the catalyst is input,
A third model that outputs the amount of additive that does not contribute to the nitrogen oxide purification reaction in the catalyst;
The said estimation part is a catalyst state estimation apparatus which estimates the quantity of the additive which does not contribute to the purification | cleaning reaction of a nitrogen oxide in the said catalyst as said catalyst purification performance using the said 3rd model. The catalyst state estimation device according to claim 6 subordinate to claim 5 subordinate to claim 4, comprising:
The catalyst state estimation model is:
The present nitrogen oxide purification rate estimated by the first model;
The current runoff estimated by the second model;
The amount of additive that does not contribute to the nitrogen oxide purification reaction in the current catalyst estimated by the third model, and
It is a model for obtaining the amount of additive adsorbed on the catalyst at the next time using a physical law,
The said estimation part is a catalyst state estimation apparatus which estimates the said adsorption amount in the next time as the purification performance of the said catalyst using the said catalyst state estimation model. A catalyst state estimation device according to claim 7, comprising:
The first model and the second model further include:
As the exhaust information, at least one of the temperature of the exhaust, the flow rate of the exhaust, and the amount of nitrogen oxide contained in the exhaust is input,
As the information of the additive, the inflow amount of the additive flowing into the catalyst is input,
The third model further includes:
As the exhaust information, the temperature of the exhaust is input,
The catalyst state estimation apparatus which uses the inflow amount of the additive flowing into the catalyst as input as information on the additive. The catalyst state estimation apparatus according to claim 7 or 8,
The catalyst state estimation model is:
A plurality of the first models created using teacher data acquired from a plurality of the catalysts having different degrees of deterioration;
And a plurality of second models created using teacher data acquired from the plurality of catalysts having different degrees of deterioration. A method for estimating the state of a catalyst, comprising:
An information processing apparatus including a storage unit that stores in advance a catalyst state estimation model including at least one mathematical model,
A step of obtaining information on a catalyst that is provided in a main flow path through which exhaust from an internal combustion engine flows and purifies harmful substances in the exhaust; and
Applying the catalyst information obtained in the obtaining step to the catalyst state estimation model to estimate the purification performance of the catalyst;
A method comprising: A computer program,
An information processing apparatus including a storage unit that stores in advance a catalyst state estimation model including at least one mathematical model,
A function of obtaining information on a catalyst that is provided in a main flow path through which exhaust from an internal combustion engine flows and purifies harmful substances in the exhaust;
A function of estimating the purification performance of the catalyst by applying the catalyst information acquired in the acquiring step to the catalyst state estimation model;
A computer program that executes The catalyst state estimation device according to claim 1,
When a plurality of the same type of catalyst is provided in the main flow path,
The first acquisition unit acquires at least information on the catalyst located in the uppermost stream in the main flow path,
The said estimation part is a catalyst state estimation apparatus which estimates the purification performance as said whole several catalyst by applying the information of the said catalyst acquired by the said 1st acquisition part to the said catalyst state estimation model. The catalyst state estimation device according to claim 12,
The catalyst state estimation model is:
Consists of machine learning models,
As the information of the catalyst, the temperature of the front end of the catalyst located in the most upstream, the temperature of each of the catalysts, and the total adsorbed amount of the additive adsorbed on the plurality of catalysts one hour before, At least one of
Including a first model whose output is the sum of the purification amounts of nitrogen oxides in the plurality of catalysts,
The said estimation part is a catalyst state estimation apparatus which estimates the purification amount of the said nitrogen oxide as the said purification performance using the said 1st model. The catalyst state estimation apparatus according to claim 12 or claim 13,
The catalyst state estimation model is:
Consists of machine learning models,
As the information of the catalyst, the temperature of the front end of the catalyst located at the most upstream, the temperature of each catalyst, the total amount of adsorbed adsorbed by the plurality of catalysts one hour before, At least one of a time differential value of the temperature of the catalyst and a ratio of the total adsorption amount to the total saturated adsorption amount of the additive in the plurality of catalysts is input,
Including a second model that outputs an outflow amount of the additive flowing out from the catalyst located on the most downstream side in the main flow path;
The said estimation part is a catalyst state estimation apparatus which estimates the said outflow amount as said purification performance using the said 2nd model. The catalyst state estimation device according to any one of claims 12 to 14,
The catalyst state estimation model is:
Composed of physical models,
As the information of the catalyst, at least one of the temperature of the front end of the catalyst located in the uppermost stream and the temperature of each catalyst is input,
A third model that outputs an amount of an additive that does not contribute to a nitrogen oxide purification reaction among the plurality of catalysts among the additives supplied to the catalyst located at the uppermost stream;
The said estimation part is a catalyst state estimation apparatus which estimates the quantity of the additive which does not contribute to the purification reaction of the said nitrogen oxide as said purification performance using the said 3rd model. The catalyst state estimation apparatus according to claim 15 dependent on claim 14 dependent on claim 13, comprising:
The catalyst state estimation model is:
The total amount of nitrogen oxide purification currently estimated by the first model;
The current runoff estimated by the second model;
The amount of additive that does not contribute to the present nitrogen oxide purification reaction estimated by the third model, and
It is a model for obtaining the total amount of additives adsorbed on the plurality of catalysts at the next time using a physical law,
The said estimation part is a catalyst state estimation apparatus which estimates the total of the said adsorption amount in the next time as said purification performance using the said catalyst state estimation model. The catalyst state estimation apparatus according to claim 16, further comprising:
A second acquisition unit for acquiring information on exhaust flowing into the catalyst located at the uppermost stream;
A third acquisition unit for acquiring information on the additive to be supplied to the catalyst located in the uppermost stream,
In the first model and the second model,
As the exhaust information, at least one of the temperature of the exhaust, the flow rate of the exhaust, and the amount of nitrogen oxide contained in the exhaust is input,
As the information on the additive, the inflow amount of the additive flowing into the catalyst located in the uppermost stream is input,
The third model further includes:
As the exhaust information, the temperature of the exhaust is input,
The catalyst state estimation apparatus which uses the inflow amount of the additive flowing into the catalyst located at the uppermost stream as input as information on the additive. The catalyst state estimation device according to claim 16 or 17,
The catalyst state estimation model is:
For the catalyst group consisting of a plurality of the same type of catalysts arranged in the exemplary main channel, the inlet of the catalyst located in the uppermost stream in the exemplary main channel is regarded as the inlet of the catalyst group, and the exemplary Of the first model and the second model created using teacher data acquired from the catalyst group, the outlet of the catalyst located at the most downstream in the main channel is regarded as the outlet of the catalyst group A catalyst state estimation device including at least one of the models. The catalyst state estimation device according to any one of claims 12 to 18, further comprising:
A catalyst state estimation device comprising a temperature estimation unit that estimates the temperature of the other catalyst provided in the main flow path from the temperature of the catalyst located in the uppermost stream.

Images