Plant performance evaluation system and plant performance evaluation method

Abstract

To provide a plant performance evaluation system and plant performance evaluation method making it possible to evaluate plant performance without depending on the expertise or labor of a person in charge.SOLUTION: A plant performance evaluation system that outputs a result of evaluation of a plant is adapted to a plant including various apparatuses and various sensors which measure the conditions of the respective apparatuses. The plant performance evaluation system includes an analysis evaluation device 40 that uses measurement values sg1 to sg3 of the sensors to inspect the relationships among the sensors, creates a model of the plant on the basis of the relationships, uses the model to calculate predictive measurement values on the basis of the measurement values of the sensors obtained at the time of creating the model and the measurement values received from the sensors in a predetermined period of time, compares the calculated measurement values with the measurement values received from the sensors in the predetermined period of time, and regards a result of the comparison as a result of evaluation of the plant.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] This invention relates to a plant performance evaluation system and a plant performance evaluation method that enable the evaluation of the performance of various plants.

Background Art

[0002] In a power generation plant of a power station, various devices such as steam turbines and boilers are used, and the power generation plant generates electricity using these devices. In such a power generation plant, the power generation efficiency is evaluated, that is, the plant performance is evaluated (see, for example, Patent Document 1). According to this document, necessary data is extracted from a huge amount of operation data of a thermal power plant to find abnormalities and optimal operation guidelines of the power generation plant.

[0003] There are various plant performance evaluation methods for power generation plants, but they are mainly classified into the following two types.

[0004] The first method is to evaluate the plant performance using software sold as an off-the-shelf product. This first method is incorporated into a computer. According to the first method, specification data of devices such as a generator, a condenser, a feed water system, a boiler, and a turbine of the plant is input into the computer. After that, the computer performs a heat balance calculation and compares it with a threshold value. Thus, according to the first method, abnormalities of the plant are detected.

[0005] For example, the plant may be a nuclear power plant that uses nuclear power. In this case, the thermal efficiency of the entire nuclear power plant is evaluated. In a general nuclear power plant, heat is generated by the nuclear fission reaction of nuclear fuel in the nuclear reactor. The generated heat turns the water in the nuclear reactor into high-temperature and high-pressure steam. This steam is sent to a turbine, and electricity is generated by the rotation of the turbine in a generator.

[0006] The steam that passes through the turbine is returned to water in the condenser. A portion of the steam is also sent to the feedwater heater, where it heats the feedwater through heat exchange, contributing to the plant's thermal efficiency. Meanwhile, the water condensed in the condenser is sent back to the reactor using equipment such as condensate pumps and feedwater pumps.

[0007] When using a computer incorporating the first method for such nuclear power plants, the design specifications of the equipment used in the plant are first entered into the computer. This input is performed by the person in charge of operating the plant. After entering the design specifications, the computer takes measurements obtained from actual operation, such as those shown below, and performs thermal efficiency calculations and evaluates the plant's performance. Note that the measurements a1 to a10 are a selection of those shown.

[0008] a1. Heat generated by nuclear fuel (converted to MW) a2. Generated electrical output (MW) a3. The amount of heat (temperature and pressure) of steam generated in a nuclear reactor. a4. Work done in a high-pressure turbine (heat input - heat output) a5. Work done in a low-pressure turbine (heat input - heat output) a6. Condenser vacuum value a7. Condensate heat (temperature and pressure) a8. Low-pressure water supply heat output (temperature and pressure) a9. High-pressure water supply heat output (temperature and pressure) a10. Work done in the condenser (heat input - heat output) The second method involves periodically plotting the parameters of equipment related to plant performance evaluation and individually checking and evaluating the changes in these parameters. In this second method, the checking and evaluation are performed by the personnel who operate the plant. [Prior art documents] [Patent Documents]

[0009] [Patent Document 1] Japanese Patent Publication No. 2005-38098 [Overview of the project] [Problems that the invention aims to solve]

[0010] The two methods described above have the following challenges. In the first method, accurate evaluation by computer is difficult unless detailed specification data and correction values ​​for the related equipment are correctly entered. Also, if equipment is updated or added, the system data needs to be reviewed each time. For this reason, the first method requires a high level of technical skill and effort from the person in charge.

[0011] In the second method, the evaluation depends on the individual's (person in charge's) competence, resulting in variability in the evaluation results. Furthermore, since the evaluation is conducted on an individual equipment basis, it is difficult to comprehensively evaluate the entire plant. Additionally, the second method involves periodically plotting equipment parameters and individually checking and evaluating the parameter trends, making real-time evaluation difficult.

[0012] The objective of this invention is to solve the aforementioned problems and to provide a plant performance evaluation system and a plant performance evaluation method that enable the evaluation of plant performance without depending on the technical skills and effort of the person in charge. [Means for solving the problem]

[0013] To solve the aforementioned problems, the invention of claim 1 is a plant performance evaluation system used in a plant equipped with various devices and various sensors for measuring the state of these devices, which outputs evaluation results for the plant, and which investigates the relationship between the sensors using the normal measurement values ​​of the sensors during the model target period, which is the period for creating a model that serves as the basis for evaluation, All of the things I investigated Based on the strength of the relationship, a model is created to represent the state of the plant at any given point in time or period, and thereafter, The normal measurement values ​​of each sensor during the aforementioned model period, Measurement values ​​received from each of the aforementioned sensors during a predetermined analysis period. and Based on the model, the predicted measurement values ​​are calculated, and the calculated measurement values ​​and the measurement values ​​received from each sensor during the predetermined analysis period are compared. use , there is a change in the strength of the relationship between the sensors. We investigate whether there is a change. and, I investigated This plant performance evaluation system is characterized by comprising a processing device that takes the results as evaluation results for the plant.

[0014] The invention of claim 1 is a plant performance evaluation system used in a plant equipped with various devices and various sensors for measuring the status of these devices, and which outputs evaluation results for the plant. In this plant performance evaluation system, the processing unit investigates the relationships between sensors using normal measurement values ​​of the sensors during the model target period, which is the period during which the processing unit creates a model that serves as the basis for evaluation, All of the things I investigated A model is created that represents the state of the plant at any given point in time or over a given period, based on the strength of the relationships. Next, the processing unit... Normal measurement values ​​for each sensor during the model period and , measurement values ​​received from each sensor during a predetermined analysis period. and Based on the model, the predicted measurement values ​​are calculated. After this, the processing unit combines the calculated measurement values ​​with the measurement values ​​received from each sensor during the predetermined analysis period. use , there is a change in the strength of the relationship between the sensors. We investigate whether there is a change. and, I investigated The results will be used as the evaluation results for the plant.

[0016] The invention of claim 2 is a plant performance evaluation system according to claim 1, wherein the processing device performs at least one of the following processes on the measurement values ​​of the sensor: discretizing the measurement values, removing values ​​that deviate from the discretized measurement values, and removing seasonal variation components from the discretized measurement values, and thereafter, From the strength of the relationship between the aforementioned sensors This is characterized by creating a model that represents the state of the plant at any given time or period.

[0017] The invention according to claim 3 is a plant performance evaluation method used in a plant equipped with various devices and various sensors for measuring the states of these devices, and for outputting an evaluation result of this plant from a processing device. In a model target period, which is a period for creating a model serving as an evaluation criterion, the processing device examines the relationship between the sensors using the normal measurement values of the sensors, All of the things I investigated and creates, using the processing device, a model representing the state of the plant at an arbitrary point in time or period based on the strength of the relationship between the sensors. After that, The measured values ​​of each sensor under normal conditions during the aforementioned model period, from the measurement values received from each of the sensors during a predetermined analysis target period to be analyzed, and the processing device calculates predicted measurement values based on the model, and compares the calculated measurement values with the measurement values received from each of the sensors during the predetermined analysis target period. use by the processing device the law of nature, if there is a change in the strength of the relationship between the sensors We investigate whether there is a change. and I investigated the processing device uses the result as the evaluation result of the plant. This is a plant performance evaluation method characterized by the above.

Advantages of the Invention

[0018] According to the inventions of claim 1 and claim 3, Normal measurement values ​​for each sensor during the model period and from the measurement values received from each sensor during a predetermined period, and the processing device calculates predicted measurement values based on the previously created model, and uses the result of comparing the calculated measurement values with the measurement values received from each sensor during the predetermined period as the evaluation result of the plant. Therefore, it is possible to evaluate the performance of the plant without being affected by the technical ability of the person in charge of operating the plant. Also, according to the inventions of claim 1 and claim 3, since the evaluation result of the plant is automatically created, the workload of the plant performance evaluation (cost reduction) can be reduced.

[0019] Also Since the model is created using the measurement values when the plant is normal, it is possible to detect early signs of equipment abnormalities through constant monitoring of the performance degradation trend of the plant, and to establish an effective replacement plan for the equipment based on the performance evaluation.

[0020] Claim 2 According to this invention, various processes are performed on the sensor's measured values ​​to correct them, thereby eliminating various influences and creating a model that corresponds to the plant's original state. [Brief explanation of the drawing]

[0021] [Figure 1] This is a diagram showing a plant performance evaluation system according to Embodiment 1 of the present invention. [Figure 2] This figure shows an example of sensor data. [Figure 3] This is a flowchart showing an example of the model creation process. [Figure 4] This is an explanatory diagram illustrating the discretization of input data. [Figure 5] This is an explanatory diagram illustrating outlier removal. [Figure 6] This is an explanatory diagram illustrating another method of outlier removal. [Figure 7] This is an explanatory diagram illustrating the removal of seasonal variations. [Figure 8] This is an explanatory diagram illustrating the extraction of data relationships. [Figure 9] This figure shows an example of a model. [Figure 10] This is a flowchart showing an example of a plant performance evaluation process. [Figure 11] This is an explanatory diagram illustrating the comparison between sensor prediction values ​​and actual sensor measurements. [Modes for carrying out the invention]

[0022] Next, embodiments of this invention will be described in detail with reference to the drawings. Figure 1 shows a plant performance evaluation system according to this embodiment. The plant performance evaluation system in Figure 1 is used in a power plant where a power plant 10 is installed, and mainly comprises a monitoring device 20 and an analysis and evaluation device 40.

[0023] The power plant 10 generates electricity using nuclear or thermal power, and this embodiment uses nuclear power generation as an example. Although not shown in the diagram, the power plant 10 uses numerous pieces of equipment such as a reactor, turbine, generator, pumps, and piping. The power plant 10 generates electricity using these pieces of equipment.

[0024] The power plant 10 is equipped with various sensors, which are also omitted from the diagram, to monitor the status of these devices. The measurements from each sensor include the following, as mentioned earlier.

[0025] a1. Heat generated by nuclear fuel (converted to MW) a2. Generated electrical output (MW) a3. The amount of heat (temperature and pressure) of steam generated in a nuclear reactor. a4. Work done in a high-pressure turbine (heat input - heat output) a5. Work done in a low-pressure turbine (heat input - heat output) a6. Condenser vacuum value a7. Condensate heat (temperature and pressure) a8. Low-pressure water supply heat output (temperature and pressure) a9. High-pressure water supply heat output (temperature and pressure) a10. Work done in the condenser (heat input - heat output) Each sensor sends the measurement values ​​sg1 to sg3 of the equipment, including the measurement values ​​a1 to a10 above, to the monitoring device 20.

[0026] The monitoring device 20 receives measurement values ​​sg1 to sg3 from sensors installed in the power plant 10. Upon receiving measurement values ​​sg1 to sg3 from each sensor, the monitoring device 20 creates sensor data as shown in Figure 2. The sensor data includes sensor identification information that represents the sensor and identifies the sensor, as well as the installation location of the sensor. The sensor data also includes the measurement value of the sensor, the date and time of measurement, etc., corresponding to the sensor identification information.

[0027] The monitoring device 20 transmits the generated sensor data to the analysis and evaluation device 40 via the company's internal communication network 30 at predetermined intervals. The predetermined interval can be set in seconds, minutes, hours, days, etc., as needed by the monitoring device 20.

[0028] The analysis and evaluation device 40 receives sensor data from the monitoring device 20 via the company's internal communication network 30, analyzes the received sensor data, and evaluates the power plant 10. For this purpose, the analysis and evaluation device 40 is equipped with a communication control unit 41, a data server 42, a management server 43, and clients 44-45. The communication control unit 41 to clients 45 of the analysis and evaluation device 40 are connected to enable data transmission and reception.

[0029] The communication control unit 41 performs communication control to connect the data server 42 to the client 45 to the internal communication network 30. For example, when the communication control unit 41 receives sensor data from the monitoring device 20 via the internal communication network 30, it sends this sensor data to the data server 42.

[0030] The data server 42 is a storage device that stores data related to the power plant. For example, when the data server 42 receives sensor data from the monitoring device 20 via the internal communication network 30 and the communication control unit 41, it stores this sensor data. When the data server 42 receives a data transmission request from the management server 43, it refers to the measurement date and time item of each sensor data, extracts the sensor data corresponding to the model target period described later, which is attached to the data transmission request, and sends it to the management server 43. Furthermore, if the analysis target period described later is attached to the data transmission request, the data server 42 extracts the sensor data corresponding to the analysis target period and sends it to the management server 43.

[0031] Clients 44-45 are computers for plant operation, operated by the personnel responsible for operating the power plant. Various instructions and other information necessary for power plant operation are entered into clients 44-45 by the personnel. For example, model creation instructions are entered into clients 44-45 to create short-term or long-term models. Upon receiving these instructions, clients 44-45 send them to the management server 43. The model creation instructions include a model target period, which is the period for which the model is to be created. The personnel create performance evaluation models using sensor data from past plant operations when the power plant is functioning normally, at least several times. These models may cover the entire plant operating period, a specific period in each season, or other areas depending on the evaluation objective.

[0032] Furthermore, in order to evaluate the performance of the power plant, for example, clients 44-45 receive performance evaluation instructions from their respective personnel. These performance evaluation instructions include the analysis period, which is the period covered by the power plant analysis, and the model period, which is the period for creating the performance evaluation model (hereinafter referred to as the "performance evaluation model"). Clients 44-45 send these performance evaluation instructions to the management server 43. Note that the analysis period may also be the time when the performance is evaluated, i.e., the performance evaluation time.

[0033] Furthermore, when clients 44-45 receive various data from the management server 43, such as the display data described later, they display the received data.

[0034] The management server 43 is a computer that performs various processes to enable the evaluation of plant performance. When the management server 43 receives a model creation instruction from clients 44-45, it checks whether the model to be created is short-term or long-term based on the model target period attached to the instruction. If the model is short-term, the management server 43 performs model creation processing to create a short-term model. An example of this model creation processing is shown in Figure 3. When the management server 43 starts the model creation processing, it selects data for the model target period (step S1).

[0035] In step S1, when the management server 43 receives a model creation instruction from, for example, the client 44, it refers to the model target period attached to the model creation instruction. The management server 43 then sends a data transmission request to the data server 42, requesting that it send sensor data corresponding to this model target period. After this, when the management server 43 receives the sensor data for each model target period from the data server 42, it arranges the measured values ​​contained in the sensor data in order of the passage of time, that is, in a time series, to create the input data. The management server 43 performs this generation of input data for each sensor.

[0036] When step S1 is completed, the management server 43 selects the first input data from the large number of input data generated (step S2). After this, the management server 43 performs data discretization processing (step S3). In step S3, the management server 43 discretizes the input data and decimates the input data. This is shown in Figure 4. The management server 43 extracts sensor data from the input data, which consists of a large number of sensor data, for example at a predetermined period, and discretizes the input data, i.e., decimates the data. In Figure 4, the dashed lines show the measured values ​​that have been removed by discretization. By discretizing the input data, the management server 43 removes short-term fluctuation elements and extracts long-term fluctuation trends. The discretization processing by the management server 43 can be adjusted according to the rate of progression of the plant performance degradation trend, and is effective in creating a model suitable for the target equipment.

[0037] When step S3 is completed, the management server 43 performs outlier removal processing (step S4). In step S4, if the management server 43 determines that the target measurement value is larger or smaller than a reference value calculated from adjacent measurements, for example as shown in Figure 5, it removes the target measurement value. Alternatively, the management server 43 may also perform the following: If the management server 43 determines that the target measurement value is larger or smaller than a reference value, it smooths the target measurement value based on adjacent measurements, as shown in Figure 6. The value obtained by smoothing here may be the average of two adjacent measurements and the target measurement value.

[0038] When step S4 is completed, the management server 43 performs a seasonal variation removal process (step S5). For example, if each measurement value in the input data is the intake temperature of cooling seawater, this intake temperature will fluctuate due to seasonal influences, as shown in the original signal in Figure 7. In such cases, in step S5, the management server 43 removes the variation elements originating from seasonal variations from each measurement value in the input data. That is, the seasonal variation component (low-frequency component), which is a variation element originating from seasonal variations, is removed (subtracted) from the original signal of the input data. The management server 43 then uses the result of removing the seasonal variation component as the measurement value after processing the input data.

[0039] In steps S3 to S5, the management server 43 performs various processes on the input data. As a result, the management server 43 eliminates various influences on the input data and converts it into input data that corresponds to the original state of the power plant.

[0040] When step S5 is completed, the management server 43 checks if there is any input data that was not selected in the previous step S2, i.e., any unselected input data (step S6). If there is any unselected input data in step S6, the management server 43 selects the next input data (step S7) and returns the process to step S3.

[0041] On the other hand, if there is no unselected input data in step S6, the management server 43 extracts relationships for each input data (step S8). In step S8, the management server 43 extracts relationships for each input data created in steps S3 to S5. Specifically, as shown in Figure 8, among the many input data a1 to an created in steps S3 to S5, if, for example, input data a1 and input data ak change in the same way, the management server 43 considers that there is a strong relationship between the sensor that outputs input data a1 and the sensor that outputs input data ak. In this way, the management server 43 checks for sensors with a strong relationship among all the input data.

[0042] After this, the management server 43 creates a model from the strength of all the relationships it has examined. In other words, the management server 43 models a set of mathematical formulas that represent the relationships between sensors that indicate the state of the power plant at any given time or period. The model thus created is represented in a pattern such as that shown in Figure 9. In Figure 9, the sensors installed in equipment B1 to B4 (shown as black circles) are each grouped together, and sensors with strong relationships are connected by lines.

[0043] The management server 43 saves the created model to the data server 42, and upon completing step S8, terminates the model creation process.

[0044] Meanwhile, the management server 43 receives model creation instructions from clients 44-45, and if the model to be created is long-term, it performs model creation processing to create the long-term model. This model creation processing is the same as the short-term model creation processing described earlier, but for example, step S4 can be omitted in the series of processes.

[0045] Thus, in the model creation process, the sensor values ​​used are assumed to include disturbances such as seasonal variations in the model. In this case, the following processing may be performed: If the model used is long-term, the predicted values ​​for the fluctuating sensor elements are calculated using quadratic and cubic equations. If the model is short-term, the model is created by processing it so that the transient changes in sensor values ​​are not used. Next, the actual measured values ​​for the fluctuating sensors at the evaluation date and time are input for the model to be evaluated. After this, performance evaluation can be performed using big data analysis processing based on the obtained results.

[0046] Incidentally, when the management server 43 receives a performance evaluation instruction from clients 44-45, it performs a plant performance evaluation process to evaluate the performance of the power plant. An example of this plant performance evaluation process is shown in Figure 10. When the management server 43 starts the plant performance evaluation process, it performs a model creation process (step S21). In step S21, the management server 43 creates a model of the power plant under normal conditions. To do this, the management server 43 refers to the model target period attached to the performance evaluation instruction, and if a model corresponding to this period is stored in the data server 42, it reads this model. If the corresponding model is not stored in the data server 42, the management server 43 performs the model creation process consisting of steps S1-S8 described above to create a reference model. The management server 43 uses the model thus created or the read model as the performance evaluation model.

[0047] When step S21 is completed, the management server 43 selects the data for the analysis period attached to the performance evaluation instruction (step S22). In step S22, the management server 43 sends a data transmission request to the data server 42 requesting that it send sensor data corresponding to the analysis period. After this, the management server 43 receives the sensor data for the requested period from the data server 42 and arranges the measured values ​​contained in this sensor data in chronological order to create input data. The management server 43 generates this input data for each sensor.

[0048] Once step S22 is completed, data analysis is performed using a plant model (step S23). In step S23, the management server 43 extracts the measured values ​​from each sensor from the input data selected in step S22, as shown in Figure 11, and uses them as sensor measurement values. Furthermore, the management server 43 extracts the sensor data measurements at the time the performance evaluation model was created, that is, the measurements under normal conditions. Then, the management server 43 records the measured values ​​during the analysis period (performance evaluation). And, the measurement value under normal conditions Based on the strength of the relationship between the sensors shown by the performance evaluation model obtained in step S21, the predicted measurement values ​​of the sensors are calculated and designated as the predicted sensor values.

[0049] Next, in step S23, the management server 43 compares the sensor prediction values ​​calculated based on the performance evaluation model with the sensor measurements taken during the analysis period (performance evaluation). In other words, the management server 43 checks whether there has been a change in the strength of the relationships between the sensors. If the performance evaluation results, which are the result of comparing the sensor prediction values ​​with the sensor measurements, show a change in the measurements taken during the analysis period (performance evaluation), that is, a change in the strength of the relationships between the sensors, the management server 43 creates a sensor list data showing the sensors whose relationship strength has changed, as well as the sensor measurements and sensor prediction values. The management server 43 then uses the sensor list data as the evaluation result for the power plant.

[0050] After this, the management server 43 outputs the evaluation results from step S23 to clients 44-45 (step S24). In other words, in step S24, the management server 43, having created the sensor list data in step S23, sends this data to clients 44-45.

[0051] Once step S24 is complete, the management server 43 terminates the plant performance evaluation process.

[0052] The above describes the configuration of the plant performance evaluation system. Next, we will explain the plant performance evaluation method using this system. When creating a performance evaluation model required for evaluating the performance of a power plant, the person in charge of operating the power plant operates, for example, client 44 to input model creation instructions for creating a short-term or long-term model of the power plant. At this time, the person in charge also inputs the model target period, which is the period for which the model will be created, into client 44. When client 44 receives the model creation instructions and the model target period, it sends the model creation instructions with the model target period added to the management server 43.

[0053] When the management server 43 receives a model creation instruction from the client 44, it performs the model creation process and creates a performance evaluation model according to the instruction. After this, the management server 43 saves the created performance evaluation model to the data server 42.

[0054] The person in charge will create a performance evaluation model using at least several sets of sensor data obtained during past plant operation when the power plant was functioning normally. The person in charge will create the performance evaluation model according to the purpose of the evaluation, such as for the entire plant operation period or for a specific period in each season.

[0055] By the way, when a person in charge performs a performance evaluation of a power plant, for example, they operate client 44 to input performance evaluation instructions. At this time, the person in charge inputs into client 44 the analysis period, which is the period for which the performance evaluation of the power plant will be evaluated, and the model period, which is the period for which the performance evaluation model that will serve as the basis for the evaluation will be created. When client 44 receives the performance evaluation instructions, model period, and analysis period, it sends the performance evaluation instructions, with the model period and analysis period added, to management server 43.

[0056] When the management server 43 receives a performance evaluation instruction from the client 44, it performs plant performance evaluation processing. This causes the management server 43 to calculate predicted sensor values ​​based on a performance evaluation model created using sensor data from the model target period. The management server 43 also obtains sensor measurement values ​​from the input data. After this, the management server 43 compares the predicted sensor values ​​calculated based on the performance evaluation model with the sensor measurement values ​​from the analysis target period (performance evaluation time). In other words, the management server 43 checks for any changes in the strength of the relationships between sensors. If there are changes in the strength of the relationships, the management server 43 sends a list of sensors with changed relationship strengths and measurement data as the evaluation result to the client 44.

[0057] When client 44 receives the sensor list data and evaluation results from management server 43, it displays this data and notifies the person in charge that the relationships between sensors have changed.

[0058] Thus, according to this embodiment, it is possible to automatically create a model of the power plant using only sensors installed on each piece of equipment in the power plant, and to evaluate the performance of the power plant based on the created model. Furthermore, since the model is created automatically, it is possible to evaluate the plant performance without depending on the technical skills or effort of the person in charge.

[0059] According to this embodiment, degradation trends are detected by examining changes in the strength of the relationships between individual sensors, rather than comparing past and present values ​​of individual sensors attached to the equipment. Therefore, it becomes unnecessary to perform equipment-specific data correction, etc. Furthermore, the evaluation target can be arbitrarily selected based on the strength of the relationships between sensors, such as the entire plant rather than individual equipment, and since the strength of the relationships between sensors is automatically extracted by the system, a broader and more comprehensive evaluation becomes possible.

[0060] According to this embodiment, when a change occurs in the strength of the relationship between sensors, the change in relationship strength is displayed to clients 44-45, making it possible to identify abnormalities in the equipment related to this change. Furthermore, if a difference is extracted in a specific part of the evaluation results, it is possible to identify whether it is an abnormality in the equipment related to that part or other causes (for example, dirt in the system piping).

[0061] According to this embodiment, by setting the analysis period to the most recent point in time, it becomes possible to accurately and in real time grasp the trend of plant performance degradation. Furthermore, by arranging and comparing the models stored in the data server 42 in chronological order, it becomes possible to investigate the performance degradation of the power plant, etc.

[0062] According to this embodiment, since data from sensors is used, there is no need for human intervention or to input the design specifications of the equipment in advance, resulting in uniform and rapid results, as well as labor savings (cost reduction). [Explanation of symbols]

[0063] 10 Power plants 20 Monitoring equipment 30. Internal communication network 40. Analysis and evaluation device (processing device) 41 Communication Control Unit 42 Data Servers 43 Management Server 44-45 Clients

Claims

1. A plant performance evaluation system used in a plant equipped with various devices and various sensors for measuring the status of these devices, which outputs evaluation results for the plant, A plant performance evaluation system characterized by comprising a processing device that: investigates the relationships between sensors using normal measurement values ​​of the sensors during a model target period, which is a period for creating a model that serves as the basis for evaluation; creates a model representing the state of the plant at any given time or period from the strength of all the relationships investigated; then calculates predicted measurement values ​​based on the model, using the normal measurement values ​​of each sensor during the model target period and the measurement values ​​received from each sensor during a predetermined analysis target period to be analyzed; and, using the calculated measurement values ​​and the measurement values ​​received from each sensor during the predetermined analysis target period, investigates whether there has been a change in the strength of the relationships between the sensors; and, if a change is found, uses the results of the investigation as the evaluation result of the plant.

2. The aforementioned processing apparatus is With respect to the measurement values ​​of the aforementioned sensor, at least one of the following processes is performed: discretizing the measurement values, removing values ​​that fall outside the discretized measurement values, and removing seasonal variation components from the discretized measurement values. Subsequently, a model representing the state of the plant at any given time or period is created based on the strength of the relationships between the sensors. The plant performance evaluation system according to feature 1.

3. A plant performance evaluation method used in a plant equipped with various devices and various sensors for measuring the status of these devices, wherein the evaluation results of the plant are output from a processing device, The processing device examines the relationship between the sensors using the normal measurement values ​​of the sensors during the model target period, which is the period for creating a model that serves as the basis for evaluation. The processing device creates a model representing the state of the plant at any given time or period based on the strength of the relationships between all the sensors examined. Subsequently, the processing unit calculates predicted measurement values ​​based on the model, using the normal measurement values ​​of each sensor during the model target period and the measurement values ​​received from each sensor during a predetermined analysis target period. Using the calculated measurement values ​​and the measurement values ​​received from each of the sensors during the predetermined analysis period, the processing device performs the following: The system checks whether there is a change in the strength of the relationship between the sensors, and if there is a change, the processing device uses the results of the check as the evaluation result of the plant. A method for evaluating plant performance characterized by the following features.

Images