JP4058870B2 - Electronic control device with floating point arithmetic function - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an electronic control device having a floating point arithmetic function.
[0002]
[Prior art]
In general, an electronic control unit (ECU) applied to automobile engine control or the like has conventionally performed various operations using fixed-point data, but in recent years, a floating-point arithmetic processor (FPU) is used. ) Has made it possible to perform operations on floating-point data. According to the floating-point type data, the operation result can be obtained with very fine accuracy compared to the fixed-point type data.
[0003]
The floating-point type data is configured in accordance with, for example, the IEEE 754 standard, and includes a 1-bit sign part, an 8-bit exponent part, and a 23-bit mantissa part as shown in FIG. Thus, in the case of 4-byte (single-precision storage format) floating-point data having a mantissa part of 23 bits, it has a resolution of 7 digits (0.0000001 resolution).
[0004]
FIG. 8B shows a bit pattern in a single precision storage format. Floating-point type data includes a normalized number, a denormalized number, an infinite number, zero, and a combination of an exponent part and a mantissa part. A distinction is made between numbers. Here, a non-numeric value represents a numerical value, and a non-numeric value represents not a numerical value. For example, when a calculation result that cannot be expressed as a numerical value such as 0/0 or + ∞−∞ is represented, a non-number is used.
[0005]
[Problems to be solved by the invention]
As described above, floating-point data has a data format of a non-number. If even one non-number is generated or exists in the electronic control device, the non-number may multiply in the device. Further, for example, the results of four arithmetic operations including non-numbers are all non-numbers and become invalid. For example, in the comparison operation, when comparing whether the non-number is 1 or more, the result is false. Therefore, when a non-number occurs in the electronic control unit, there is a problem that the calculation result (output value) cannot be guaranteed at all.
[0006]
There are mainly two conditions for generating non-numbers in engine control. The first condition is when the floating point type RAM value is rewritten due to noise during operation of the electronic control unit or during battery backup, and the RAM value itself changes to an innumerable number. For example, there is a case where the floating-point RAM value is rewritten to FFFFFFFFh (all 1 bits) due to noise. The second condition is a case where an argument used for floating-point arithmetic is rewritten due to noise or the like, and an arithmetic operation such as 0/0 is performed to generate a secondary number. The case will be described in the calculation processing of the engine speed shown in FIG.
[0007]
That is, in the process of FIG. 9, the time T360 required for the engine to rotate at 360 ° CA (crank angle) is calculated and stored as the FRO value (step 501). 1 sec "is set in FR1 and" 60 "is set in FR2 (steps 502 to 504).
[0008]
After that, the arithmetic expression FR0 = FR1 / FR0 * FR2
Thus, the FR0 value is calculated (step 505), and the FR0 value is set as the engine speed Ne (step 506).
[0009]
In the process of FIG. 9, it is assumed that the FR0 and FR1 values change to 0 due to noise or the like immediately after step 504. In this case, the calculation of “0/0” is performed in step 505, and the FR0 value that is the calculation result of step 505 becomes a non-number. As a result, the engine speed cannot be calculated correctly.
[0010]
The present invention has been made paying attention to the above-mentioned problem, and the object of the present invention is to prevent a control failure caused by the occurrence of a non-number in an electronic control device having a floating-point arithmetic function. It is to provide an electronic control device capable of
[0011]
[Means for Solving the Problems]
In the electronic control unit according to claim 1, the microcomputer determines whether or not the data in the storage area of the floating-point data in the nonvolatile memory is non -numeric (non-numeric judging means), and judges that there is a non-numeric. At this time, the data in the storage area of the floating point type data in the nonvolatile memory is initialized (data initialization means). In such a case, the non-number is cleared by initializing the memory data in which the non-number exists. Therefore, various inconveniences caused by the existence of non-numerical data can be avoided in various controls by the microcomputer. As a result, it is possible to prevent a control failure associated with the occurrence of a non-number. The non-numeric determination means performs non-numeric determination on the storage area of the floating-point type data in the nonvolatile memory at the time of system initialization processing associated with power-on of the microcomputer. In this case, even if the floating point type data in the non-volatile memory is destroyed due to noise, etc. while the microcomputer is stopped, it will be detected properly immediately after the power is turned on, that is, before the control is started. This can prevent adverse effects on the control.
The non-volatile memory includes a backup RAM (also referred to as a standby RAM) that stores data in a non-volatile state by supplying power from a battery power source, an EEPROM, a flash ROM, and the like.
[0012]
In particular, as described in claim 2, upon initialization of the data in the storage area of the floating-point data in the nonvolatile memory, and writes the default value does not hinder the control in the storage area of the floating-point data in the nonvolatile memory Thus, various controls can be performed using this default value.
[0015]
In the third aspect of the invention, since the non-numeric determination is performed in the idle time of various controls even during the normal operation of the microcomputer, there is a large amount of floating-point type data, and the non-numeric determination requires time. In addition, the influence on other controls can be suppressed. This configuration is also effective from the viewpoint of the computational load of the microcomputer.
[0016]
According to a fourth aspect of the present invention, the microcomputer resets the microcomputer when the non-number determination means determines that there is a non-number in its normal operation. Therefore, the non-number is cleared in the initialization process associated with the resetting of the microcomputer.
[0017]
In the fifth aspect of the invention, the microcomputer stops the inversion operation of the watchdog clear (WDC) signal when the non-number determination means determines that there is a non-number in the normal operation . That is, by stopping the inversion operation of the watchdog clear signal, the periodicity of the signal is lost, and the watchdog circuit outputs a reset signal to the microcomputer. As a result, the microcomputer is reset, and the non-number is cleared in the initialization process accompanying the reset.
[0018]
In such a case, as described in claim 6 , when it is determined that there is a non-number during normal operation of the microcomputer , it is preferable to prohibit all interruptions. As a result, it is possible to surely prevent the inconvenience that the controllability is deteriorated due to the non-number after the presence of the non-number is confirmed until the deletion.
[0019]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an outline of an engine control system. In this system, the vehicle-mounted engine 1 is configured as, for example, a gasoline injection type multi-cylinder internal combustion engine.
[0020]
The engine control ECU 10 includes a microcomputer (hereinafter referred to as a microcomputer) 11, and the microcomputer 11 includes a CPU 12, a RAM 13, a ROM 14, an FPU (floating point arithmetic processor) 15, and an I / O (input / output device) 16. Here, the FPU 15 performs a calculation in the floating point format, and the CPU 12 performs a calculation other than the floating point format. The I / O 16 includes a well-known A / D converter. The RAM 13 is a storage element that retains stored contents by receiving power from an in-vehicle battery (not shown) even when the ignition switch is OFF. In the present embodiment, the RAM 13 corresponds to a nonvolatile memory (backup RAM). However, the nonvolatile memory may include an EEPROM, a flash ROM, or the like.
[0021]
The ECU 10 also includes a WDC (watchdog clear) monitoring device 18 as a watchdog circuit. The WDC monitoring device 18 monitors a WDC signal periodically output from the microcomputer 11 and outputs a reset signal to the microcomputer 11 every time the periodicity is lost, such as when the WDC signal is interrupted. The watch dog circuit may be built in the microcomputer 11.
[0022]
For example, as shown in FIG. 2, the WDC signal is output as a signal that inverts H / L at a cycle of 4 ms, and if the inversion edge is not detected for a predetermined time (32 ms in the figure), the WDC monitoring device 18 sets a reset signal. Lower. After that, when the reset is released, the WDC signal that is inverted at a cycle of 4 ms is taken into the WDC monitoring device 18 again.
[0023]
Various information representing the engine operating state is input to the ECU 10 from the sensor 2 provided in the engine 1. The sensor 2 includes, for example, an air flow meter for detecting the intake air amount, a water temperature sensor for detecting the cooling water temperature, a throttle opening sensor for detecting the throttle opening, and the like. Then, the ECU 10 performs fuel injection control, fuel gas purge control, and the like by an injector (not shown) based on the input sensor signal.
[0024]
As described above, the floating-point data calculated by the FPU 15 is configured according to, for example, the IEEE 754 standard, and has a data format shown in FIG. Further, as shown in FIG. 8B, the floating point type data is classified into a normalized number, a denormalized number, an infinite number, zero, and a non-number according to the combination of the exponent part and the mantissa part.
[0025]
On the other hand, FIG. 3 shows a configuration example of the RAM 13. The RAM 13 can be roughly divided into a floating-point RAM area, a fixed-point RAM area, and a stack area. The floating-point RAM area is divided by 4 bytes, “START” is assigned to the start address, and “END” is assigned to the end address. In this case, in the RAM 13, it is preferable that the floating-point RAM area is separated from the fixed-point RAM area and the stack area.
[0026]
In the present embodiment, the floating point type data held in the RAM 13 becomes non-numbered due to the influence of noise or the like before the operation power supply to the ECU 10 is turned on, or the floating point type data during the normal operation of the ECU 10. When the number becomes a non-number, it is determined to that effect and a data initialization process is performed to remove the non-number. The objects to be subjected to the non-numeric check are constants and RAM data defined in the software. Hereinafter, the gist of the present embodiment will be described in detail according to each process of FIGS. 4 to 7 executed by the CPU 12.
[0027]
First, a non-numerical check procedure immediately after power-on to the ECU 10 will be described with reference to the flowchart of FIG. FIG. 4 shows an initial routine that is started immediately after power-on, and this routine corresponds to “data initialization means”.
[0028]
In step 101, a non-number occurrence flag indicating the presence or absence of a non-number is set to “0”. Next, in step 102, a non-number check routine is called. In this non-number check routine, it is determined whether or not there is a non-number in the floating-point RAM area, details of which will be described later with reference to FIG.
[0029]
In the next step 103, it is determined whether or not the non-number generation flag is “0”. If the non-number generation flag is “0”, there is no non-number and the process proceeds to step 106. In step 106, after performing other initialization processes, this process is terminated.
[0030]
If there is a non-number in any of the floating-point RAM areas, the non-number generation flag is set to “1”. Therefore, the process proceeds to step 104 to initialize the floating-point RAM area. That is, a default value that does not hinder control is written in the floating-point RAM area, and the non-number is erased. Since this initialization process eliminates non-numbers, the non-number generation flag is set to “0” in step 105. In the subsequent step 106, other initialization processing is executed, and then this processing is terminated.
[0031]
Next, details of the non-number check routine will be described with reference to the flowchart of FIG. This routine corresponds to “non-number determination means”. In this processing, for each area from the start address START to the end address END of the floating-point RAM, combinations in which the exponent parts (bits 30 to 23) are all “1” and the mantissa parts (bits 22 to 0) are not all “0”. Whether or not there is a non-number is determined by whether or not.
[0032]
Specifically, the following loop processing is performed while setting the address by sequentially changing the variable R1 in units of 4 bytes from the start value START to the end value END. At this time, in step 201, the contents of the address indicated by the variable R1 are captured as the R0 value. In step 202, it is determined whether or not the bits 30 to 23 of the R0 value are “11111111”. In step 203, it is determined whether or not all the bits 22 to 0 of the R0 value are 0.
[0033]
If step 202 is NO, or if both steps 202 and 203 are YES, the floating-point type data at the corresponding address is not a non-numeric value, and the address of the next floating-point type data is transferred. If step 202 is YES and step 203 is NO, it is assumed that there is a non-number in the floating-point RAM area, and in step 204, “1” is set in the non-number generation flag, and the process ends.
[0034]
On the other hand, during normal operation of the ECU 10, the idle routine shown in FIG. Here, the idle routine is a process that is executed during idle time of various controls that are processed in a time cycle or rotation signal synchronization, and is executed in a task that has the lowest priority among the various processes that are normally handled by the ECU 10. Process.
[0035]
In step 301 of FIG. 6, the non-number check routine of FIG. 5 is called, and the non-number generation flag is operated in accordance with the presence or absence of the number. In the subsequent step 302, various controls such as a checksum of the ROM 14 and a register confirmation process in the microcomputer 11 are performed.
[0036]
In the 4 ms processing shown in FIG. 7, the result of the non-number check performed in the idle routine of FIG. 6 is monitored. That is, in step 401 of FIG. 7, it is determined whether or not the non-number occurrence flag is “1”. If the non-number generation flag = 0 and there is no non-number, the process proceeds to step 402, and the WDC signal is inverted. In such a case, since the WDC signal is inverted at a cycle of 4 ms, the WDC monitoring device 18 can be informed that the CPU 12 is operating normally. In the subsequent step 403, various controls such as A / D conversion processing of various sensor detection values are performed at a cycle of 4 ms, and then this processing is temporarily terminated.
[0037]
If the non-number occurrence flag = 1 and there is a non-number, the process proceeds to step 404 to prohibit all interrupts, and then enters an infinite loop. In other words, since the WDC signal is not inverted after entering the infinite loop, the periodicity of the WDC signal is lost, and a reset signal is output from the WDC monitoring device 18 to the microcomputer 11 after a predetermined period (after 32 ms). The Therefore, in the initial routine (the process of FIG. 4) accompanying the resetting of the microcomputer 11, the non-number generation flag is confirmed again, and the non-number is cleared in the same process.
[0038]
In the process of FIG. 7 above, if there are non-numbers, all interrupts are prohibited and an infinite loop is entered. Therefore, after determining that there are non-numbers, there are various numbers depending on the number before the reset signal is issued. Inconveniences such as adverse effects on the control are avoided. That is, if interruption is permitted after determination that there is a non-number, the non-number may be used in the generated interrupt processing, and control may become abnormal, but such inconvenience does not occur according to FIG. .
[0039]
According to the embodiment described in detail above, the following effects can be obtained.
(A) Whether or not there is a non-number is determined for the floating-point RAM area, and when it is determined that there is a non-number, the non-number is cleared by initializing the RAM data. Therefore, in the various controls by the microcomputer 11, various inconveniences caused by the existence of data including non-numbers can be avoided. As a result, it is possible to prevent a control failure associated with the occurrence of a non-number.
[0040]
(B) Since a default value that does not hinder control is written as RAM data and the floating-point RAM area is initialized, various controls can be performed using this default value.
[0041]
(C) When the system is initialized (initial routine) when the power to the microcomputer 11 is turned on, the non-number determination is performed on the floating-point RAM area of the RAM 13. In this case, even if the floating point type data in the RAM 13 is destroyed due to noise or the like while the operation of the microcomputer 11 is stopped, it can be properly detected immediately after the power is turned on, that is, before the control is started. An adverse effect on the control is prevented.
[0042]
(D) Since the non-number determination is performed during the idle time of various controls during the normal operation of the microcomputer 11, a large amount of floating-point data exists, and even when the non-number determination requires time, other control is performed. There is no impact. That is, the inconvenience that other control is not executed according to the predetermined scheduling is avoided. This configuration can also be said to be effective from the viewpoint of the computational load of the microcomputer 11.
[0043]
(E) When it is determined that there is a non-number, the inversion operation of the WDC signal is stopped, so that the microcomputer 11 is reset, and the non-number can be cleared by an initialization process associated with the reset.
[0044]
(F) When it is determined that there is a non-number, all interrupts are prohibited and an infinite loop is entered, so control is performed due to the non-number between the confirmation of the existence of the non-number and the deletion. Inconveniences such as deterioration in performance can be reliably prevented.
[0045]
In addition to the above, the present invention can be realized in the following forms.
Although the non-number check routine of FIG. 5 has been described as a configuration for realizing the non-number determination means, the means may be embodied by a hardware configuration of a microcomputer in addition to such software implementation.
[0046]
In the above embodiment, the microcomputer 11 is reset using the WDC monitoring device 18 during normal operation. However, the reset operation is not limited to the configuration described above. The microcomputer 11 itself may be configured to be reset by software.
[0047]
In the above embodiment, the electronic control unit (ECU) that handles floating-point type data in the single-precision storage format is exemplified, but the present invention can also be applied to an electronic control unit that handles floating-point type data in the double-precision storage format.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an outline of an engine control system in an embodiment of the invention.
FIG. 2 is a time chart showing the form of a WDC signal.
FIG. 3 is a diagram illustrating a configuration example of a RAM.
FIG. 4 is a flowchart showing an initial routine.
FIG. 5 is a flowchart showing a non-number check routine.
FIG. 6 is a flowchart showing an idle routine.
FIG. 7 is a flowchart showing 4 ms processing.
FIG. 8 is a diagram showing a configuration of floating point type data.
FIG. 9 is a flowchart showing a rotational speed calculation process.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Engine, 10 ... Engine control ECU (electronic control apparatus), 11 ... Microcomputer, 12 ... CPU which comprises a non-number determination means, a data initialization means, 13 ... RAM (memory), 15 ... FPU (floating point arithmetic processor) 18) WDC monitoring device.

Claims (6)

A microcomputer that has a floating-point arithmetic function for calculating a floating-point value, performs various controls according to a predetermined control program , retains stored contents even when the operation power to the microcomputer is shut off, and is a floating-point type met electronic control device and a nonvolatile memory having a storage area of the data,
The microcomputer includes a non -numeric determination means for determining whether or not the data in the storage area of the floating-point data in the nonvolatile memory is non-numeric.
Data initialization means for initializing data in the storage area of the floating-point type data in the nonvolatile memory when the non-number determination means determines that there is a non-number;
Equipped with a,
The electronic control device according to claim 1, wherein the non-number determination means performs a non-number determination on a storage area of the floating point type data in the nonvolatile memory at the time of initialization of the system accompanying power-on of the microcomputer . Said data initialization means, claim wherein upon initialization of the data in the storage area of the floating-point data in the nonvolatile memory, for writing the default value does not hinder the control in the storage area of the floating point data of the non-volatile memory The electronic control device according to 1. 3. The electronic control according to claim 1, wherein the non-number determination means performs the non-number determination on the storage area of the floating-point type data in the nonvolatile memory during idle times of various controls even during normal operation of the microcomputer. apparatus. The non-numeric determination means performs non-numeric determination on the storage area of the floating-point data in the nonvolatile memory even during normal operation of the microcomputer, and the microcomputer performs the non-numeric determination means during the normal operation. The electronic control device according to any one of claims 1 to 3 , wherein the microcomputer is reset when it is determined that there is a non-number . A watchdog circuit that monitors a watchdog clear signal periodically output from the microcomputer, and further outputs a reset signal to the microcomputer every time the periodicity of inversion of the watchdog clear signal is lost,
The non-numeric determination means performs non-numeric determination on the storage area of the floating-point data in the nonvolatile memory even during normal operation of the microcomputer, and the microcomputer performs the non-numeric determination means during the normal operation. The electronic control device according to any one of claims 1 to 4 , wherein when it is determined that there is a non-number, the inversion operation of the watchdog clear signal is stopped . The non-number determination means performs non-number determination on the storage area of the floating point type data in the nonvolatile memory even during normal operation of the microcomputer,
The electronic control device according to any one of claims 1 to 5 , wherein during the normal operation of the microcomputer, when the non-number determination means determines that there is a non-number, all interrupts are prohibited .

Images