JP2020095451A - Parameter estimation device and method - Google Patents

Abstract

To improve accuracy of estimating an unknown parameter related to a facility.SOLUTION: A parameter estimation device 1 estimating a parameter related to a facility 3 acquires an observed value from a sensor 2 measuring a physical amount related to the facility, and acquires a facility model formula to calculate an observed value based on a target parameter (41), calculates likelihood of an estimated observed value, which is obtained by substituting the estimated value of the target parameter into the facility model formula, to the observed value by a first likelihood function, calculates likelihood of the estimated value of the target parameter based on the first likelihood function and a second likelihood function for obtaining probability that the estimated value of the target parameter is within a predetermined range (42), estimates the target parameter based on the calculated likelihood (43), and evaluates the estimated target parameter according to a predetermined evaluation criterion (44).SELECTED DRAWING: Figure 1

Description

The present invention relates to a parameter estimation device and method.

BEMS (Building Energy Management System) is becoming widespread in buildings and commercial facilities in order to use energy efficiently. BEMS manages utility equipment such as heat source equipment, air conditioning equipment, and electric power equipment of a building. BEMS carries out maintenance of facilities and optimization of facilities by attaching sensors and communication devices to utility facilities, measuring, visualizing and managing them.

Furthermore, in recent years, the importance of CEMS (Community Energy Management System) is increasing. CEMS manages several BEMS and manages the energy of the whole area. The main function of BEMS and CEMS is visualization of equipment status. The equipment state is a value representing not only power consumption but also an operation state of utility equipment managed by BEMS and CEMS, such as efficiency, flow rate, temperature, and pressure. However, it is expensive and impractical to provide various kinds of sensors in the utility equipment in order to know all the states of the equipment.

Therefore, it is conceivable to estimate the value of another state based on the sensor data for detecting one state. That is, by estimating the unknown state by using the measurement data under the control of BEMS or CEMS, the equipment state can be managed without attaching a sensor for each state of the detection target.

By the way, in order to estimate the state of utility equipment, it is necessary to estimate parameters. However, when estimating an unknown parameter, the estimated value may converge to a value that is not a physical value. As a technique for improving the estimation accuracy of unknown parameters, the technique of Patent Document 1 has been proposed.

In patent document 1, "a parameter for estimating a model parameter when a working model of a plurality of electric devices is modeled by a probabilistic model using a total value of power consumption by a plurality of electric devices connected to a distribution board. In the parameter estimation step, including the estimation step, the likelihood calculated by the likelihood function is maximized based on the characteristics of the power data that can be predetermined as prior knowledge from the operation tendency of each of the plurality of electric devices. Such a model parameter is estimated, the probabilistic model is a Factorial HMM (Hidden Markov Model), and the likelihood is the actually measured consumption of the total power consumption pattern modeled by the Factorial HMM. It is a value indicating the certainty with respect to the total value of electric power."

JP, 2017-49221, A

However, Patent Document 1 is limited to the case of using a model other than the Physical HMM, and it is difficult to estimate the unknown parameter using another model.

By the way, in order to easily estimate many parameters of the utility equipment model, it is necessary to have a large number of types of observation values. The observed value is a value actually measured by a measuring device such as a sensor. However, in order to obtain many kinds of observation values, it is necessary to attach a measuring device such as a sensor to the utility equipment as described above, which causes a cost problem.

Furthermore, when the number of types of observed values is small, it becomes a so-called inferior problem, and thus the estimated values may not converge. Therefore, there is a need for a method of obtaining an estimated value close to the true value by using a limited number of observation values.

The present invention has been made in view of the above problems, and an object thereof is to provide a parameter estimation device and method capable of improving the estimation accuracy of an unknown parameter regarding equipment.

In order to solve the above-mentioned problems, a parameter estimation device according to the present invention is a device for estimating a parameter related to equipment, and acquires an observed value from a sensor that measures a physical quantity related to the equipment, and calculates the observed value based on a target parameter. Obtain the equipment model formula and calculate the likelihood of the estimated observation value obtained by substituting the estimated value of the target parameter into the equipment model formula by the first likelihood function, and the estimated value of the target parameter is prepared in advance. The likelihood of the estimated value of the target parameter is calculated based on the second likelihood function and the first likelihood function for obtaining the probability of belonging to the predetermined range, and the target parameter is estimated based on the calculated likelihood. , The estimated target parameter is evaluated according to a predetermined evaluation criterion.

According to the present invention, the target parameter can be estimated and evaluated from the likelihood of the estimated value of the target parameter calculated based on the first likelihood function and the second likelihood function.

It is a block diagram of the whole system including a parameter estimation device. It is a block diagram of a parameter estimation apparatus. It is a flowchart which shows a parameter estimation process. It is a flowchart which shows the process which determines an equipment model type|formula. It is a flow chart which shows the processing which sets up the proper range of a target parameter. It is explanatory drawing which shows the example of setting of the suitable range of a target parameter. It is explanatory drawing which shows the other example of setting of the suitable range of a target parameter. It is explanatory drawing which shows another example of setting of the appropriate range of a target parameter. 9 is a flow chart showing a setting process of an appropriate range according to the second embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the parameter estimation device according to the present embodiment, the likelihood is set and learning is performed so that the estimated value of the target parameter (hereinafter, sometimes referred to as parameter or unknown parameter) has a high distribution density in a predetermined range.

In this embodiment, the estimated value of the unknown parameter is stochastically automatically calculated and evaluated. The quality of the target parameter is determined by assuming that the estimated value of the parameter follows a predetermined distribution in which the likelihood increases in a predetermined range.

According to the present embodiment, the likelihood function based on the appropriate range of the parameter is set, the parameter is estimated using the likelihood function of the estimated value of the observed value and the likelihood function based on the appropriate range, and the quality of the parameter is determined. You can judge. Therefore, in the present embodiment, when estimating a larger number of parameters than the number of types of observed values in utility equipment, it is possible to calculate an estimated value close to a true value.

The parameter estimation device and the parameter estimation method according to the present embodiment can also be expressed as follows, for example.

"Expression 1" A parameter estimating device for utility equipment, which sets a likelihood function, estimates a parameter, and determines a parameter based on an appropriate range of the parameter.

“Expression 2” In the parameter estimation method for utility equipment, a parameter estimation method that sets a likelihood function, estimates the parameter, and determines the parameter based on the appropriate range of the parameter.

“Expression 3” The parameter estimation method according to Expression 2, including the step of performing likelihood setting based on one continuous appropriate range.

“Expression 4” The parameter estimation method according to Expression 2, including a step of performing likelihood setting based on an appropriate range having a plurality of continuous sections.

“Expression 5” The parameter estimation method described in Expression 2, including a step of performing likelihood setting on the estimated parameter based on an appropriate range of the amount of change for a certain period of time.

“Expression 6” The parameter estimation method according to Expression 2, including the step of searching the storage unit for the equipment model formula and determining the parameter to be estimated.

[Expression 7] A method of estimating parameters of utility equipment, which is a method of estimating parameters by estimating and estimating parameters by generating and setting combinations of a plurality of likelihood functions based on appropriate ranges of the parameters.

[Expression 8] In the parameter estimation method described in Expression 2, even in a combination of likelihood settings, even if a parameter does not use an appropriate range, the appropriate range set in the appropriate range setting step is applied in the parameter determination step. , A parameter estimation method for determining pass/fail.

A first embodiment will be described with reference to FIGS. In the present embodiment, the case of estimating the unknown parameter for the utility equipment in the building will be described as an example, but the present invention is not limited to this. For example, factory production equipment, refrigeration equipment for commercial facilities, play equipment, elevators, It can also be applied to other facilities of other facilities such as escalator and food delivery facility.

<Overall structure>

FIG. 1 shows the overall configuration of a system including the parameter estimation device 1. The parameter estimation device 1 is connected to, for example, the BEMS 4 and the sensor 2 via the communication network CN. The sensor 2 is connected to the utility equipment (hereinafter, equipment) 3 and transmits measurement information (also referred to as sensor data or observation value) to the BEMS 4 or the parameter estimation device 1.

The facility 3 is, for example, a boiler facility, an air conditioner, a gas turbine power generator, a power-related device, or the like. The sensor 2 is a device that measures various physical quantities of the equipment 3 and outputs measurement information that is the measurement result. The sensor 2 measures, for example, power consumption, efficiency, flow rate, temperature, pressure, liquid level, mass, sound, light amount, and the like. It may be a sensor that measures psychophysical quantities such as color, taste, and smell.

The data (measurement information) measured by the sensor 2 may be directly transmitted to the parameter estimation device 1 via the communication network CN, or may be transmitted via a device such as the BEMS 4 and the communication network CN. May be sent to. Furthermore, the sensor 2 may transmit data regularly, or may transmit data in response to a request from the parameter estimation device 1 or the BEMS 4. Alternatively, the data measured by the sensor 2 may be stored in a recording medium and read into the parameter estimation device 1 from the recording medium.

In the present embodiment, the BEMS 4 is taken as an example, but the present invention is not limited to this, and any system that manages the utility equipment 3 such as CEMS may be used.

The parameter estimation device 1 is a device including, for example, a CPU (microprocessor), a memory, a communication function, a display function, an input function (none of which is shown). The parameter estimation device 1 can be configured using a personal computer or a dedicated computer.

Details of the functional configuration of the parameter estimation device 1 will be described later with reference to FIG. Here, the outline of the parameter estimation device 1 will be described first. The parameter estimation device 1 has functions such as an equipment model formula determination unit 41, an appropriate range setting unit 42, a parameter estimation unit 43, and a parameter determination unit 44.

FIG. 2 shows a functional configuration of the parameter estimation device 1. The parameter estimation device 1 includes, for example, an acquisition unit 10, an interface unit 20, a storage unit 30, and a calculation unit 40. Each of these functions 10 to 40 is realized by the microprocessor reading and executing a predetermined computer program (not shown) stored in the memory. At least a part of any one of the functions 10 to 40 may be configured as a hardware circuit.

The interface unit 20 has, for example, an input unit 21 and an output unit 22. The calculation unit 40 includes, for example, an equipment model formula determination unit 41, an appropriate range setting unit 42, a parameter estimation unit 43, and a parameter determination unit 44.

The acquisition unit 10 acquires data (also referred to as information) from the sensor 2 of the equipment 3, the BEMS 4 (or CEMS), and an information source (not shown) on the Internet. The acquired data is stored in the storage unit 30.

The input unit 21 of the interface unit 20 is a function for the user to input information to the parameter estimation device 1. The user uses the input device (not shown) such as a keyboard or a mouse to input information about the facility 3 and the like to the input unit 21. An input device such as a voice recognition device may be used instead of or in addition to the keyboard.

Information of the equipment 3 and the like input to the input unit 21 is transferred to and stored in the storage unit 30. A part or all of the information input to the input unit 21 can be displayed and output by the output unit 22. Further, the information input to the input unit 21 is sent to the calculation unit 40 via the storage unit 30.

The input information includes, for example, the specifications of the equipment 3, the installation location of the equipment 3, the connection relationship between the equipment 3 and other equipment, and parameters for estimating the equipment 3. Examples of input information are not limited to these.

The output unit 22 has a function of providing information to the user. The output unit 22 displays the parameter estimation result and the like using an output device (not shown) such as a display, a printer, and a voice synthesizer. The output unit 22 displays the information input to the input unit 21, reads and displays the information stored in the storage unit 30 from the calculation unit 40, and reads and displays the information stored in the storage unit 30. You can

The storage unit 30 transmits/receives information among the acquisition unit 10, the interface unit 20, and the calculation unit 40. The storage unit 30 stores the input information, extracts the requested information, and transmits it to the request source. In addition to the specification information of the equipment 3, the storage unit 30 stores an equipment model formula (to be described later), an appropriate range, parameter initial values, and the like.

The equipment model formula is a formula indicating the relationship between the data acquired from the acquisition unit 10 and the estimation parameter acquired from the interface unit 20. The equipment model formula can be configured as a plurality of formulas that can represent the relationship between the data acquired from the acquisition unit 10 and the estimated parameter, and may include a physical formula. The appropriate range as the “predetermined range” is an appropriate range of each parameter in which tags (type information) such as equipment type, equipment specifications, and installation environment are set.

The calculation unit 40 includes an equipment model formula determination unit 41, an appropriate range setting unit 42, a parameter estimation unit 43, and a parameter determination unit 44. The calculation unit 40 acquires information necessary for calculation from the interface unit 20 and the storage unit 30, estimates the target parameter, and transmits the estimation result to the interface unit 20. The estimation result of the target parameter is displayed on the output unit 22.

Information including the parameter estimation result may be provided to an external device (not shown) such as a facility introduction simulation system. Information may be transferred from the parameter estimation device 1 to an external device by wire or wirelessly, or may be stored in a recording medium and provided to the external device.

<Outline of Overall Processing of Parameter Estimation Device 1>

FIG. 3 is a flowchart showing the overall processing of the parameter estimation device 1. Details of steps S12 to S15 will be described later.

The parameter estimation device 1 acquires data from the interface unit 20 and the storage unit 30 (S11). The parameter estimation device 1 determines an equipment model formula based on the data acquired in step S11 (S12). The parameter estimation device 1 sets an appropriate range based on the data acquired in step S11 (S13).

Here, step S12 and step S13 may be processed in parallel. Step S12 may be processed first, and step S13 may be processed later. The order of the processes of step S12 and step S13 does not matter.

The parameter estimation device 1 estimates the target parameter based on the equipment model formula determined in step S12 and the appropriate range set in step S13 (S14).

The parameter estimation device 1 evaluates the estimated value of the parameter calculated in step S14 (S15). The evaluation is, for example, to judge the quality of the estimated value of the calculated parameter.

When the parameter estimation device 1 determines “NO” in step S15, the parameter estimation device 1 returns to step S14 and performs parameter estimation again. On the other hand, when the parameter estimation device 1 determines "good" in step S15, the parameter estimation device 1 outputs the estimated value of the parameter calculated in step S14 (S16).

<Process of Equipment Model Formula Determination Unit 41>

FIG. 4 is a flowchart showing details of the processing in the equipment model formula determination unit 41 shown in step S12 in FIG.

The equipment model formula determination unit 41 searches for one or more facility model formulas to be used for calculation from the facility model formulas stored in the storage unit 30 based on the data acquired in step S11 of FIG. 3 ( S121).

The equipment model formulas stored in the storage unit 30 are tagged with, for example, “equipment type”, “parameter on left side of formula”, and “parameter on right side of formula”. Therefore, the equipment model formula determination unit 41 searches for the equipment model formula based on the tag. Clustering may be used to retrieve the equipment model formula, but this is not the only option.

The equipment model formula determination unit 41 confirms the parameters of the equipment model formula retrieved in step S121 and determines parameters to be estimated (S122). Hereinafter, an example of step S121 and step S122 will be described using an equation for calculating the gas consumption of the steam boiler.

G=(H+Hb)/Hrout×Grout (1)

H=f(S, Es, Ew)... Formula (2)

Hb=f(S, Ek, Ew, Br)... Formula (3)

Es=f(P)... Formula (4)

Ew=f(Tw)... Formula (5)

Ek=f(Tk) Equation (6)

Tk=g(P)... Formula (7)

Here, "G" is the gas consumption amount, "H" is the heat amount, "Hb" is the blow loss heat amount, "Hrout" is the rated heat output, "Grout" is the rated gas consumption amount, "S" is the steam amount, " "Es" is steam enthalpy, "Ew" is water supply enthalpy, "Ek" is can water enthalpy, "Br" is blow coefficient, "P" is working pressure, "Tw" is water supply temperature, and "Tk" is water temperature. is there.

Each formula is a physical model formula or an approximate formula. For example, the formula (5) of the water supply enthalpy Ew is a quadratic approximation formula of the water supply temperature, and the formula (1) of the gas consumption G is a physical model formula.

In step S11, the blow coefficient Br is acquired as the estimation parameter and the steam boiler is acquired as the equipment type from the interface unit 20. Furthermore, in step S11, the amount of steam S, the working pressure P, the feed water temperature Tw, the rated heat output Hrout, and the rated gas consumption amount Grout are acquired from the acquisition unit 10 and the input unit 21. In this case, in step S121, by searching the equipment model formula in the storage unit 30, the formula (1), formula (2), formula (3), formula (4), formula (5), formula (6), Expression (7) is acquired.

The equipment model formula determination unit 41 determines parameters that need to be estimated in addition to the blow coefficient Br acquired as the estimation parameter in step S11 (S122). Here, in step S11, all parameters other than the blow coefficient Br have been acquired, so the estimated parameter is determined to be the blow coefficient Br. On the other hand, if the steam amount S cannot be acquired from the acquisition unit 10, the steam amount S is also determined to be an estimation parameter.

<Process of Appropriate Range Setting Unit 42>

FIG. 5 is a flowchart showing details of the processing in the proper range setting unit 42 shown in step S13 in FIG.

The appropriate range setting unit 42 searches the storage unit 30 for the appropriate range of the estimation parameter determined by the equipment model formula determination unit 41, and sets the retrieved appropriate range as the estimation target parameter (S131). That is, the appropriate range setting unit 42 searches the appropriate range stored in the storage unit 30 using the equipment type and the name of the estimation target parameter as tags, and sets the appropriate range for each estimation target parameter (S131). ).

For example, clustering can be used to search for the appropriate range, but the searching method does not matter. The appropriate range is not set for the estimation target parameter for which the appropriate range cannot be found.

As shown in FIGS. 6 to 8, there are various patterns in the proper range. FIG. 6 shows a case where the proper range of the feed water temperature is, for example, one continuous range of “20° C. to 40° C.”. FIG. 7 shows a case where the proper range of the supply water temperature has a plurality of continuous sections of “10° C. to 20° C.” and “30° C. to 40° C.”.

FIG. 8 shows an appropriate range of parameters that change the temperature during a certain fixed time (t1-t2). Point A is the temperature T1 at time t1. The point B1 is the lowest temperature T2 that can be taken at the time t2. The point B2 is the maximum temperature T3 that can be taken at the time t2.

In the example of FIG. 8, the temperature change range for a certain fixed time (difference between t1 and t2) is between ΔT1 and ΔT2. In this way, it is also possible to set the parameter variation amount ΔT due to the time difference (t1−t2) as an appropriate range.

Returning to FIG. The appropriate range setting unit 42 sets the likelihood function using the appropriate range set in step S131 (S132). A method of setting the likelihood function will be described in detail by taking a steam boiler as an example.

The equipment model formulas determined by the equipment model formula determination unit 41 are the formula (1), the formula (2), the formula (3), the formula (4), the formula (5), the formula (6), and the formula (7). In addition, it is assumed that the measured value of the gas consumption G is stored in the storage unit 30.

By inputting values to the parameters of the formula (1), the formula (2), the formula (3), the formula (4), the formula (5), the formula (6), and the formula (7), the gas consumption G An estimated value (estimated observed value) can be calculated. The likelihood of the calculated estimated value of the gas flow rate G can be calculated by a likelihood function. For example, when the observation error, which is the error between the measured value and the estimated value, has a normal distribution, the likelihood function is calculated by the following equation (8). In Expression (8), “X” is a probability function, “μ” is an average, and “σ ² ”is a variance. As an explanation of step S132, equation (8) is also shown in FIG.

f(X:μ,σ ² )=(1/(2πσ ² ) ^½ )exp(−(X−μ) ² /2σ ² ).
...Formula (8)

The appropriate range setting unit 42 updates the likelihood function using the appropriate range of the parameters set in step S131.

For example, when the water supply temperature is the estimation target parameter and the appropriate range is set to “20° C. to 40° C.” in step S131, the uniform distribution is applied as the likelihood function of the parameter. A uniform distribution is used here, but a distribution other than the uniform distribution can also be applied. When the estimated value of the water supply temperature is in the range of 20°C to 40°C, the likelihood is "1/(40-20)", and in other cases, the likelihood is "0".

The appropriate range setting unit 42 sets the likelihood function by multiplying the likelihood function shown in Expression (8) by the likelihood function of the parameter. Here, the setting of the likelihood function based on the appropriate range in FIG. 6 has been described, but the likelihood function setting based on the appropriate range shown in FIG. 7 or 8 is also applied by taking the likelihood function setting in the case of FIG. 6 as an example. It is possible to When there are a plurality of estimation target parameters and there are respective appropriate ranges, the likelihood function may be set by multiplying the likelihood function of equation (8) by the likelihood function of each parameter.

<Process of Parameter Estimating Unit 43>

The parameter estimation unit 43 estimates a parameter by using a likelihood function based on the equipment model formula determined by the equipment model formula determination unit 41 and the appropriate range of the estimation target parameter set by the appropriate range setting unit 42. . As the parameter estimation method, for example, a particle filter method or a Markov chain Monte Carlo method can be used.

<Process of Parameter Determination Unit 44>

The parameter determination unit 44 determines the quality of the estimated value of the parameter calculated by the parameter estimation unit 43. Examples of the quality criterion include the degree of convergence of the estimated value, the appropriate range, and the observation error.

The variance of the estimated value for each estimation step may be used as the degree of convergence of the estimated value. Instead of this, a standard deviation or the like may be used. In the pass/fail judgment based on the degree of convergence of the estimated value, the judgment is made based on a preset threshold value. In the pass/fail determination based on the appropriate range, it is determined as “good” when the estimated value converges within the appropriate range, and is determined as “fail” when the estimated value does not converge. In the quality judgment based on the observation error, the judgment is made based on a preset threshold value. In the quality judgment based on a plurality of judgment criteria (evaluation criteria) of the convergence of the estimated value, the appropriate range, and the observation error, if any one of the judgments is “No”, the parameter estimation unit 43 performs the estimation processing again. Is performed (S14).

<Processing of output unit 22>

The output unit 22 receives the information from the input unit 21, the storage unit 30, and the calculation unit 40, and displays it on the output device. The output unit 22 displays the determination result of each determination criterion by the parameter determination unit 44 of the calculation unit 40, the estimated value, and the appropriate range on the output device.

According to this example configured as described above, the estimation target parameter can be estimated efficiently and accurately. According to the present embodiment, even when estimating more parameters than the number of types of observed values, by using the likelihood function set so that the likelihood of estimated values within the appropriate range is high. The estimated values of the estimation target parameters can be converged within the proper range, and the estimated value close to the true value can be efficiently acquired.

In addition, in the present embodiment, by using the appropriate ranges of various patterns in the appropriate range setting unit 42, it is possible to obtain more information about the estimation target parameter. Therefore, in this case, the estimated value closer to the true value can be acquired more efficiently.

A second embodiment will be described with reference to FIG. Since this embodiment is a modification of the first embodiment, the description will focus on the differences from the first embodiment.

In the first embodiment, when setting the likelihood function in the appropriate range setting unit 42, the case where the likelihood function as a whole is set by multiplying the likelihood functions of all estimation target parameters has been described. However, if the determination result by the parameter determination unit 44 is “No”, the parameter estimation unit 43 will re-estimate the result, but even if it is re-estimated, the determination result will be “No”, The estimation process (S14) and the determination process (S15) may be repeated.

FIG. 9 shows processing S13A of the appropriate range setting unit 42 in this embodiment. Comparing the flowchart shown in FIG. 9 with the flowchart described in FIG. 5, step S132A in this embodiment is different from step S132 in the first embodiment. Therefore, the different step S132A will be described.

When setting the likelihood function based on the appropriate range set in step S131, the appropriate range setting unit 42 calculates the combination of the appropriate ranges of the estimation target parameters and sets the likelihood function (S132A). ..

For example, when there are two parameters to be estimated, that is, the parameter Pa and the parameter Pb, the combinations of the likelihood settings include the likelihood setting based on the appropriate range of the parameter Pa, the likelihood setting based on the appropriate range of the parameter Pb, and the appropriateness of the parameter Pa. There are a total of three likelihood settings based on the range and the appropriate range of the parameter Pb.

Then, the parameter estimation unit 43 of the present embodiment performs the parameter estimation process using the likelihood setting of each combination. The parameter determination unit 44 of this embodiment determines the quality of the parameter estimation value based on the likelihood setting of each combination.

The parameter determination unit 44 can use the determination criteria described in the first embodiment. Among the criteria, the appropriate range set in step S131 is applied to the estimation target parameter for which the appropriate range is not set, and the quality is determined.

For example, in the case of the combination in which the likelihood function is set only based on the proper range of the parameter Pa, the parameter determination unit 44 determines whether the estimated value of the parameter Pb is good or bad based on the proper range of the parameter Pb set at step S131. judge.

This embodiment, which is configured in this way, also exhibits the same effects as the first embodiment. Furthermore, in the present embodiment, since the estimation is performed using a combination of a plurality of likelihood settings, it is possible to increase the probability of acquiring the parameter estimation value that is determined as “good” by the parameter determination unit 44.

The present invention is not limited to the above embodiment. Those skilled in the art can make various additions and changes within the scope of the present invention. The above embodiment is not limited to the configuration example illustrated in the accompanying drawings. The configuration and the processing method of the embodiment can be appropriately changed within the scope of achieving the object of the present invention.

Further, each constituent element of the present invention can be arbitrarily selected, and an invention having a selected configuration is also included in the present invention. Furthermore, the configurations described in the claims can be combined in addition to the combinations specified in the claims.

1: Parameter estimation device, 2: Sensor, 3: Equipment, 41: Equipment model formula determination unit, 42: Appropriate range setting unit, 41: Parameter estimation unit, 42: Parameter determination unit

Claims (12)

A device for estimating parameters relating to equipment,
Obtaining an observation value from a sensor that measures a physical quantity related to the equipment,
Acquire the equipment model formula to calculate the observed value based on the target parameter,
The likelihood for the observed value of the estimated observed value obtained by substituting the estimated value of the target parameter into the equipment model formula is calculated by the first likelihood function,
The likelihood of the estimated value of the target parameter is calculated based on the second likelihood function and the first likelihood function for obtaining the probability that the estimated value of the target parameter belongs to a predetermined range prepared in advance,
Estimating the target parameter based on the calculated likelihood,
Evaluating the estimated target parameter according to a predetermined evaluation criterion,
Parameter estimation device. Output the evaluation result of the target parameter,
Parameter estimation device. The likelihood of the target parameter is set such that the distribution density of the estimated value of the target parameter in the predetermined range is larger than the distribution density outside the predetermined range,
The parameter estimation device according to claim 1. The equipment model formula is prepared in advance according to the type of equipment,
The parameter estimation device according to claim 1. The number of the target parameter is set to be larger than the number of observation values obtained from the sensor,
The parameter estimation device according to claim 1. The equipment model formula includes a plurality of different target parameters,
The parameter estimation device according to claim 1. The predetermined range is one continuous range,
The parameter estimation device according to claim 1. The predetermined range includes a plurality of continuous ranges,
The parameter estimation device according to claim 1. The predetermined range changes in a preset unit time,
The parameter estimation device according to claim 1. In the predetermined range, the likelihoods of the estimated values of the plurality of target parameters are calculated based on the second likelihood function and the first likelihood function according to a combination of the predetermined ranges for each of the plurality of target parameters. To do
The parameter estimation device according to claim 1. In the case of evaluating the estimated plurality of target parameters according to a predetermined evaluation criterion, the predetermined range regarding the target parameter is set for a target parameter having a predetermined range not included in the combination of the predetermined ranges. evaluate,
The parameter estimation device according to claim 10. A method for estimating parameters relating to equipment by a computer,
The calculator is
Obtaining an observation value from a sensor that measures a physical quantity related to the equipment,
Acquire the equipment model formula to calculate the observed value based on the target parameter,
The likelihood for the observed value of the estimated observed value obtained by substituting the estimated value of the target parameter into the equipment model formula is calculated by the first likelihood function,
The likelihood of the estimated value of the target parameter is calculated based on the second likelihood function and the first likelihood function for obtaining the probability that the estimated value of the target parameter belongs to a predetermined range prepared in advance,
Estimating the target parameter based on the calculated likelihood,
Evaluating the estimated target parameter according to a predetermined evaluation criterion,
Parameter estimation method.

Images