JP6473538B1 - Lifetime evaluation apparatus and lifetime evaluation method for particle beam therapy system - Google Patents

Abstract

A particle beam therapy system lifetime evaluation apparatus and lifetime evaluation method capable of performing a highly accurate lifetime evaluation for a particle beam therapy system with large fluctuations in sensor data obtained in a normal state. A first evaluation model is created by clustering a cathode voltage, a cathode current, and a grid current acquired in a certain period of a high-frequency amplifier circuit 57A, 58A of a particle beam therapy system 100 to define a normal data group. The components of the created first evaluation model are converted by a conversion function (sigmoid function that is an activation function), a normal data group is determined, and a second evaluation model is created. Sensor data is evaluated by the first evaluation model, if normal, the life is evaluated, if abnormal, the data is converted by a conversion function, and the converted data is evaluated by the second evaluation model as normal or abnormal. If normal, evaluate life. [Selection] Figure 3

Description

  The present invention relates to a lifetime evaluation apparatus and lifetime evaluation method for a particle beam therapy system.

  The particle beam therapy system, which is a cancer treatment system using radiation, uses the dose concentration of the particle beam to concentrate and irradiate the particle beam to the affected area to suppress the influence on the surrounding healthy tissue. Is a feature.

  The structure includes an injector, an accelerator, and a treatment room. Then, a particle beam is emitted from the electron gun of the incident device into the ring-shaped accelerator, and when the necessary energy amount is reached by the accelerator, the particle beam is separated from the circular orbit and taken out as a beam to irradiate the affected part.

  The injection device has a high-frequency amplifier (RF amplifier), and the RF amplifier has a short lifetime among the components constituting the particle beam therapy system. To evaluate the lifetime of the particle beam therapy system, the lifetime evaluation of the RF amplifier is performed. is important.

  The RF amplifier includes an amplifying element having at least a cathode and a grid. For the life evaluation of the RF amplifier, it is important to monitor the cathode and the grid, which are main components of the amplifying element.

  As a technique that can be used for the life evaluation of the particle beam therapy system, there is an abnormal sign evaluation technique for evaluating the presence or absence of an abnormal sign of mechanical equipment.

  In this abnormality sign evaluation technique, when the machine facility is operating normally, an evaluation model is created by clustering sensor data acquired from a plurality of sensors installed in the machine facility. After that, the plurality of sensor data installed in the machine equipment in operation is acquired, the degree of divergence between the acquired sensor data and the created evaluation model is calculated, and the degree of abnormality indicated by the calculated degree of divergence is calculated. Evaluates whether there is any sign of abnormal equipment.

  For example, when sensor data having a degree of deviation within a predetermined range is acquired from the evaluation model, the sensor data is evaluated as “normal”, and the sensor data exceeding the sensor data is evaluated as “abnormal sign”.

  As an example of the abnormality sign evaluation technique, there is a technique described in Patent Document 1.

Japanese Patent No. 4832609

  However, the inventors of the present application have detected the sensor data having a large degree of deviation from the evaluation model in the RF amplifier of the injection device in the particle beam therapy system from the normal RF amplifier in the normal particle beam therapy system. I found that there was something.

  This is because the particle beam therapy system targets various specimens, the measurement conditions are likely to change, and even if the RF amplifier is normal, the obtained sensor data is outside the range of normal data collected so far. It is because it may become.

  In the prior art, even if the device is normal, there is a lack of recognition that when the device is evaluated from normal data collected so far, it may be determined to be abnormal.

  For this reason, the lifetime evaluation of the particle beam therapy system could not be performed with high accuracy.

  An object of the present invention is to solve the above-mentioned problems, and a particle beam therapy system capable of highly accurate life evaluation for a particle beam therapy system with a large fluctuation in sensor data obtained in a normal state. It is to implement a life evaluation apparatus and a life evaluation method.

  In order to achieve the above object, the present invention is configured as follows.

  An input unit for acquiring, in a time series, data indicating a plurality of cathode voltages, a plurality of cathode currents, and a plurality of grid currents of a plurality of high-frequency amplifiers arranged in the particle beam therapy system in a lifetime evaluation apparatus for a particle beam therapy system A storage unit storing time series data of data indicating the plurality of cathode voltages, data indicating the plurality of cathode currents and data indicating the plurality of grid currents, and the particle beam therapy system is operating normally Sometimes obtained cathode voltage average value, cathode current average value and grid current average value are average values of time series data indicating the plurality of cathode voltages, the plurality of cathode currents, and the plurality of grid currents. The average value of the cathode voltage average value, the cathode current average value, and the grid current average value is calculated. A first evaluation model is created to create a second evaluation model by converting the components of the first evaluation model by a predetermined conversion function and stored in the storage unit; and the input Data indicating the cathode voltage, cathode current, and grid current obtained from the first evaluation mode is used to evaluate the life of the plurality of high-frequency amplifiers using the first evaluation model, and the first evaluation mode is used. The cathode voltage, the cathode current, or the grid current evaluated as needing detailed evaluation by using the second evaluation model to evaluate the life of the plurality of high-frequency amplifiers using the second evaluation model. A life evaluation unit for evaluating the life of the amplifier.

Further, a lifetime estimation method of a plurality of high-frequency amplifier in the high-frequency power supply device provided to the accelerator, was obtained when the previous SL plurality of high-frequency amplifier is operating normally, a plurality of cathode voltage, a plurality of cathode current And a first step of calculating a cathode voltage average value, a cathode current average value, and a grid current average value that are respective average values of time series data indicating a plurality of grid currents, the cathode voltage average value, and the cathode current average A second step of creating a first evaluation model by clustering values and the grid current average value, and a third step of creating a second evaluation model by converting components of the first evaluation model by a predetermined conversion function And the cathode voltage, the cathode current, and the grid current are increased using the first evaluation model. A fourth step of evaluating a circuit of life, assessed using the first evaluation model, when evaluated as detailed evaluation required, the cathode voltage, the cathode current or the grid current, the conversion function And a fifth step of evaluating the lifetimes of the plurality of high-frequency amplifiers by using and converting and evaluating using the second evaluation model.

  ADVANTAGE OF THE INVENTION According to this invention, about the particle beam therapy system with the fluctuation | variation of the sensor data obtained in a normal state is large, the lifetime evaluation apparatus and lifetime evaluation method of the particle beam therapy system which can perform lifetime evaluation with high precision are realizable. it can.

1 is a schematic configuration diagram of a particle beam therapy system to which an embodiment of the present invention is applied. It is internal structure explanatory drawing of the high frequency amplifier circuit in a high frequency power supply device. 1 is an overall configuration diagram of a life evaluation apparatus according to an embodiment of the present invention. It is an operation | movement flowchart of the lifetime evaluation in the Example of this invention. It is a figure which shows the example of the 1st evaluation model used for lifetime evaluation. It is a figure which shows the example of the 2nd evaluation model used for lifetime evaluation.

  In the present invention, monitoring of the cathode and grid of the RF amplifier of the injection device in the particle beam therapy system is performed for each signal with respect to the cathode voltage (Vk), current (Ik), and grid current (Ig). An evaluation model is created by clustering learning data using data when operating normally, and a life evaluation is performed by comparing these signals with a dedicated evaluation model.

  Further, the incident device has a plurality of RF amplifiers, and output signals from the respective RF amplifiers vary greatly, and it is necessary to take measures against this.

  In the present invention, life evaluation is performed as follows.

  (1) The cathode voltage (Vk), current (Ik), and grid current (Ig) from a plurality of RF amplifiers when operating normally are obtained, the average value of each is obtained, and the reference value of each signal To do.

  (2) A direction and a distance with respect to the reference value are determined according to the characteristics of the particle beam therapy system with the reference value as a center, and a first evaluation model is created by clustering for each reference value.

  (3) The second evaluation model is created by subjecting the constituent elements of the first evaluation model to predetermined conversion.

  (4) Particle beam therapy by evaluating the acquired cathode voltage (Vk), current (Ik), and grid current (Ig) signals using the corresponding first evaluation model and evaluating the life of the RF amplifier. Assess system life.

  (5) When an abnormality sign is detected by the first evaluation model, the sensor data is subjected to predetermined conversion and evaluated by the second evaluation model.

  (6) The lifetime of each part is evaluated according to the degree of deviation between the sensor data and the second evaluation model, and the lifetime of the RF amplifier is evaluated to evaluate the lifetime of the particle beam therapy system.

  Examples of the present invention will be described below.

  FIG. 1 is a schematic configuration diagram of a particle beam therapy system (particle beam therapy system) 100 to which an embodiment of the present invention is applied.

  In FIG. 1, a particle beam therapy system 100 includes a synchrotron accelerator 50, a linac 56, an ion source 59, a beam transport system 52, an irradiation device 54, a treatment table 40, a life evaluation device (control device) 1, and a plurality of high frequencies. Power supply devices 57 and 58 are arranged.

  The ion source 59 is a device that generates ions used for treatment. The linac 56 is a device that accelerates ions generated by the ion source 59 to a speed suitable for entering the synchrotron accelerator 50.

  The synchrotron accelerator 50 is an accelerator that further accelerates ions accelerated up to a predetermined speed by the linac 56 to form an ion beam. The synchrotron accelerator 50 includes a bending electromagnet, a high-frequency acceleration cavity, an emission device 50a, and the like. In addition, although the case where the synchrotron type accelerator 50 was used was demonstrated, the accelerator should just be provided with the high frequency power supply device, for example, it can be set as a cyclotron type accelerator etc.

  The beam transport system 52 is connected to the synchrotron accelerator 50 and guides the ion beam emitted from the synchrotron accelerator 50 to the irradiation device 54.

  The irradiation device 54 is provided in the treatment room and is a device for irradiating an ion beam, and two scanning electromagnets that independently scan the beam in two orthogonal directions within a plane perpendicular to the beam trajectory. And a beam monitor.

  The treatment table 40 is a bed on which a patient 45 is placed.

  The lifetime evaluation apparatus 1 also serves as a control apparatus, and not only evaluates the lifetime of each part constituting the particle beam therapy apparatus, but also includes the devices and devices in the particle beam therapy apparatus 100 including the synchrotron accelerator 50. Controls actions and alarms.

  The high frequency power supply device 57 is a device that generates a high frequency to be supplied to the linac 56, and the high frequency power supply device 58 is a device that generates a high frequency to be supplied to the emission device 50 a in the synchrotron accelerator 50.

  In the particle beam therapy system 100, ions generated by the ion source 59 are accelerated by the linac 56 and the synchrotron accelerator 50 to form an ion beam. The accelerated ion beam is emitted from the synchrotron accelerator 50 and transported to the irradiation device 54 by the beam transport system 52. The transported ion beam is shaped by the irradiation device 54 so as to match the shape of the affected area, and a predetermined amount is irradiated to the target of the patient 45 lying on the treatment table 40.

  The ion source 59 is provided with a vacuum sensor 60 that measures the degree of vacuum in the ion source 59. The linac 56 is also provided with vacuum sensors 61 and 62. The degree of vacuum detected by the vacuum sensors 60, 61, and 62 is input to the abnormality sign evaluation apparatus 1 as sensor data.

  The high frequency power supply device 57 includes a high frequency amplification circuit (RF amplifier) 57A, and the high frequency power supply device 58 includes a high frequency amplification circuit (RF amplifier) 58A. That is, the high-frequency amplifier circuit 57A and the high-frequency amplifier circuit 58A are arranged in the particle beam therapy system 100.

  FIG. 2 is an explanatory diagram of the internal configuration of the high frequency amplifier circuit 57A in the high frequency power supply device 57. As shown in FIG. Since the internal configuration of the high frequency amplifier circuit 58A is the same as the internal configuration of the high frequency amplifier circuit 57A, only the internal configuration of the high frequency amplifier circuit 57A is shown, and the internal configuration of the high frequency amplifier circuit 58A and its detailed description are omitted.

  In FIG. 2, a high frequency amplifier circuit 57A schematically shows a grid grounding type amplifier circuit in which the grid 21 of the vacuum tube 23 is grounded. The vacuum tube 23, the cathode heating power source 24, the cathode power source 28, the plate power source 27, and the high frequency filter. 13, coupling capacitors 25 and 26, a grid current detector 29, and a cathode voltage detector 30.

  The vacuum tube 23 is used for high-frequency amplification, and a plurality of vacuum tubes having the same configuration are connected in parallel (in the example shown in FIG. 2, three vacuum tubes are connected in parallel to each other). The vacuum tube 23 is a triode having the cathode 20, the grid 21, and the plate 22, but may be a vacuum tube having a configuration such as a quadrupole or a pentode. FIG. 2 shows an example of a triode.

  Further, FIG. 2 shows a configuration in which three vacuum tubes 23 are connected in parallel. However, a plurality of vacuum tubes 23 can be connected according to the required output power, and the number is not limited as long as they are connected in parallel. . Although FIG. 2 shows grid grounding, the life prediction method for the vacuum tube 23 according to the present invention can be similarly implemented even if the cathode grounding or other circuit system is used in parallel.

  The cathode heating power supply 24 uses alternating current, and in this case, the high-frequency filter 13 is used together. A DC power supply may be used.

  The cathode power source 28 is a direct current power source, and is connected between the cathode 20 and the grid 21 with a positive ground, and the plate current Ip (Ip1, Ip2, Ip3) is adjusted by changing this value.

  The plate power source 27 is a direct current power source and is connected to the plate 22 through the high frequency filter 13 with a negative electrode grounded. It is assumed that the plate power source 27 of the present embodiment can detect the plate current Ip flowing through the plate 22 of each vacuum tube 23 therein. The plate current Ip is not limited to the form measured by the plate power source 27, and an ammeter may be arranged between each plate 22 and the plate power source 27 and measured by the ammeter.

  The values of both the cathode power supply 28 and the plate power supply 27 depend on the vacuum tube used. In FIG. 2, a single plate power supply 27 is shared, but separate power supplies may be used.

  The grid current detector 29 includes, for example, a shunt resistor or a current converter (CT), and detects the grid current Ig (Ig1, Ig2, Ig3) of each vacuum tube 23.

  The cathode voltage detector 30 detects the cathode voltage Vk (Vk1, Vk2, Vk3) by detecting the monitor output of the cathode power supply 28. Note that the cathode voltage Vk may be directly measured.

  Cathode currents Ik (Ik1, Ik2, Ik3) flowing through the cathode power supply 28 are calculated by the cathode current calculator 12 from the voltages detected by the cathode voltage detector 30. As can be seen from FIG. 2, since the cathode current Ik is the sum of the grid current Ig and the plate current Ip, the plate current Ip measured by the plate power source 27 and the grid current measured by the grid current detector 29 are displayed. It is also possible to calculate from Ig.

  Data S1-1, S1-4, S1-7 indicating cathode currents calculated by the three cathode current calculators 12, and data S1-2, S1 indicating cathode voltages detected by the three cathode voltage detectors 30 −5, S1-8 Data S1-3, S1-6, S1-9 indicating grid currents detected by the three grid current detectors 29 are input to the life evaluation apparatus 1.

  FIG. 3 is an overall configuration diagram of the life evaluation apparatus 1 according to the embodiment of the present invention, and FIG. 4 is an operation flowchart of life evaluation in the embodiment.

  FIG. 5 is a diagram illustrating an example of a first evaluation model used for life evaluation, and FIG. 6 is a diagram illustrating an example of a second evaluation model used for life evaluation.

  First, the first evaluation model and the second evaluation model will be described with reference to FIGS. 5 and 6.

  In the graph shown in FIG. 5, the horizontal axis indicates the parameter α, and the vertical axis indicates the parameter β. The parameter α and the parameter β indicate sensor data values from a plurality of sensors installed in the particle therapy apparatus 100, the parameter α is a voltage value detected by the cathode voltage detector 30, and the parameter β is grid current detection. The current value detected by the device 29.

  In FIG. 5, a, b, c, and d indicate a plurality of high-frequency amplifier circuits. In the example shown in FIG. 1, there are only two high-frequency amplifier circuits, but four examples are used as evaluation model examples. Then, the cathode voltage value data detected by the cathode voltage detector 30 of each high-frequency amplifier circuit and the grid current value data detected by the grid current detector 29 are plotted.

  A normal data group is defined by clustering a plurality of sensor data during a period in which the particle beam therapy system 100 is normally operated (operated) and the high-frequency amplifier circuit is determined to be normally operated (operated). An abnormal data group is defined by clustering a plurality of sensor data during a period in which it is determined that the circuit has operated abnormally.

  FIG. 6 is a graph showing the result of calculating the parameters α ′ and β ′ by converting the data of the parameters α and β shown in FIG. 5 using a predetermined conversion function (activation function such as a sigmoid function). It is. Then, a plurality of sensor data in a period in which it is determined that the high-frequency amplifier circuit of the particle beam therapy system 100 is normally operated are clustered into a normal data group, and a plurality of periods in which the high-frequency amplifier circuit is determined to be abnormally operated are clustered. Sensor data is clustered into an abnormal data group.

  Comparing FIG. 5 and FIG. 6, it can be understood that the concentration degree of the normal data group of the second evaluation model is higher than the concentration degree of the normal data group of the first evaluation model. That is, as compared with the first evaluation model, the second evaluation model can perform the range of the normal data group more clearly, and even if the data is out of the normal data group of the first evaluation model. Therefore, the possibility of being within the range of normal data is increased, and the accuracy of abnormality sign evaluation and the accuracy of life evaluation can be improved.

  Although it is conceivable to perform the abnormal sign evaluation and the life evaluation using only the second evaluation model, it is necessary to convert the sensor data by the activation function (sigmoid function), and therefore it is necessary for the abnormal sign evaluation and the life evaluation. The time will be longer.

  For this reason, in the present invention, first, as a result of the evaluation by the first evaluation model, when it is determined as abnormal and it is evaluated that the detailed evaluation is necessary, the sensor data is converted by the conversion function, and the converted data is converted into the second evaluation. The life is evaluated by evaluating whether the model is normal or abnormal. Thereby, as a result of the evaluation using the first evaluation model, only when it is determined that there is an abnormality, the sensor data only needs to be converted by the conversion function, so that the time required for the abnormality sign evaluation and the life evaluation can be shortened.

  The example shown in FIGS. 5 and 6 is an example in the case of two-dimensional data, but is also applicable to three-dimensional data. Therefore, the present invention can also be applied to the case of using the cathode voltage, the cathode current, and the grid current in the embodiment of the present invention.

  Next, the configuration and operation of the first embodiment will be described with reference to FIGS. 3 and 4.

  In FIG. 3, the life evaluation apparatus 1 includes an input unit 2, a memory (storage unit) 3, an average value calculation unit 4, a data clustering unit 5, a first evaluation model creation unit 6, and a conversion unit 7. The second evaluation model creation unit 8 and the life evaluation unit 9 are provided.

  In step S1 of FIG. 4, the data S1-1 to S1-9 from the high frequency amplifier circuit 57A and the data S2-1 to S2-9 from the high frequency amplifier circuit 58A when the particle beam therapy system 100 is normal are input to the input unit 2. It is input and collected in time series and is stored in the memory 3 (the memory 3 stores time series data).

  Next, in step S2, the average value calculation unit 4 calculates the average values (cathode voltage average value, cathode current average value, grid current average value) of the cathode voltage, cathode current, and grid current input in time series. Calculate and store in memory 3.

  In steps S3 and S4, the data clustering unit 5 creates a normal data group by clustering the average value data stored in the memory 3, and creates a first evaluation model by the first evaluation model creating unit 6. And stored (stored) in the memory 3.

  In step S <b> 5, the conversion unit 7 converts the components of the first evaluation model stored in the memory 3 using a sigmoid function that is an activation function, and stores it in the memory 3.

  Subsequently, in step S <b> 6, the second evaluation model creation unit 8 creates a second evaluation model based on the data converted by the conversion unit 7 and stores it in the memory 3.

  Steps S7 and thereafter are sensor data evaluation steps from the high-frequency amplifier circuits 57A and 58A.

  In step S7, sensor data from the high frequency amplifier circuits 57A and 58A are stored in the memory 3 via the input unit 2, and the life evaluation unit 9 uses the first evaluation model stored in the memory 3 for the stored data. Is used to evaluate whether the high-frequency amplifier circuits 57A and 58A are normal. That is, it is determined whether the average value of the input data is within the normal data range (first normal data range) set in the first evaluation model (determined (evaluated) in the first evaluation mode).

  In step S8, if the data is within the normal data range (when the deviation from the center of the normal data range is within a predetermined level), it can be determined that there is no abnormality sign (S8: Yes), and the process goes to step S9. Then, the high frequency amplifier circuits 57A and 58A are evaluated to be normal, and the lifetime is evaluated (steps S7, S8, and S9 are set to the first evaluation mode). The evaluation of the lifetime is determined by the degree of deviation from the center of the normal data range. The life evaluation based on the degree of deviation can be arbitrarily set for each device.

  In step S8, if the sensor data is not in the normal data range (when the deviation from the center of the first normal data range exceeds a predetermined size) (S8: No), the process proceeds to step S10, and step S10 The life evaluation unit 9 converts the sensor data by the conversion unit 7 and determines the converted data using the second evaluation model (determines (evaluates) in the second evaluation mode).

  And it progresses to step S11, and the lifetime evaluation part 9 judges whether the converted data are in the normal data range (2nd normal data range) in a 2nd evaluation model in step S11. If the converted data is within the normal data range in the second evaluation model (when the deviation from the center of the second normal data range is within a predetermined magnitude) (S11: Yes), the life evaluation is performed in step S9. (Steps S10, S11, and S9 are set as the second evaluation mode). If the converted data does not fall within the normal data range (second normal data range) in the second evaluation model (S11: No), it is evaluated in step S12 that the life of the particle beam therapy system 100 is complete.

  The life evaluation results in step S9 and step S12 are displayed on the display unit 10.

  Here, when the sensor data is evaluated as normal in step S8, the data evaluated as normal is added to the components of the first evaluation model based on the command of the life evaluation unit 9, and the memory 3 Is remembered. Further, these data evaluated as normal are converted by the conversion unit 7 using the conversion function based on the command of the life evaluation unit 9, added to the components of the second evaluation model, and stored in the memory 3. .

  As described above, according to the embodiment of the present invention, the cathode voltage, the cathode current, and the grid current acquired during the period when the high-frequency amplification circuits 57A and 58A of the particle beam therapy system 100 are normally operated are clustered. A normal data group is determined to create a first evaluation model, and the constituent elements of the created first evaluation model are converted by a conversion function (sigmoid function which is an activation function) to determine a normal data group and a second evaluation model Create Then, the sensor data is evaluated by the first evaluation model. If the sensor data is normal, the life is evaluated. If the sensor data is abnormal, the data is converted by a conversion function, and whether the converted data is normal or abnormal by the second evaluation model. Evaluate, and if normal, evaluate life.

  Therefore, it is possible to realize a life evaluation apparatus and a life evaluation method of a particle beam therapy system capable of highly accurate life evaluation even for a particle radiotherapy apparatus having a large variation in sensor data obtained in a normal state.

  In the above-described embodiment, the average value calculation unit 4, the data clustering unit 5, and the first evaluation model creation unit 6 are configured as separate units. However, the average value calculation unit 4 and the data clustering unit 5 and the first evaluation model creation unit 6 can be combined into a first evaluation model creation unit.

  Further, the conversion unit 7 and the second evaluation model creation unit 8 can be combined into a single second evaluation model creation unit.

  Further, the average value calculation unit 4, the data clustering unit 5, the first evaluation model creation unit 6, the conversion unit 7, and the second evaluation model creation unit 8 are integrated into an evaluation model creation unit. You can also.

  In the embodiment of the present invention, the sigmoid function is given as an example of the conversion function. However, a step function, softmax, identity function, etc. can be used.

  DESCRIPTION OF SYMBOLS 1 ... Life evaluation apparatus, 2 ... Input part, 3 ... Memory (memory | storage part), 4 ... Average value calculating part, 5 ... Data clustering part, 6 ... 1st evaluation Model creation unit, 7 ... Conversion unit, 8 ... Second evaluation model creation unit, 9 ... Life evaluation unit, 10 ... Display unit, 12 ... Cathode current calculator, 13 ... High frequency filter, 20 ... cathode, 21 ... grid, 22 ... plate, 23 ... vacuum tube, 24 ... heating power source for cathode, 25, 26 ... coupling capacitor, 27 ... plate Power source 28 ... Cathode power source 29 ... Grid current detector 30 ... Cathode voltage detector 40 ... Treatment table 45 ... Patient 50 ... Synchrotron accelerator 50a ... Ejecting device, 52 ... Bi Transport system, 54 ... irradiation device, 56 ... linac, 57, 58 ... high frequency power supply device, 57A, 58A ... high frequency amplification circuit, 59 ... ion source, 60-62 ... Vacuum sensor, 100 ... Particle beam therapy device

Claims (9)

An input unit for acquiring, in time series, data indicating a plurality of cathode voltages, a plurality of cathode currents, and a plurality of grid currents of a plurality of high-frequency amplifiers arranged in the particle beam therapy system;
A storage unit that stores time series data of data indicating the plurality of cathode voltages, data indicating the plurality of cathode currents, and data indicating the plurality of grid currents;
Cathode voltage average which is an average value of each of the time series data indicating the plurality of cathode voltages, the plurality of cathode currents, and the plurality of grid currents acquired when the particle beam therapy system is operating normally. Value, cathode current average value, and grid current average value are calculated, and a first evaluation model is created by clustering the respective average values of the cathode voltage average value, cathode current average value, and grid current average value. An evaluation model creating unit that converts the components of the evaluation model by a predetermined conversion function to create a second evaluation model, and stores the second evaluation model in the storage unit;
Data indicating the cathode voltage, cathode current, and grid current acquired from the input unit is used to evaluate the life of the plurality of high-frequency amplifiers using the first evaluation model, and the first evaluation mode. The cathode voltage, the cathode current, or the grid current evaluated as needing detailed evaluation using the plurality of the plurality of the plurality of high-frequency amplifiers in the second evaluation mode using the second evaluation model. A life evaluation unit for evaluating the life of the high frequency amplifier of
An apparatus for evaluating the lifetime of a particle beam therapy system comprising: In the lifetime evaluation apparatus of the particle beam therapy system according to claim 1,
The life evaluation unit is
In the first evaluation mode, detailed evaluation is performed when the degree of deviation between the predetermined position of the first evaluation model and the cathode voltage, the cathode current, and the grid current acquired from the input unit exceeds a predetermined magnitude. Evaluate it as necessary,
In the second evaluation mode, the degree of divergence between the predetermined position of the second evaluation model and the conversion data obtained by converting the cathode voltage, the cathode current, and the grid current acquired from the input unit by the conversion function is a predetermined amount. A life evaluation apparatus for a particle beam therapy system, wherein the life of the plurality of high-frequency amplifiers is evaluated as having expired when the length is exceeded. In the lifetime evaluation apparatus of the particle beam therapy system according to claim 1 or 2,
The life evaluation unit is
When the cathode voltage, cathode current or grid current acquired from the input unit is evaluated as normal in the first evaluation mode, the acquired cathode voltage, cathode current or grid current is used as a component of the first evaluation model. In addition to
A lifetime evaluation apparatus for a particle beam therapy system, wherein conversion data obtained by converting the acquired cathode voltage, cathode current, or grid current using the conversion function is added to a component of the second evaluation model. In the lifetime evaluation apparatus of the particle beam therapy system according to any one of claims 1, 2, and 3,
The lifetime evaluation apparatus for a particle beam therapy system, wherein the predetermined conversion function is a sigmoid function. In the lifetime evaluation apparatus of the particle beam therapy system according to claim 2,
The predetermined position of the first evaluation model is the center of the normal data range of the first evaluation model, and the predetermined position of the second evaluation model is the center of the normal data range of the second evaluation model. Lifetime evaluation device for a featured particle beam therapy system. A life evaluation method for a plurality of high-frequency amplifiers in a high-frequency power supply device provided in an accelerator ,
Was obtained when the previous SL plurality of high-frequency amplifier is operating normally, a plurality of cathode voltage, cathode voltage average value respectively the average values of the time series data shown multiple cathode current, and a plurality of grid current, A first step of calculating a cathode current average value and a grid current average value;
A second step of clustering the cathode voltage average value, the cathode current average value, and the grid current average value to create a first evaluation model;
A third step of creating a second evaluation model by converting the components of the first evaluation model by a predetermined conversion function;
A fourth step of evaluating the cathode voltage, the cathode current, and the grid current by using the first evaluation model to evaluate the lifetimes of the plurality of high-frequency amplifier circuits;
When the evaluation is performed using the first evaluation model and the detailed evaluation is required, the cathode voltage, the cathode current, or the grid current is converted using the conversion function, and the second evaluation model is used. A fifth step of evaluating the life of the plurality of high-frequency amplifiers by using
A method for evaluating the lifetime of a high-frequency amplifier, comprising: In the life evaluation method of the high frequency amplifier according to claim 6,
The fourth step includes
When the degree of deviation between the predetermined position of the first evaluation model and the cathode voltage, the cathode current, and the grid current exceeds a predetermined magnitude, it is evaluated that detailed evaluation is necessary,
Life of the plurality of high-frequency amplifiers when a deviation degree between a predetermined position of the second evaluation model and conversion data obtained by converting the cathode voltage, cathode current, and grid current by the conversion function exceeds a predetermined magnitude A method for evaluating the life of a high-frequency amplifier, characterized in that the evaluation is completed. In the life evaluation method of the high frequency amplifier of Claim 6 or 7,
The method for evaluating the life of a high-frequency amplifier, wherein the predetermined conversion function is a sigmoid function. In the life evaluation method of the high frequency amplifier of Claim 7,
The predetermined position of the first evaluation model is the center of the normal data range of the first evaluation model, and the predetermined position of the second evaluation model is the center of the normal data range of the second evaluation model. A method for evaluating the life of a high frequency amplifier.

Images